Since the dawn of time (aka Postgres95) multiple pins of the same
buffer by one backend have been optimized not to modify the shared
refcount more than once. This optimization has always used a NBuffer
sized array in each backend keeping track of a backend's pins.
That array (PrivateRefCount) was one of the biggest per-backend memory
allocations, depending on the shared_buffers setting. Besides the
waste of memory it also has proven to be a performance bottleneck when
assertions are enabled as we make sure that there's no remaining pins
left at the end of transactions. Also, on servers with lots of memory
and a correspondingly high shared_buffers setting the amount of random
memory accesses can also lead to poor cpu cache efficiency.
Because of these reasons a backend's buffers pins are now kept track
of in a small statically sized array that overflows into a hash table
when necessary. Benchmarks have shown neutral to positive performance
results with considerably lower memory usage.
Patch by me, review by Robert Haas.
Discussion: 20140321182231.GA17111@alap3.anarazel.de
Per recent analysis by Andres Freund, this implementation is in fact
unsafe, because ARMv5 has weak memory ordering, which means tha the
CPU could move loads or stores across the volatile store performed by
the default S_UNLOCK. We could try to fix this, but have no ARMv5
hardware to test on, so removing support seems better. We can still
support ARMv5 systems on GCC versions new enough to have built-in
atomics support for this platform, and can also re-add support for
the old way if someone has hardware that can be used to test a fix.
However, since the requirement to use a relatively-new GCC hasn't
been an issue for ARMv6 or ARMv7, which lack the swpb instruction
altogether, perhaps it won't be an issue for ARMv5 either.
The assertion failed if WAL_DEBUG or LWLOCK_STATS was enabled; fix that by
using separate memory contexts for the allocations made within those code
blocks.
This patch introduces a mechanism for marking any memory context as allowed
in a critical section. Previously ErrorContext was exempt as a special case.
Instead of a blanket exception of the checkpointer process, only exempt the
memory context used for the pending ops hash table.
Support for running postgres on Alpha hasn't been tested for a long
while. Due to Alpha's uniquely lax cache coherency model it's a hard
to develop for platform (especially blindly!) and thought to be
unlikely to currently work correctly.
As Alpha is the only supported architecture for Tru64 drop support for
it as well. Tru64's support has ended 2012 and it has been in
maintenance-only mode for much longer.
Also remove stray references to __ksr__ and ultrix defines.
Commit ea9df812d8 failed to include
NUM_BUFFER_PARTITIONS in this offset, resulting in a bad offset.
Ultimately this threw off NUM_FIXED_LWLOCKS which is based on
earlier offsets, leading to memory allocation problems. It seems
likely to have also caused increased LWLOCK contention when
serializable transactions were used, because lightweight locks used
for that overlapped others.
Reported by Amit Kapila with analysis and fix.
Backpatch to 9.4, where the bug was introduced.
It's critical that the backend's idea of LOBLKSIZE match the way data has
actually been divided up in pg_largeobject. While we don't provide any
direct way to adjust that value, doing so is a one-line source code change
and various people have expressed interest recently in changing it. So,
just as with TOAST_MAX_CHUNK_SIZE, it seems prudent to record the value in
pg_control and cross-check that the backend's compiled-in setting matches
the on-disk data.
Also tweak the code in inv_api.c so that fetches from pg_largeobject
explicitly verify that the length of the data field is not more than
LOBLKSIZE. Formerly we just had Asserts() for that, which is no protection
at all in production builds. In some of the call sites an overlength data
value would translate directly to a security-relevant stack clobber, so it
seems worth one extra runtime comparison to be sure.
In the back branches, we can't change the contents of pg_control; but we
can still make the extra checks in inv_api.c, which will offer some amount
of protection against running with the wrong value of LOBLKSIZE.
Commit a730183926 created rather a mess by
putting dependencies on backend-only include files into include/common.
We really shouldn't do that. To clean it up:
* Move TABLESPACE_VERSION_DIRECTORY back to its longtime home in
catalog/catalog.h. We won't consider this symbol part of the FE/BE API.
* Push enum ForkNumber from relfilenode.h into relpath.h. We'll consider
relpath.h as the source of truth for fork numbers, since relpath.c was
already partially serving that function, and anyway relfilenode.h was
kind of a random place for that enum.
* So, relfilenode.h now includes relpath.h rather than vice-versa. This
direction of dependency is fine. (That allows most, but not quite all,
of the existing explicit #includes of relpath.h to go away again.)
* Push forkname_to_number from catalog.c to relpath.c, just to centralize
fork number stuff a bit better.
* Push GetDatabasePath from catalog.c to relpath.c; it was rather odd
that the previous commit didn't keep this together with relpath().
* To avoid needing relfilenode.h in common/, redefine the underlying
function (now called GetRelationPath) as taking separate OID arguments,
and make the APIs using RelFileNode or RelFileNodeBackend into macro
wrappers. (The macros have a potential multiple-eval risk, but none of
the existing call sites have an issue with that; one of them had such a
risk already anyway.)
* Fix failure to follow the directions when "init" fork type was added;
specifically, the errhint in forkname_to_number wasn't updated, and neither
was the SGML documentation for pg_relation_size().
* Fix tablespace-path-too-long check in CreateTableSpace() to account for
fork-name component of maximum-length pathnames. This requires putting
FORKNAMECHARS into a header file, but it was rather useless (and
actually unreferenced) where it was.
The last couple of items are potentially back-patchable bug fixes,
if anyone is sufficiently excited about them; but personally I'm not.
Per a gripe from Christoph Berg about how include/common wasn't
self-contained.
Instead of storing the ID of the dynamic shared memory control
segment in a file within the data directory, store it in the main
control segment. This avoids a number of nasty corner cases,
most seriously that doing an online backup and then using it on
the same machine (e.g. to fire up a standby) would result in the
standby clobbering all of the master's dynamic shared memory
segments.
Per complaints from Heikki Linnakangas, Fujii Masao, and Tom
Lane.
The special feature the XLogInsert slots had over regular LWLocks is the
insertingAt value that was updated atomically with releasing backends
waiting on it. Add new functions to the LWLock API to do that, and replace
the slots with LWLocks. This reduces the amount of duplicated code.
(There's still some duplication, but at least it's all in lwlock.c now.)
Reviewed by Andres Freund.
With this in place, a session blocking behind another one because of
tuple locks will get a context line mentioning the relation name, tuple
TID, and operation being done on tuple. For example:
LOG: process 11367 still waiting for ShareLock on transaction 717 after 1000.108 ms
DETAIL: Process holding the lock: 11366. Wait queue: 11367.
CONTEXT: while updating tuple (0,2) in relation "foo"
STATEMENT: UPDATE foo SET value = 3;
Most usefully, the new line is displayed by log entries due to
log_lock_waits, although of course it will be printed by any other log
message as well.
Author: Christian Kruse, some tweaks by Álvaro Herrera
Reviewed-by: Amit Kapila, Andres Freund, Tom Lane, Robert Haas
Revise the original decision to expose a uint64-based interface and
use Size everywhere possible. Avoid assuming that MAXIMUM_ALIGNOF is
8, or making any assumption about the relationship between that value
and sizeof(Size). If MAXIMUM_ALIGNOF is bigger, we'll now insert
padding after the length word; if it's smaller, we are now prepared
to read and write the length word in chunks.
Per discussion with Tom Lane.
The new function dsm_detach_all() can be used either by postmaster
children that don't wish to take any risk of accidentally corrupting
shared memory; or by forked children of regular backends with
the same need. This patch also updates the postmaster children that
already do PGSharedMemoryDetach() to do dsm_detach_all() as well.
Per discussion with Tom Lane.
If a postmaster child invokes fork() and then calls on_exit_reset, that
should be sufficient to let it exit() without breaking anything, but
dynamic shared memory broke that by not updating on_exit_reset() to
discard callbacks registered with dynamic shared memory segments.
Per investigation of a complaint from Tom Lane.
A fake relcache entry can "own" a SmgrRelation object, like a regular
relcache entry. But when it was free'd, the owner field in SmgrRelation
was not cleared, so it was left pointing to free'd memory.
Amazingly this apparently hasn't caused crashes in practice, or we would've
heard about it earlier. Andres found this with Valgrind.
Report and fix by Andres Freund, with minor modifications by me. Backpatch
to all supported versions.
This feature, building on previous commits, allows the write-ahead log
stream to be decoded into a series of logical changes; that is,
inserts, updates, and deletes and the transactions which contain them.
It is capable of handling decoding even across changes to the schema
of the effected tables. The output format is controlled by a
so-called "output plugin"; an example is included. To make use of
this in a real replication system, the output plugin will need to be
modified to produce output in the format appropriate to that system,
and to perform filtering.
Currently, information can be extracted from the logical decoding
system only via SQL; future commits will add the ability to stream
changes via walsender.
Andres Freund, with review and other contributions from many other
people, including Álvaro Herrera, Abhijit Menon-Sen, Peter Gheogegan,
Kevin Grittner, Robert Haas, Heikki Linnakangas, Fujii Masao, Abhijit
Menon-Sen, Michael Paquier, Simon Riggs, Craig Ringer, and Steve
Singer.
These are needed in HEAD to make assorted contrib modules build on Windows.
Now that all the MSVC and Mingw buildfarm members seem to be on the same
page about the need for them, we can have some confidence that future
problems of this ilk will be detected promptly; there seems nothing more
to be learned by delaying this fix further.
I chose to mark QueryCancelPending as well, since it's easy to imagine code
that wants to touch ProcDiePending also caring about QueryCancelPending.
Replication slots are a crash-safe data structure which can be created
on either a master or a standby to prevent premature removal of
write-ahead log segments needed by a standby, as well as (with
hot_standby_feedback=on) pruning of tuples whose removal would cause
replication conflicts. Slots have some advantages over existing
techniques, as explained in the documentation.
In a few places, we refer to the type of replication slots introduced
by this patch as "physical" slots, because forthcoming patches for
logical decoding will also have slots, but with somewhat different
properties.
Andres Freund and Robert Haas
This patch adds an option, huge_tlb_pages, which allows requesting the
shared memory segment to be allocated using huge pages, by using the
MAP_HUGETLB flag in mmap(). This can improve performance.
The default is 'try', which means that we will attempt using huge pages,
and fall back to non-huge pages if it doesn't work. Currently, only Linux
has MAP_HUGETLB. On other platforms, the default 'try' behaves the same as
'off'.
In the passing, don't try to round the mmap() size to a multiple of
pagesize. mmap() doesn't require that, and there's no particular reason for
PostgreSQL to do that either. When using MAP_HUGETLB, however, round the
request size up to nearest 2MB boundary. This is to work around a bug in
some Linux kernel versions, but also to avoid wasting memory, because the
kernel will round the size up anyway.
Many people were involved in writing this patch, including Christian Kruse,
Richard Poole, Abhijit Menon-Sen, reviewed by Peter Geoghegan, Andres Freund
and me.
This makes it possible to store lwlocks as part of some other data
structure in the main shared memory segment, or in a dynamic shared
memory segment. There is still a main LWLock array and this patch does
not move anything out of it, but it provides necessary infrastructure
for doing that in the future.
This change is likely to increase the size of LWLockPadded on some
platforms, especially 32-bit platforms where it was previously only
16 bytes.
Patch by me. Review by Andres Freund and KaiGai Kohei.
Previously, we did this just once per checkpoint, but that could make
Hot Standby take a long time to initialize. To avoid busying an
otherwise-idle system, we don't do this if no WAL has been written
since we did it last.
Andres Freund
In ordinary operation, VACUUM must be careful to take a cleanup lock on
each leaf page of a btree index; this ensures that no indexscans could
still be "in flight" to heap tuples due to be deleted. (Because of
possible index-tuple motion due to concurrent page splits, it's not enough
to lock only the pages we're deleting index tuples from.) In Hot Standby,
the WAL replay process must likewise lock every leaf page. There were
several bugs in the code for that:
* The replay scan might come across unused, all-zero pages in the index.
While btree_xlog_vacuum itself did the right thing (ie, nothing) with
such pages, xlogutils.c supposed that such pages must be corrupt and
would throw an error. This accounts for various reports of replication
failures with "PANIC: WAL contains references to invalid pages". To
fix, add a ReadBufferMode value that instructs XLogReadBufferExtended
not to complain when we're doing this.
* btree_xlog_vacuum performed the extra locking if standbyState ==
STANDBY_SNAPSHOT_READY, but that's not the correct test: we won't open up
for hot standby queries until the database has reached consistency, and
we don't want to do the extra locking till then either, for fear of reading
corrupted pages (which bufmgr.c would complain about). Fix by exporting a
new function from xlog.c that will report whether we're actually in hot
standby replay mode.
* To ensure full coverage of the index in the replay scan, btvacuumscan
would emit a dummy WAL record for the last page of the index, if no
vacuuming work had been done on that page. However, if the last page
of the index is all-zero, that would result in corruption of said page,
since the functions called on it weren't prepared to handle that case.
There's no need to lock any such pages, so change the logic to target
the last normal leaf page instead.
The first two of these bugs were diagnosed by Andres Freund, the other one
by me. Fixes based on ideas from Heikki Linnakangas and myself.
This has been wrong since Hot Standby was introduced, so back-patch to 9.0.
This code provides infrastructure for user backends to communicate
relatively easily with background workers. The message queue is
structured as a ring buffer and allows messages of arbitary length
to be sent and received.
Patch by me. Review by KaiGai Kohei and Andres Freund.
This interface is intended to make it simple to divide a dynamic shared
memory segment into different regions with distinct purposes. It
therefore serves much the same purpose that ShmemIndex accomplishes for
the main shared memory segment, but it is intended to be more
lightweight.
Patch by me. Review by Andres Freund.
Minor improvement to commit daa7527afc2274432094ebe7ceb03aa41f916607:
s_lock.h no longer has any need to mention PGSemaphoreData, so we can
rip out the #include that supplies that. In a non-HAVE_SPINLOCKS
build, this doesn't really buy much since we still need the #include
in spin.h --- but everywhere else, this reduces #include footprint by
some trifle, and helps keep the different locking facilities separate.
Instead of allocating a semaphore from the operating system for every
spinlock, allocate a fixed number of semaphores (by default, 1024)
from the operating system and multiplex all the spinlocks that get
created onto them. This could self-deadlock if a process attempted
to acquire more than one spinlock at a time, but since processes
aren't supposed to execute anything other than short stretches of
straight-line code while holding a spinlock, that shouldn't happen.
One motivation for this change is that, with the introduction of
dynamic shared memory, it may be desirable to create spinlocks that
last for less than the lifetime of the server. Without this change,
attempting to use such facilities under --disable-spinlocks would
quickly exhaust any supply of available semaphores. Quite apart
from that, it's desirable to contain the quantity of semaphores
needed to run the server simply on convenience grounds, since using
too many may make it harder to get PostgreSQL running on a new
platform, which is mostly the point of --disable-spinlocks in the
first place.
Patch by me; review by Tom Lane.
Just as backends must clean up their shared memory state (releasing
lwlocks, buffer pins, etc.) before exiting, they must also perform
any similar cleanups related to dynamic shared memory segments they
have mapped before unmapping those segments. So add a mechanism to
ensure that.
Existing on_shmem_exit hooks include both "user level" cleanup such
as transaction abort and removal of leftover temporary relations and
also "low level" cleanup that forcibly released leftover shared
memory resources. On-detach callbacks should run after the first
group but before the second group, so create a new before_shmem_exit
function for registering the early callbacks and keep on_shmem_exit
for the regular callbacks. (An earlier draft of this patch added an
additional argument to on_shmem_exit, but that had a much larger
footprint and probably a substantially higher risk of breaking third
party code for no real gain.)
Patch by me, reviewed by KaiGai Kohei and Andres Freund.
When acquiring a lock in fast-path mode, we must reset the locallock
object's lock and proclock fields to NULL. They are not necessarily that
way to start with, because the locallock could be left over from a failed
lock acquisition attempt earlier in the transaction. Failure to do this
led to all sorts of interesting misbehaviors when LockRelease tried to
clean up no-longer-related lock and proclock objects in shared memory.
Per report from Dan Wood.
In passing, modify LockRelease to elog not just Assert if it doesn't find
lock and proclock objects for a formerly fast-path lock, matching the code
in FastPathGetRelationLockEntry and LockRefindAndRelease. This isn't a
bug but it will help in diagnosing any future bugs in this area.
Also, modify FastPathTransferRelationLocks and FastPathGetRelationLockEntry
to break out of their loops over the fastpath array once they've found the
sole matching entry. This was inconsistently done in some search loops
and not others.
Improve assorted related comments, too.
Back-patch to 9.2 where the fast-path mechanism was introduced.
Development of IRIX has been discontinued, and support is scheduled
to end in December of 2013. Therefore, there will be no supported
versions of this operating system by the time PostgreSQL 9.4 is
released. Furthermore, we have no maintainer for this platform.
All of these platforms are very much obsolete.
As far as I can determine, the last version of SINIX, later renamed
Reliant, occurred some time between 2002 and 2005.
The last release of SunOS that would run on a sun3 was released in
November of 1991; the last release of OpenBSD which supported that
platform was in 2001. The highest clock speed of any processor in
the family was 25MHz.
The NS32K (national semiconductor 320xx) architecture was retired
in 1990.
Support can be re-added if a maintainer emerges for any of these
platforms, but it seems unlikely.
Reviewed by Andres Freund.
If a tuple was frozen while its predicate locks mattered,
read-write dependencies could be missed, resulting in failure to
detect conflicts which could lead to anomalies in committed
serializable transactions.
This field was added to the tag when we still thought that it was
necessary to carry locks forward to a new version of an updated
row. That was later proven to be unnecessary, which allowed
simplification of the code, but elimination of xmin from the tag
was missed at the time.
Per report and analysis by Heikki Linnakangas.
Backpatch to 9.1.
Testing done in 2011 by Tom Lane concluded that this is a win on Intel Xeons
and AMD Opterons, but it was not changed back then, because of an old
comment in tas() that suggested that it's a huge loss on older Opterons.
However, didn't have separate TAS() and TAS_SPIN() macros back then, so the
comment referred to doing a non-locked initial test even on the first
access, in uncontended case. I don't have access to older Opterons, but I'm
pretty sure that doing an initial unlocked test is unlikely to be a loss
while spinning, even though it might be for the first access.
We probably should do the same on 32-bit x86, but I'm afraid of changing it
without any testing. Hence just add a note to the x86 implementation
suggesting that we probably should do the same there.
Using the infrastructure provided by this patch, it's possible either
to wait for the startup of a dynamically-registered background worker,
or to poll the status of such a worker without waiting. In either
case, the current PID of the worker process can also be obtained.
As usual, worker_spi is updated to demonstrate the new functionality.
Patch by me. Review by Andres Freund.
On HPPA, implement pg_memory_barrier() as pg_compiler_barrier(), which
should be correct since this arch doesn't do memory access reordering,
and is anyway better than the completely-nonfunctional-on-this-arch
dummy_spinlock code. (But note this patch only fixes things for gcc,
not for builds with HP's compiler.)
Also, fix incorrect default definition of pg_memory_barrier as a macro
requiring an argument.
Also, fix incorrect spelling of "#elif" as "#else if" in icc code path
(spotted by pgindent).
This doesn't come close to fixing all of the functional and stylistic
deficiencies in barrier.h, but at least it un-breaks my personal build.
Now that we're actually using barriers in the code, this file is going
to need some serious attention.
There is a new API, RegisterDynamicBackgroundWorker, which allows
an ordinary user backend to register a new background writer during
normal running. This means that it's no longer necessary for all
background workers to be registered during processing of
shared_preload_libraries, although the option of registering workers
at that time remains available.
When a background worker exits and will not be restarted, the
slot previously used by that background worker is automatically
released and becomes available for reuse. Slots used by background
workers that are configured for automatic restart can't (yet) be
released without shutting down the system.
This commit adds a new source file, bgworker.c, and moves some
of the existing control logic for background workers there.
Previously, there was little enough logic that it made sense to
keep everything in postmaster.c, but not any more.
This commit also makes the worker_spi contrib module into an
extension and adds a new function, worker_spi_launch, which can
be used to demonstrate the new facility.
Itanium doesn't have the mfence instruction - that's a 386 thing. Use the
"mf" instruction instead.
This reverts the previous commit to add "#include <emmintrinsic.h>"; the
problem was not with a missing #include.
This patch replaces WALInsertLock with a number of WAL insertion slots,
allowing multiple backends to insert WAL records to the WAL buffers
concurrently. This is particularly useful for parallel loading large amounts
of data on a system with many CPUs.
This has one user-visible change: switching to a new WAL segment with
pg_switch_xlog() now fills the remaining unused portion of the segment with
zeros. This potentially adds some overhead, but it has been a very common
practice by DBA's to clear the "tail" of the segment with an external
pg_clearxlogtail utility anyway, to make the WAL files compress better.
With this patch, it's no longer necessary to do that.
This patch adds a new GUC, xloginsert_slots, to tune the number of WAL
insertion slots. Performance testing suggests that the default, 8, works
pretty well for all kinds of worklods, but I left the GUC in place to allow
others with different hardware to test that easily. We might want to remove
that before release.
Reviewed by Andres Freund.
SnapshotNow scans have the undesirable property that, in the face of
concurrent updates, the scan can fail to see either the old or the new
versions of the row. In many cases, we work around this by requiring
DDL operations to hold AccessExclusiveLock on the object being
modified; in some cases, the existing locking is inadequate and random
failures occur as a result. This commit doesn't change anything
related to locking, but will hopefully pave the way to allowing lock
strength reductions in the future.
The major issue has held us back from making this change in the past
is that taking an MVCC snapshot is significantly more expensive than
using a static special snapshot such as SnapshotNow. However, testing
of various worst-case scenarios reveals that this problem is not
severe except under fairly extreme workloads. To mitigate those
problems, we avoid retaking the MVCC snapshot for each new scan;
instead, we take a new snapshot only when invalidation messages have
been processed. The catcache machinery already requires that
invalidation messages be sent before releasing the related heavyweight
lock; else other backends might rely on locally-cached data rather
than scanning the catalog at all. Thus, making snapshot reuse
dependent on the same guarantees shouldn't break anything that wasn't
already subtly broken.
Patch by me. Review by Michael Paquier and Andres Freund.
In some cases with higher numbers of subtransactions
it was possible for us to incorrectly initialize
subtrans leading to complaints of missing pages.
Bug report by Sergey Konoplev
Analysis and fix by Andres Freund
MarkBufferDirtyHint() writes WAL, and should know if it's got a
standard buffer or not. Currently, the only callers where buffer_std
is false are related to the FSM.
In passing, rename XLOG_HINT to XLOG_FPI, which is more descriptive.
Back-patch to 9.3.
pg_filedump and other external utility programs are likely to want to be
able to check Postgres page checksums. To avoid messy duplication of code,
move the checksumming functionality into an exported header file, much as
we did awhile back for the CRC code.
In passing, get rid of an unportable assumption that a static char[] array
will be word-aligned, and do some other minor code beautification.
Use the same gcc atomic functions as we do on newer ARM chips.
(Basically this is a copy and paste of the __arm__ code block,
but omitting the SWPB option since that definitely won't work.)
Back-patch to 9.2. The patch would work further back, but we'd also
need to update config.guess/config.sub in older branches to make them
build out-of-the-box, and there hasn't been demand for it.
Mark Salter
Isolate checksum calculation to its own module, so that bufpage
knows little if anything about the details of the calculation.
This implementation is a modified FNV-1a hash checksum, details
of which are given in the new checksum.c header comments.
Basic implementation only, so we fix the output value.
Later related commits will add version numbers to pg_control,
compiler optimization flags and memory barriers.
Ants Aasma, reviewed by Jeff Davis and Simon Riggs
Checksums are set immediately prior to flush out of shared buffers
and checked when pages are read in again. Hint bit setting will
require full page write when block is dirtied, which causes various
infrastructure changes. Extensive comments, docs and README.
WARNING message thrown if checksum fails on non-all zeroes page;
ERROR thrown but can be disabled with ignore_checksum_failure = on.
Feature enabled by an initdb option, since transition from option off
to option on is long and complex and has not yet been implemented.
Default is not to use checksums.
Checksum used is WAL CRC-32 truncated to 16-bits.
Simon Riggs, Jeff Davis, Greg Smith
Wide input and assistance from many community members. Thank you.
Remove use of PageSetTLI() from all page manipulation functions
and adjust README to indicate change in the way we make changes
to pages. Repurpose those bytes into the pd_checksum field and
explain how that works in comments about page header.
Refactoring ahead of actual feature patch which would make use
of the checksum field, arriving later.
Jeff Davis, with comments and doc changes by Simon Riggs
Direction suggested by Robert Haas; many others providing
review comments.
This GUC allows limiting the time spent waiting to acquire any one
heavyweight lock.
In support of this, improve the recently-added timeout infrastructure
to permit efficiently enabling or disabling multiple timeouts at once.
That reduces the performance hit from turning on lock_timeout, though
it's still not zero.
Zoltán Böszörményi, reviewed by Tom Lane,
Stephen Frost, and Hari Babu
This includes backend "COPY TO/FROM PROGRAM '...'" syntax, and corresponding
psql \copy syntax. Like with reading/writing files, the backend version is
superuser-only, and in the psql version, the program is run in the client.
In the passing, the psql \copy STDIN/STDOUT syntax is subtly changed: if you
the stdin/stdout is quoted, it's now interpreted as a filename. For example,
"\copy foo from 'stdin'" now reads from a file called 'stdin', not from
standard input. Before this, there was no way to specify a filename called
stdin, stdout, pstdin or pstdout.
This creates a new function in pgport, wait_result_to_str(), which can
be used to convert the exit status of a process, as returned by wait(3),
to a human-readable string.
Etsuro Fujita, reviewed by Amit Kapila.
This patch introduces two additional lock modes for tuples: "SELECT FOR
KEY SHARE" and "SELECT FOR NO KEY UPDATE". These don't block each
other, in contrast with already existing "SELECT FOR SHARE" and "SELECT
FOR UPDATE". UPDATE commands that do not modify the values stored in
the columns that are part of the key of the tuple now grab a SELECT FOR
NO KEY UPDATE lock on the tuple, allowing them to proceed concurrently
with tuple locks of the FOR KEY SHARE variety.
Foreign key triggers now use FOR KEY SHARE instead of FOR SHARE; this
means the concurrency improvement applies to them, which is the whole
point of this patch.
The added tuple lock semantics require some rejiggering of the multixact
module, so that the locking level that each transaction is holding can
be stored alongside its Xid. Also, multixacts now need to persist
across server restarts and crashes, because they can now represent not
only tuple locks, but also tuple updates. This means we need more
careful tracking of lifetime of pg_multixact SLRU files; since they now
persist longer, we require more infrastructure to figure out when they
can be removed. pg_upgrade also needs to be careful to copy
pg_multixact files over from the old server to the new, or at least part
of multixact.c state, depending on the versions of the old and new
servers.
Tuple time qualification rules (HeapTupleSatisfies routines) need to be
careful not to consider tuples with the "is multi" infomask bit set as
being only locked; they might need to look up MultiXact values (i.e.
possibly do pg_multixact I/O) to find out the Xid that updated a tuple,
whereas they previously were assured to only use information readily
available from the tuple header. This is considered acceptable, because
the extra I/O would involve cases that would previously cause some
commands to block waiting for concurrent transactions to finish.
Another important change is the fact that locking tuples that have
previously been updated causes the future versions to be marked as
locked, too; this is essential for correctness of foreign key checks.
This causes additional WAL-logging, also (there was previously a single
WAL record for a locked tuple; now there are as many as updated copies
of the tuple there exist.)
With all this in place, contention related to tuples being checked by
foreign key rules should be much reduced.
As a bonus, the old behavior that a subtransaction grabbing a stronger
tuple lock than the parent (sub)transaction held on a given tuple and
later aborting caused the weaker lock to be lost, has been fixed.
Many new spec files were added for isolation tester framework, to ensure
overall behavior is sane. There's probably room for several more tests.
There were several reviewers of this patch; in particular, Noah Misch
and Andres Freund spent considerable time in it. Original idea for the
patch came from Simon Riggs, after a problem report by Joel Jacobson.
Most code is from me, with contributions from Marti Raudsepp, Alexander
Shulgin, Noah Misch and Andres Freund.
This patch was discussed in several pgsql-hackers threads; the most
important start at the following message-ids:
AANLkTimo9XVcEzfiBR-ut3KVNDkjm2Vxh+t8kAmWjPuv@mail.gmail.com1290721684-sup-3951@alvh.no-ip.org1294953201-sup-2099@alvh.no-ip.org1320343602-sup-2290@alvh.no-ip.org1339690386-sup-8927@alvh.no-ip.org4FE5FF020200002500048A3D@gw.wicourts.gov4FEAB90A0200002500048B7D@gw.wicourts.gov
When relations are dropped, at end of transaction we need to remove the
files and clean the buffer pool of buffers containing pages of those
relations. Previously we would scan the buffer pool once per relation
to clean up buffers. When there are many relations to drop, the
repeated scans make this process slow; so we now instead pass a list of
relations to drop and scan the pool once, checking each buffer against
the passed list. When the number of relations is larger than a
threshold (which as of this patch is being set to 20 relations) we sort
the array before starting, and bsearch the array; when it's smaller, we
simply scan the array linearly each time, because that's faster. The
exact optimal threshold value depends on many factors, but the
difference is not likely to be significant enough to justify making it
user-settable.
This has been measured to be a significant win (a 15x win when dropping
100,000 relations; an extreme case, but reportedly a real one).
Author: Tomas Vondra, some tweaks by me
Reviewed by: Robert Haas, Shigeru Hanada, Andres Freund, Álvaro Herrera
The patch that turned XLogRecPtr into a uint64 inadvertently changed the
on-disk format of GiST indexes, because the NSN field in the GiST page
opaque is an XLogRecPtr. That breaks pg_upgrade. Revert the format of that
field back to the two-field struct that XLogRecPtr was before. This is the
same we did to LSNs in the page header to avoid changing on-disk format.
Bump catversion, as this invalidates any existing GiST indexes built on
9.3devel.
In situations where there are over 8MB of empty pages at the end of
a table, the truncation work for trailing empty pages takes longer
than deadlock_timeout, and there is frequent access to the table by
processes other than autovacuum, there was a problem with the
autovacuum worker process being canceled by the deadlock checking
code. The truncation work done by autovacuum up that point was
lost, and the attempt tried again by a later autovacuum worker. The
attempts could continue indefinitely without making progress,
consuming resources and blocking other processes for up to
deadlock_timeout each time.
This patch has the autovacuum worker checking whether it is
blocking any other thread at 20ms intervals. If such a condition
develops, the autovacuum worker will persist the work it has done
so far, release its lock on the table, and sleep in 50ms intervals
for up to 5 seconds, hoping to be able to re-acquire the lock and
try again. If it is unable to get the lock in that time, it moves
on and a worker will try to continue later from the point this one
left off.
While this patch doesn't change the rules about when and what to
truncate, it does cause the truncation to occur sooner, with less
blocking, and with the consumption of fewer resources when there is
contention for the table's lock.
The only user-visible change other than improved performance is
that the table size during truncation may change incrementally
instead of just once.
This problem exists in all supported versions but is infrequently
reported, although some reports of performance problems when
autovacuum runs might be caused by this. Initial commit is just the
master branch, but this should probably be backpatched once the
build farm and general developer usage confirm that there are no
surprising effects.
Jan Wieck
Background workers are postmaster subprocesses that run arbitrary
user-specified code. They can request shared memory access as well as
backend database connections; or they can just use plain libpq frontend
database connections.
Modules listed in shared_preload_libraries can register background
workers in their _PG_init() function; this is early enough that it's not
necessary to provide an extra GUC option, because the necessary extra
resources can be allocated early on. Modules can install more than one
bgworker, if necessary.
Care is taken that these extra processes do not interfere with other
postmaster tasks: only one such process is started on each ServerLoop
iteration. This means a large number of them could be waiting to be
started up and postmaster is still able to quickly service external
connection requests. Also, shutdown sequence should not be impacted by
a worker process that's reasonably well behaved (i.e. promptly responds
to termination signals.)
The current implementation lets worker processes specify their start
time, i.e. at what point in the server startup process they are to be
started: right after postmaster start (in which case they mustn't ask
for shared memory access), when consistent state has been reached
(useful during recovery in a HOT standby server), or when recovery has
terminated (i.e. when normal backends are allowed).
In case of a bgworker crash, actions to take depend on registration
data: if shared memory was requested, then all other connections are
taken down (as well as other bgworkers), just like it were a regular
backend crashing. The bgworker itself is restarted, too, within a
configurable timeframe (which can be configured to be never).
More features to add to this framework can be imagined without much
effort, and have been discussed, but this seems good enough as a useful
unit already.
An elementary sample module is supplied.
Author: Álvaro Herrera
This patch is loosely based on prior patches submitted by KaiGai Kohei,
and unsubmitted code by Simon Riggs.
Reviewed by: KaiGai Kohei, Markus Wanner, Andres Freund,
Heikki Linnakangas, Simon Riggs, Amit Kapila
Rename PGXACT->inCommit flag into delayChkpt flag,
and generalise comments to allow use in other situations,
such as the forthcoming potential use in checksum patch.
Replace wait loop to look for VXIDs with delayChkpt set.
No user visible changes, not behaviour changes at present.
Simon Riggs, reviewed and rebased by Jeff Davis
This reverts commit c11130690d in favor of
actually fixing the problem: namely, that we should never have been
modifying the checkpoint record's nextXid at this point to begin with.
The nextXid should match the state as of the checkpoint's logical WAL
position (ie the redo point), not the state as of its physical position.
It's especially bogus to advance it in some wal_levels and not others.
In any case there is no need for the checkpoint record to carry the
same nextXid shown in the XLOG_RUNNING_XACTS record just emitted by
LogStandbySnapshot, as any replay operation will already have adopted
that value as current.
This fixes bug #7710 from Tarvi Pillessaar, and probably also explains bug
#6291 from Daniel Farina, in that if a checkpoint were in progress at the
instant of XID wraparound, the epoch bump would be lost as reported.
(And, of course, these days there's at least a 50-50 chance of a checkpoint
being in progress at any given instant.)
Diagnosed by me and independently by Andres Freund. Back-patch to all
branches supporting hot standby.
Previously we stored all xids mixed together.
Now we store top-level xids first, followed
by all subxids. Also skip logging any subxids
if the snapshot is suboverflowed, since there
are potentially large numbers of them and they
are not useful in that case anyway. Has value
in the envisaged design for decoding of WAL.
No planned effect on Hot Standby.
Andres Freund, reviewed by me
Files opened with BasicOpenFile or PathNameOpenFile are not automatically
cleaned up on error. That puts unnecessary burden on callers that only want
to keep the file open for a short time. There is AllocateFile, but that
returns a buffered FILE * stream, which in many cases is not the nicest API
to work with. So add function called OpenTransientFile, which returns a
unbuffered fd that's cleaned up like the FILE* returned by AllocateFile().
This plugs a few rare fd leaks in error cases:
1. copy_file() - fixed by by using OpenTransientFile instead of BasicOpenFile
2. XLogFileInit() - fixed by adding close() calls to the error cases. Can't
use OpenTransientFile here because the fd is supposed to persist over
transaction boundaries.
3. lo_import/lo_export - fixed by using OpenTransientFile instead of
PathNameOpenFile.
In addition to plugging those leaks, this replaces many BasicOpenFile() calls
with OpenTransientFile() that were not leaking, because the code meticulously
closed the file on error. That wasn't strictly necessary, but IMHO it's good
for robustness.
The same leaks exist in older versions, but given the rarity of the issues,
I'm not backpatching this. Not yet, anyway - it might be good to backpatch
later, after this mechanism has had some more testing in master branch.
If the sleep is interrupted by a signal, we must recompute the remaining
time to wait; otherwise, a steady stream of non-wait-terminating interrupts
could delay return from WaitLatch indefinitely. This has been shown to be
a problem for the autovacuum launcher, and there may well be other places
now or in the future with similar issues. So we'd better make the function
robust, even though this'll add at least one gettimeofday call per wait.
Back-patch to 9.2. We might eventually need to fix 9.1 as well, but the
code is quite different there, and the usage of WaitLatch in 9.1 is so
limited that it's not clearly important to do so.
Reported and diagnosed by Jeff Janes, though I rewrote his patch rather
heavily.
If an SMgrRelation is not "owned" by a relcache entry, don't allow it to
live past transaction end. This design allows the same SMgrRelation to be
used for blind writes of multiple blocks during a transaction, but ensures
that we don't hold onto such an SMgrRelation indefinitely. Because an
SMgrRelation typically corresponds to open file descriptors at the fd.c
level, leaving it open when there's no corresponding relcache entry can
mean that we prevent the kernel from reclaiming deleted disk space.
(While CacheInvalidateSmgr messages usually fix that, there are cases
where they're not issued, such as DROP DATABASE. We might want to add
some more sinval messaging for that, but I'd be inclined to keep this
type of logic anyway, since allowing VFDs to accumulate indefinitely
for blind-written relations doesn't seem like a good idea.)
This code replaces a previous attempt towards the same goal that proved
to be unreliable. Back-patch to 9.1 where the previous patch was added.
This reverts commit fba105b109.
That approach had problems with the smgr-level state not tracking what
we really want to happen, and with the VFD-level state not tracking the
smgr-level state very well either. In consequence, it was still possible
to hold kernel file descriptors open for long-gone tables (as in recent
report from Tore Halset), and yet there were also cases of FDs being closed
undesirably soon. A replacement implementation will follow.
In the previous coding, new backend processes would attempt to create their
self-pipe during the OwnLatch call in InitProcess. However, pipe creation
could fail if the kernel is short of resources; and the system does not
recover gracefully from a FATAL error right there, since we have armed the
dead-man switch for this process and not yet set up the on_shmem_exit
callback that would disarm it. The postmaster then forces an unnecessary
database-wide crash and restart, as reported by Sean Chittenden.
There are various ways we could rearrange the code to fix this, but the
simplest and sanest seems to be to split out creation of the self-pipe into
a new function InitializeLatchSupport, which must be called from a place
where failure is allowed. For most processes that gets called in
InitProcess or InitAuxiliaryProcess, but processes that don't call either
but still use latches need their own calls.
Back-patch to 9.1, which has only a part of the latch logic that 9.2 and
HEAD have, but nonetheless includes this bug.
Do read/write permissions checks at most once per large object descriptor,
not once per lo_read or lo_write call as before. The repeated tests were
quite useless in the read case since the snapshot-based tests were
guaranteed to produce the same answer every time. In the write case,
the extra tests could in principle detect revocation of write privileges
after a series of writes has started --- but there's a race condition there
anyway, since we'd check privileges before performing and certainly before
committing the write. So there's no real advantage to checking every
single time, and we might as well redefine it as "only check the first
time".
On the same reasoning, remove the LargeObjectExists checks in inv_write
and inv_truncate. We already checked existence when the descriptor was
opened, and checking again doesn't provide any real increment of safety
that would justify the cost.
Get rid of the fundamentally indefensible assumption that "long long int"
exists and is exactly 64 bits wide on every platform Postgres runs on.
Instead let the configure script select the type to use for "pg_int64".
This is a bit of a pain in the rear since we do not want to pollute client
namespace with all the random symbols that pg_config.h defines; instead
we have to create a separate generated header file, "pg_config_ext.h".
But now that the infrastructure is there, we might have the ability to
add some other stuff that's long been wanting in this area.
4TB large objects (standard 8KB BLCKSZ case). For this purpose new
libpq API lo_lseek64, lo_tell64 and lo_truncate64 are added. Also
corresponding new backend functions lo_lseek64, lo_tell64 and
lo_truncate64 are added. inv_api.c is changed to handle 64-bit
offsets.
Patch contributed by Nozomi Anzai (backend side) and Yugo Nagata
(frontend side, docs, regression tests and example program). Reviewed
by Kohei Kaigai. Committed by Tatsuo Ishii with minor editings.
In all branches back to 8.3, this patch fixes a questionable assumption in
CompactCheckpointerRequestQueue/CompactBgwriterRequestQueue that there are
no uninitialized pad bytes in the request queue structs. This would only
cause trouble if (a) there were such pad bytes, which could happen in 8.4
and up if the compiler makes enum ForkNumber narrower than 32 bits, but
otherwise would require not-currently-planned changes in the widths of
other typedefs; and (b) the kernel has not uniformly initialized the
contents of shared memory to zeroes. Still, it seems a tad risky, and we
can easily remove any risk by pre-zeroing the request array for ourselves.
In addition to that, we need to establish a coding rule that struct
RelFileNode can't contain any padding bytes, since such structs are copied
into the request array verbatim. (There are other places that are assuming
this anyway, it turns out.)
In 9.1 and up, the risk was a bit larger because we were also effectively
assuming that struct RelFileNodeBackend contained no pad bytes, and with
fields of different types in there, that would be much easier to break.
However, there is no good reason to ever transmit fsync or delete requests
for temp files to the bgwriter/checkpointer, so we can revert the request
structs to plain RelFileNode, getting rid of the padding risk and saving
some marginal number of bytes and cycles in fsync queue manipulation while
we are at it. The savings might be more than marginal during deletion of
a temp relation, because the old code transmitted an entirely useless but
nonetheless expensive-to-process ForgetRelationFsync request to the
background process, and also had the background process perform the file
deletion even though that can safely be done immediately.
In addition, make some cleanup of nearby comments and small improvements to
the code in CompactCheckpointerRequestQueue/CompactBgwriterRequestQueue.
Management of timeouts was getting a little cumbersome; what we
originally had was more than enough back when we were only concerned
about deadlocks and query cancel; however, when we added timeouts for
standby processes, the code got considerably messier. Since there are
plans to add more complex timeouts, this seems a good time to introduce
a central timeout handling module.
External modules register their timeout handlers during process
initialization, and later enable and disable them as they see fit using
a simple API; timeout.c is in charge of keeping track of which timeouts
are in effect at any time, installing a common SIGALRM signal handler,
and calling setitimer() as appropriate to ensure timely firing of
external handlers.
timeout.c additionally supports pluggable modules to add their own
timeouts, though this capability isn't exercised anywhere yet.
Additionally, as of this commit, walsender processes are aware of
timeouts; we had a preexisting bug there that made those ignore SIGALRM,
thus being subject to unhandled deadlocks, particularly during the
authentication phase. This has already been fixed in back branches in
commit 0bf8eb2a, which see for more details.
Main author: Zoltán Böszörményi
Some review and cleanup by Álvaro Herrera
Extensive reworking by Tom Lane
There was a wild mix of calling conventions: Some were declared to
return void and didn't return, some returned an int exit code, some
claimed to return an exit code, which the callers checked, but
actually never returned, and so on.
Now all of these functions are declared to return void and decorated
with attribute noreturn and don't return. That's easiest, and most
code already worked that way.
This simplifies code that needs to do arithmetic on XLogRecPtrs.
To avoid changing on-disk format of data pages, the LSN on data pages is
still stored in the old format. That should keep pg_upgrade happy. However,
we have XLogRecPtrs embedded in the control file, and in the structs that
are sent over the replication protocol, so this changes breaks compatibility
of pg_basebackup and server. I didn't do anything about this in this patch,
per discussion on -hackers, the right thing to do would to be to change the
replication protocol to be architecture-independent, so that you could use
a newer version of pg_receivexlog, for example, against an older server
version.
This speeds up reassigning locks to the parent owner, when the transaction
holds a lot of locks, but only a few of them belong to the current resource
owner. This is particularly helps pg_dump when dumping a large number of
objects.
The cache can hold up to 15 locks in each resource owner. After that, the
cache is marked as overflowed, and we fall back to the old method of
scanning the whole local lock table. The tradeoff here is that the cache has
to be scanned whenever a lock is released, so if the cache is too large,
lock release becomes more expensive. 15 seems enough to cover pg_dump, and
doesn't have much impact on lock release.
Jeff Janes, reviewed by Amit Kapila and Heikki Linnakangas.
This provides a speedup of about 4X when NBuffers is large enough.
There is also a useful reduction in sinval traffic, since we
only do CacheInvalidateSmgr() once not once per fork.
Simon Riggs, reviewed and somewhat revised by Tom Lane
When the "hot" members of PGPROC were split off to separate PGXACT structs,
many PGPROC fields referred to in comments were moved to PGXACT, but the
comments were neglected in the commit. Mostly this is just a search/replace
of PGPROC with PGXACT, but the way the dummy PGPROC entries are created for
prepared transactions changed more, making some of the comments totally
bogus.
Noah Misch
Commit 6d90eaaa89 added a hibernation mode
to the bgwriter to reduce the server's idle-power consumption. However,
its interaction with the detailed behavior of BgBufferSync's feedback
control loop wasn't very well thought out. That control loop depends
primarily on the rate of buffer allocation, not the rate of buffer
dirtying, so the hibernation mode has to be designed to operate only when
no new buffer allocations are happening. Also, the check for whether the
system is effectively idle was not quite right and would fail to detect
a constant low level of activity, thus allowing the bgwriter to go into
hibernation mode in a way that would let the cycle time vary quite a bit,
possibly further confusing the feedback loop. To fix, move the wakeup
support from MarkBufferDirty and SetBufferCommitInfoNeedsSave into
StrategyGetBuffer, and prevent the bgwriter from entering hibernation mode
unless no buffer allocations have happened recently.
In addition, fix the delaying logic to remove the problem of possibly not
responding to signals promptly, which was basically caused by trying to use
the process latch's is_set flag for multiple purposes. I can't prove it
but I'm suspicious that that hack was responsible for the intermittent
"postmaster does not shut down" failures we've been seeing in the buildfarm
lately. In any case it did nothing to improve the readability or
robustness of the code.
In passing, express the hibernation sleep time as a multiplier on
BgWriterDelay, not a constant. I'm not sure whether there's any value in
exposing the longer sleep time as an independently configurable setting,
but we can at least make it act like this for little extra code.
This patch modifies the walwriter process so that, when it has not found
anything useful to do for many consecutive wakeup cycles, it extends its
sleep time to reduce the server's idle power consumption. It reverts to
normal as soon as it's done any successful flushes. It's still true that
during any async commit, backends check for completed, unflushed pages of
WAL and signal the walwriter if there are any; so that in practice the
walwriter can get awakened and returned to normal operation sooner than the
sleep time might suggest.
Also, improve the checkpointer so that it uses a latch and a computed delay
time to not wake up at all except when it has something to do, replacing a
previous hardcoded 0.5 sec wakeup cycle. This also is primarily useful for
reducing the server's power consumption when idle.
In passing, get rid of the dedicated latch for signaling the walwriter in
favor of using its procLatch, since that comports better with possible
generic signal handlers using that latch. Also, fix a pre-existing bug
with failure to save/restore errno in walwriter's signal handlers.
Peter Geoghegan, somewhat simplified by Tom
Commit 62c7bd31c8 had assorted problems, most
visibly that it broke PREPARE TRANSACTION in the presence of session-level
advisory locks (which should be ignored by PREPARE), as per a recent
complaint from Stephen Rees. More abstractly, the patch made the
LockMethodData.transactional flag not merely useless but outright
dangerous, because in point of fact that flag no longer tells you anything
at all about whether a lock is held transactionally. This fix therefore
removes that flag altogether. We now rely entirely on the convention
already in use in lock.c that transactional lock holds must be owned by
some ResourceOwner, while session holds are never so owned. Setting the
locallock struct's owner link to NULL thus denotes a session hold, and
there is no redundant marker for that.
PREPARE TRANSACTION now works again when there are session-level advisory
locks, and it is also able to transfer transactional advisory locks to the
prepared transaction, but for implementation reasons it throws an error if
we hold both types of lock on a single lockable object. Perhaps it will be
worth improving that someday.
Assorted other minor cleanup and documentation editing, as well.
Back-patch to 9.1, except that in the 9.1 branch I did not remove the
LockMethodData.transactional flag for fear of causing an ABI break for
any external code that might be examining those structs.
Remove the following ports:
- dgux
- nextstep
- sunos4
- svr4
- ultrix4
- univel
These are obsolete and not worth rescuing. In most cases, there is
circumstantial evidence that they wouldn't work anymore anyway.
The previous code could cause a backend crash after BEGIN; SAVEPOINT a;
LOCK TABLE foo (interrupted by ^C or statement timeout); ROLLBACK TO
SAVEPOINT a; LOCK TABLE foo, and might have leaked strong-lock counts
in other situations.
Report by Zoltán Böszörményi; patch review by Jeff Davis.
Postmaster sets max_safe_fds by testing how many open file descriptors it
can open, and that is normally inherited by all child processes at fork().
Not so on EXEC_BACKEND, ie. Windows, however. Because of that, we
effectively ignored max_files_per_process on Windows, and always assumed
a conservative default of 32 simultaneous open files. That could have an
impact on performance, if you need to access a lot of different files
in a query. After this patch, the value is passed to child processes by
save/restore_backend_variables() among many other global variables.
It has been like this forever, but given the lack of complaints about it,
I'm not backpatching this.
Currently, the only way to see the numbers this gathers is via
EXPLAIN (ANALYZE, BUFFERS), but the plan is to add visibility through
the stats collector and pg_stat_statements in subsequent patches.
Ants Aasma, reviewed by Greg Smith, with some further changes by me.
When a backend needs to flush the WAL, and someone else is already flushing
the WAL, wait until it releases the WALInsertLock and check if we still need
to do the flush or if the other backend already did the work for us, before
acquiring WALInsertLock. This helps group commit, because when the WAL flush
finishes, all the backends that were waiting for it can be woken up in one
go, and the can all concurrently observe that they're done, rather than
waking them up one by one in a cascading fashion.
This is based on a new LWLock function, LWLockWaitUntilFree(), which has
peculiar semantics. If the lock is immediately free, it grabs the lock and
returns true. If it's not free, it waits until it is released, but then
returns false without grabbing the lock. This is used in XLogFlush(), so
that when the lock is acquired, the backend flushes the WAL, but if it's
not, the backend first checks the current flush location before retrying.
Original patch and benchmarking by Peter Geoghegan and Simon Riggs, although
this patch as committed ended up being very different from that.
To make it wake up promptly when activity starts again, backends nudge it
by setting a latch in MarkBufferDirty(). The latch is kept set while
bgwriter is active, so there is very little overhead from that when the
system is busy. It is only armed before going into longer sleep.
Peter Geoghegan, with some changes by me.
We log AccessExclusiveLocks for replay onto standby nodes,
but because of timing issues on ProcArray it is possible to
log a lock that is still held by a just committed transaction
that is very soon to be removed. To avoid any timing issue we
avoid applying locks made by transactions with InvalidXid.
Simon Riggs, bug report Tom Lane, diagnosis Pavan Deolasee
Historically we've used the SWPB instruction for TAS() on ARM, but this
is deprecated and not available on ARMv6 and later. Instead, make use
of a GCC builtin if available. We'll still fall back to SWPB if not,
so as not to break existing ports using older GCC versions.
Eventually we might want to try using __sync_lock_test_and_set() on some
other architectures too, but for now that seems to present only risk and
not reward.
Back-patch to all supported versions, since people might want to use any
of them on more recent ARM chips.
Martin Pitt
Further testing convinces me that this is helpful at sufficiently high
contention levels, though it's still worrisome that it loses slightly
at lower contention levels.
Per Manabu Ori.
This is allegedly a win, at least on some PPC implementations, according
to the PPC ISA documents. However, as with LWARX hints, some PPC
platforms give an illegal-instruction failure. Use the same trick as
before of assuming that PPC64 platforms will accept it; we might need to
refine that based on experience, but there are other projects doing
likewise according to google.
I did not add an assembler compatibility test because LWSYNC has been
around much longer than hint bits, and it seems unlikely that any
toolchains currently in use don't recognize it.
Previously we defined slock_t as 8 bytes on PPC64, but the TAS assembly
code uses word-wide operations regardless, so that the second word was
just wasted space. There doesn't appear to be any performance benefit
in adding the second word, so get rid of it to simplify the code.
The hint bit makes for a small but measurable performance improvement
in access to contended spinlocks.
On the other hand, some PPC chips give an illegal-instruction failure.
There doesn't seem to be a completely bulletproof way to tell whether the
hint bit will cause an illegal-instruction failure other than by trying
it; but most if not all 64-bit PPC machines should accept it, so follow
the Linux kernel's lead and assume it's okay to use it in 64-bit builds.
Of course we must also check whether the assembler accepts the command,
since even with a recent CPU the toolchain could be old.
Patch by Manabu Ori, significantly modified by me.
This speeds up snapshot-taking and reduces ProcArrayLock contention.
Also, the PGPROC (and PGXACT) structures used by two-phase commit are
now allocated as part of the main array, rather than in a separate
array, and we keep ProcArray sorted in pointer order. These changes
are intended to minimize the number of cache lines that must be pulled
in to take a snapshot, and testing shows a substantial increase in
performance on both read and write workloads at high concurrencies.
Pavan Deolasee, Heikki Linnakangas, Robert Haas
This reverts commit 0180bd6180.
contrib/userlock is gone, but user-level locking still exists,
and is exposed via the pg_advisory* family of functions.
There was a timing window between when oldestActiveXid was derived
and when it should have been derived that only shows itself under
heavy load. Move code around to ensure correct timing of derivation.
No change to StartupSUBTRANS() code, which is where this failed.
Bug report by Chris Redekop
bgwriter is now a much less important process, responsible for page
cleaning duties only. checkpointer is now responsible for checkpoints
and so has a key role in shutdown. Later patches will correct doc
references to the now old idea that bgwriter performs checkpoints.
Has beneficial effect on performance at high write rates, but mainly
refactoring to more easily allow changes for power reduction by
simplifying previously tortuous code around required to allow page
cleaning and checkpointing to time slice in the same process.
Patch by me, Review by Dickson Guedes
We need not wait until the commit record is durably on disk, because
in the event of a crash the page we're updating with hint bits will
be gone anyway. Per off-list report from Heikki Linnakangas, this
can significantly degrade the performance of unlogged tables; I was
able to show a 2x speedup from this patch on a pgbench run with scale
factor 15. In practice, this will mostly help small, heavily updated
tables, because on larger tables you're unlikely to run into the same
row again before the commit record makes it out to disk.
A transaction can export a snapshot with pg_export_snapshot(), and then
others can import it with SET TRANSACTION SNAPSHOT. The data does not
leave the server so there are not security issues. A snapshot can only
be imported while the exporting transaction is still running, and there
are some other restrictions.
I'm not totally convinced that we've covered all the bases for SSI (true
serializable) mode, but it works fine for lesser isolation modes.
Joachim Wieland, reviewed by Marko Tiikkaja, and rather heavily modified
by Tom Lane
In REPEATABLE READ (nee SERIALIZABLE) mode, an attempt to do
GetTransactionSnapshot() between AbortTransaction and CleanupTransaction
failed, because GetTransactionSnapshot would recompute the transaction
snapshot (which is already wrong, given the isolation mode) and then
re-register it in the TopTransactionResourceOwner, leading to an Assert
because the TopTransactionResourceOwner should be empty of resources after
AbortTransaction. This is the root cause of bug #6218 from Yamamoto
Takashi. While changing plancache.c to avoid requesting a snapshot when
handling a ROLLBACK masks the problem, I think this is really a snapmgr.c
bug: it's lower-level than the resource manager mechanism and should not be
shutting itself down before we unwind resource manager resources. However,
just postponing the release of the transaction snapshot until cleanup time
didn't work because of the circular dependency with
TopTransactionResourceOwner. Fix by managing the internal reference to
that snapshot manually instead of depending on TopTransactionResourceOwner.
This saves a few cycles as well as making the module layering more
straightforward. predicate.c's dependencies on TopTransactionResourceOwner
go away too.
I think this is a longstanding bug, but there's no evidence that it's more
than a latent bug, so it doesn't seem worth any risk of back-patching.
This is not actually used anywhere yet, but it gets the basic
infrastructure in place. It is fairly likely that there are bugs, and
support for some important platforms may be missing, so we'll need to
refine this as we go along.
As per my recent proposal, this refactors things so that these typedefs and
macros are available in a header that can be included in frontend-ish code.
I also changed various headers that were undesirably including
utils/timestamp.h to include datatype/timestamp.h instead. Unsurprisingly,
this showed that half the system was getting utils/timestamp.h by way of
xlog.h.
No actual code changes here, just header refactoring.
walsender.h should depend on xlog.h, not vice versa. (Actually, the
inclusion was circular until a couple hours ago, which was even sillier;
but Bruce broke it in the expedient rather than logically correct
direction.) Because of that poor decision, plus blind application of
pgrminclude, we had a situation where half the system was depending on
xlog.h to include such unrelated stuff as array.h and guc.h. Clean up
the header inclusion, and manually revert a lot of what pgrminclude had
done so things build again.
This episode reinforces my feeling that pgrminclude should not be run
without adult supervision. Inclusion changes in header files in particular
need to be reviewed with great care. More generally, it'd be good if we
had a clearer notion of module layering to dictate which headers can sanely
include which others ... but that's a big task for another day.
storage/proc.h should not include replication/syncrep.h, especially not
when the latter includes storage/proc.h; but in any case this was a pretty
poor thing from a modular layering standpoint.
Per my testing, this works just as well with gcc as it does with HP's
compiler; and there is no reason to think that the effect doesn't occur
with icc, either.
Also, rewrite the header comment about enforcing sequencing around spinlock
operations, per Robert's gripe that it was misleading.
At least on this architecture, it's very important to spin on a
non-atomic instruction and only retry the atomic once it appears
that it will succeed. To fix this, split TAS() into two macros:
TAS(), for trying to grab the lock the first time, and TAS_SPIN(),
for spinning until we get it. TAS_SPIN() defaults to same as TAS(),
but we can override it when we know there's a better way.
It's likely that some of the other cases in s_lock.h require
similar treatment, but this is the only one we've got conclusive
evidence for at present.
This requires adjusting the API for syscache callback functions: they now
get a hash value, not a TID, to identify the target tuple. Most of them
weren't paying any attention to that argument anyway, but plancache did
require a small amount of fixing.
Also, improve performance a trifle by avoiding sending duplicate inval
messages when a heap_update isn't changing the catcache lookup columns.
In pursuit of this (and with the expectation that WaitLatch will be needed
in more places), convert the latch field that was already added to PGPROC
for sync rep into a generic latch that is activated for all PGPROC-owning
processes, and change many of the standard backend signal handlers to set
that latch when a signal happens. This will allow WaitLatch callers to be
wakened properly by these signals.
In passing, fix a whole bunch of signal handlers that had been hacked to do
things that might change errno, without adding the necessary save/restore
logic for errno. Also make some minor fixes in unix_latch.c, and clean
up bizarre and unsafe scheme for disowning the process's latch. Much of
this has to be back-patched into 9.1.
Peter Geoghegan, with additional work by Tom
Improve the documentation around weak-memory-ordering risks, and do a pass
of general editorialization on the comments in the latch code. Make the
Windows latch code more like the Unix latch code where feasible; in
particular provide the same Assert checks in both implementations.
Fix poorly-placed WaitLatch call in syncrep.c.
This patch resolves, for the moment, concerns around weak-memory-ordering
bugs in latch-related code: we have documented the restrictions and checked
that existing calls meet them. In 9.2 I hope that we will install suitable
memory barrier instructions in SetLatch/ResetLatch, so that their callers
don't need to be quite so careful.
Instead of entering them on transaction startup, we materialize them
only when someone wants to wait, which will occur only during CREATE
INDEX CONCURRENTLY. In Hot Standby mode, the startup process must also
be able to probe for conflicting VXID locks, but the lock need never be
fully materialized, because the startup process does not use the normal
lock wait mechanism. Since most VXID locks never need to touch the
lock manager partition locks, this can significantly reduce blocking
contention on read-heavy workloads.
Patch by me. Review by Jeff Davis.
This kluge was inserted in a spot apparently chosen at random: the lock
manager's state is not yet fully set up for the wait, and in particular
LockWaitCancel hasn't been armed by setting lockAwaited, so the ProcLock
will not get cleaned up if the ereport is thrown. This seems to not cause
any observable problem in trivial test cases, because LockReleaseAll will
silently clean up the debris; but I was able to cause failures with tests
involving subtransactions.
Fixes breakage induced by commit c85c941470.
Back-patch to all affected branches.
It was initialized in the wrong place and to the wrong value. With bad
luck this could result in incorrect query-cancellation failures in hot
standby sessions, should a HS backend be holding pin on buffer number 1
while trying to acquire a lock.
When an AccessShareLock, RowShareLock, or RowExclusiveLock is requested
on an unshared database relation, and we can verify that no conflicting
locks can possibly be present, record the lock in a per-backend queue,
stored within the PGPROC, rather than in the primary lock table. This
eliminates a great deal of contention on the lock manager LWLocks.
This patch also refactors the interface between GetLockStatusData() and
pg_lock_status() to be a bit more abstract, so that we don't rely so
heavily on the lock manager's internal representation details. The new
fast path lock structures don't have a LOCK or PROCLOCK structure to
return, so we mustn't depend on that for purposes of listing outstanding
locks.
Review by Jeff Davis.
In the previous coding, we would look up a relation in RangeVarGetRelid,
lock the resulting OID, and then AcceptInvalidationMessages(). While
this was sufficient to ensure that we noticed any changes to the
relation definition before building the relcache entry, it didn't
handle the possibility that the name we looked up no longer referenced
the same OID. This was particularly problematic in the case where a
table had been dropped and recreated: we'd latch on to the entry for
the old relation and fail later on. Now, we acquire the relation lock
inside RangeVarGetRelid, and retry the name lookup if we notice that
invalidation messages have been processed meanwhile. Many operations
that would previously have failed with an error in the presence of
concurrent DDL will now succeed.
There is a good deal of work remaining to be done here: many callers
of RangeVarGetRelid still pass NoLock for one reason or another. In
addition, nothing in this patch guards against the possibility that
the meaning of an unqualified name might change due to the creation
of a relation in a schema earlier in the user's search path than the
one where it was previously found. Furthermore, there's nothing at
all here to guard against similar race conditions for non-relations.
For all that, it's a start.
Noah Misch and Robert Haas
detect postmaster death. Postmaster keeps the write-end of the pipe open,
so when it dies, children get EOF in the read-end. That can conveniently
be waited for in select(), which allows eliminating some of the polling
loops that check for postmaster death. This patch doesn't yet change all
the loops to use the new mechanism, expect a follow-on patch to do that.
This changes the interface to WaitLatch, so that it takes as argument a
bitmask of events that it waits for. Possible events are latch set, timeout,
postmaster death, and socket becoming readable or writeable.
The pipe method behaves slightly differently from the kill() method
previously used in PostmasterIsAlive() in the case that postmaster has died,
but its parent has not yet read its exit code with waitpid(). The pipe
returns EOF as soon as the process dies, but kill() continues to return
true until waitpid() has been called (IOW while the process is a zombie).
Because of that, change PostmasterIsAlive() to use the pipe too, otherwise
WaitLatch() would return immediately with WL_POSTMASTER_DEATH, while
PostmasterIsAlive() would claim it's still alive. That could easily lead to
busy-waiting while postmaster is in zombie state.
Peter Geoghegan with further changes by me, reviewed by Fujii Masao and
Florian Pflug.
transactions might not match the order the work done in those transactions
become visible to others. The logic in SSI, however, assumed that it does.
Fix that by having two sequence numbers for each serializable transaction,
one taken before a transaction becomes visible to others, and one after it.
This is easier than trying to make the the transition totally atomic, which
would require holding ProcArrayLock and SerializableXactHashLock at the same
time. By using prepareSeqNo instead of commitSeqNo in a few places where
commit sequence numbers are compared, we can make those comparisons err on
the safe side when we don't know for sure which committed first.
Per analysis by Kevin Grittner and Dan Ports, but this approach to fix it
is different from the original patch.
As Tom Lane pointed out, "const Relation foo" doesn't guarantee that you
can't modify the data the "foo" pointer points to. It just means that you
can't change the pointer to point to something else within the function,
which is not very useful.
MARKED_FOR_DEATH flags into one. We still need the ROLLED_BACK flag to
mark transactions that are in the process of being rolled back. To be
precise, ROLLED_BACK now means that a transaction has already been
discounted from the count of transactions with the oldest xmin, but not
yet removed from the list of active transactions.
Dan Ports
the marked-for-death flag. It was only set for a fleeting moment while a
transaction was being cleaned up at rollback. All the places that checked
for the rolled-back flag should also check the marked-for-death flag, as
both flags mean that the transaction will roll back. I also renamed the
marked-for-death into "doomed", which is a lot shorter name.
snapshots, like in REINDEX, are basically non-transactional operations. The
DDL operation itself might participate in SSI, but there's separate
functions for that.
Kevin Grittner and Dan Ports, with some changes by me.
Even if a flag is modified only by the backend owning the transaction, it's
not safe to modify it without a lock. Another backend might be setting or
clearing a different flag in the flags field concurrently, and that
operation might be lost because setting or clearing a bit in a word is not
atomic.
Make did-write flag a simple backend-private boolean variable, because it
was only set or tested in the owning backend (except when committing a
prepared transaction, but it's not worthwhile to optimize for the case of a
read-only prepared transaction). This also eliminates the need to add
locking where that flag is set.
Also, set the did-write flag when doing DDL operations like DROP TABLE or
TRUNCATE -- that was missed earlier.
"Blind writes" are a mechanism to push buffers down to disk when
evicting them; since they may belong to different databases than the one
a backend is connected to, the backend does not necessarily have a
relation to link them to, and thus no way to blow them away. We were
keeping those files open indefinitely, which would cause a problem if
the underlying table was deleted, because the operating system would not
be able to reclaim the disk space used by those files.
To fix, have bufmgr mark such files as transient to smgr; the lower
layer is allowed to close the file descriptor when the current
transaction ends. We must be careful to have any other access of the
file to remove the transient markings, to prevent unnecessary expensive
system calls when evicting buffers belonging to our own database (which
files we're likely to require again soon.)
This commit fixes a bug in the previous one, which neglected to cleanly
handle the LRU ring that fd.c uses to manage open files, and caused an
unacceptable failure just before beta2 and was thus reverted.
"Blind writes" are a mechanism to push buffers down to disk when
evicting them; since they may belong to different databases than the one
a backend is connected to, the backend does not necessarily have a
relation to link them to, and thus no way to blow them away. We were
keeping those files open indefinitely, which would cause a problem if
the underlying table was deleted, because the operating system would not
be able to reclaim the disk space used by those files.
To fix, have bufmgr mark such files as transient to smgr; the lower
layer is allowed to close the file descriptor when the current
transaction ends. We must be careful to have any other access of the
file to remove the transient markings, to prevent unnecessary expensive
system calls when evicting buffers belonging to our own database (which
files we're likely to require again soon.)
Truncating or dropping a table is treated like deletion of all tuples, and
check for conflicts accordingly. If a table is clustered or rewritten by
ALTER TABLE, all predicate locks on the heap are promoted to relation-level
locks, because the tuple or page ids of any existing tuples will change and
won't be valid after rewriting the table. Arguably ALTER TABLE should be
treated like a mass-UPDATE of every row, but if you e.g change the datatype
of a column, you could also argue that it's just a change to the physical
layout, not a logical change. Reindexing promotes all locks on the index to
relation-level lock on the heap.
Kevin Grittner, with a lot of cosmetic changes by me.
On further analysis, it turns out that it is not needed to duplicate predicate
locks to the new row version at update, the lock on the version that the
transaction saw as visible is enough. However, there was a different bug in
the code that checks for dangerous structures when a new rw-conflict happens.
Fix that bug, and remove all the row-version chaining related code.
Kevin Grittner & Dan Ports, with some comment editorialization by me.
If a smart shutdown occurs just as a child is starting up, and the
child subsequently becomes a walsender, there is a race condition:
the postmaster might count the exstant backends, determine that there
is one normal backend, and wait for it to die off. Had the walsender
transition already occurred before the postmaster counted, it would
have proceeded with the shutdown.
To fix this, have each child that transforms into a walsender kick
the postmaster just after doing so, so that the state machine is
certain to advance.
Fujii Masao
than replication_timeout (a new GUC) milliseconds. The TCP timeout is often
too long, you want the master to notice a dead connection much sooner.
People complained about that in 9.0 too, but with synchronous replication
it's even more important to notice dead connections promptly.
Fujii Masao and Heikki Linnakangas
than doing it aggressively whenever the tail-XID pointer is advanced, because
this way we don't need to do it while holding SerializableXactHashLock.
This also fixes bug #5915 spotted by YAMAMOTO Takashi, and removes an
obsolete comment spotted by Kevin Grittner.
If a standby is broadcasting reply messages and we have named
one or more standbys in synchronous_standby_names then allow
users who set synchronous_replication to wait for commit, which
then provides strict data integrity guarantees. Design avoids
sending and receiving transaction state information so minimises
bookkeeping overheads. We synchronize with the highest priority
standby that is connected and ready to synchronize. Other standbys
can be defined to takeover in case of standby failure.
This version has very strict behaviour; more relaxed options
may be added at a later date.
Simon Riggs and Fujii Masao, with reviews by Yeb Havinga, Jaime
Casanova, Heikki Linnakangas and Robert Haas, plus the assistance
of many other design reviewers.
Change the way UPDATEs are handled. Instead of maintaining a chain of
tuple-level locks in shared memory, copy any existing locks on the old
tuple to the new tuple at UPDATE. Any existing page-level lock needs to
be duplicated too, as a lock on the new tuple. That was neglected
previously.
Store xmin on tuple-level predicate locks, to distinguish a lock on an old
already-recycled tuple from a new tuple at the same physical location.
Failure to distinguish them caused loops in the tuple-lock chains, as
reported by YAMAMOTO Takashi. Although we don't use the chain representation
of UPDATEs anymore, it seems like a good idea to store the xmin to avoid
some false positives if no other reason.
CheckSingleTargetForConflictsIn now correctly handles the case where a lock
that's being held is not reflected in the local lock table. That happens
if another backend acquires a lock on our behalf due to an UPDATE or a page
split.
PredicateLockPageCombine now retains locks for the page that is being
removed, rather than removing them. This prevents a potentially dangerous
false-positive inconsistency where the local lock table believes that a lock
is held, but it is actually not.
Dan Ports and Kevin Grittner
They share the same locking namespace with the existing session-level
advisory locks, but they are automatically released at the end of the
current transaction and cannot be released explicitly via unlock
functions.
Marko Tiikkaja, reviewed by me.
Until now, our Serializable mode has in fact been what's called Snapshot
Isolation, which allows some anomalies that could not occur in any
serialized ordering of the transactions. This patch fixes that using a
method called Serializable Snapshot Isolation, based on research papers by
Michael J. Cahill (see README-SSI for full references). In Serializable
Snapshot Isolation, transactions run like they do in Snapshot Isolation,
but a predicate lock manager observes the reads and writes performed and
aborts transactions if it detects that an anomaly might occur. This method
produces some false positives, ie. it sometimes aborts transactions even
though there is no anomaly.
To track reads we implement predicate locking, see storage/lmgr/predicate.c.
Whenever a tuple is read, a predicate lock is acquired on the tuple. Shared
memory is finite, so when a transaction takes many tuple-level locks on a
page, the locks are promoted to a single page-level lock, and further to a
single relation level lock if necessary. To lock key values with no matching
tuple, a sequential scan always takes a relation-level lock, and an index
scan acquires a page-level lock that covers the search key, whether or not
there are any matching keys at the moment.
A predicate lock doesn't conflict with any regular locks or with another
predicate locks in the normal sense. They're only used by the predicate lock
manager to detect the danger of anomalies. Only serializable transactions
participate in predicate locking, so there should be no extra overhead for
for other transactions.
Predicate locks can't be released at commit, but must be remembered until
all the transactions that overlapped with it have completed. That means that
we need to remember an unbounded amount of predicate locks, so we apply a
lossy but conservative method of tracking locks for committed transactions.
If we run short of shared memory, we overflow to a new "pg_serial" SLRU
pool.
We don't currently allow Serializable transactions in Hot Standby mode.
That would be hard, because even read-only transactions can cause anomalies
that wouldn't otherwise occur.
Serializable isolation mode now means the new fully serializable level.
Repeatable Read gives you the old Snapshot Isolation level that we have
always had.
Kevin Grittner and Dan Ports, reviewed by Jeff Davis, Heikki Linnakangas and
Anssi Kääriäinen
backend, as far as the postmaster shutdown logic is concerned. That means,
fast shutdown will wait for WAL sender processes to exit before signaling
bgwriter to finish. This avoids race conditions between a base backup stopping
or starting, and bgwriter writing the shutdown checkpoint WAL record. We don't
want e.g the end-of-backup WAL record to be written after the shutdown
checkpoint.
The contents of an unlogged table are WAL-logged; thus, they are not
available on standby servers and are truncated whenever the database
system enters recovery. Indexes on unlogged tables are also unlogged.
Unlogged GiST indexes are not currently supported.
First, avoid scanning the whole ProcArray once we know there
are at least commit_siblings active; second, skip the check
altogether if commit_siblings = 0.
Greg Smith
an old transaction running in the master, and a lot of transactions have
started and finished since, and a WAL-record is written in the gap between
the creating the running-xacts snapshot and WAL-logging it, recovery will fail
with "too many KnownAssignedXids" error. This bug was reported by
Joachim Wieland on Nov 19th.
In the same scenario, when fewer transactions have started so that all the
xids fit in KnownAssignedXids despite the first bug, a more serious bug
arises. We incorrectly initialize the clog code with the oldest still running
transaction, and when we see the WAL record belonging to a transaction with
an XID larger than one that committed already before the checkpoint we're
recovering from, we zero the clog page containing the already committed
transaction, leading to data loss.
In hindsight, trying to track xids in the known-assigned-xids array before
seeing the running-xacts record was too complicated. To fix that, hold
XidGenLock while the running-xacts snapshot is taken and WAL-logged. That
ensures that no transaction can begin or end in that gap, so that in recvoery
we know that the snapshot contains all transactions running at that point in
WAL.
removing an infrequently occurring race condition in Hot Standby.
An xid must be assigned before a lock appears in shared memory,
rather than immediately after, else GetRunningTransactionLocks()
may see InvalidTransactionId, causing assertion failures during
lock processing on standby.
Bug report and diagnosis by Fujii Masao, fix by me.
Having this in src/include/port.h makes no sense, now that copydir.c lives
in src/backend/strorage rather than src/port. Along the way, remove an
obsolete comment from contrib/pg_upgrade that makes reference to the old
location.
dynamic pool of event handles, we can permanently assign one for each
shared latch. Thanks to that, we no longer need a separate shared memory
block for latches, and we don't need to know in advance how many shared
latches there is, so you no longer need to remember to update
NumSharedLatches when you introduce a new latch to the system.
wait until it is set. Latches can be used to reliably wait until a signal
arrives, which is hard otherwise because signals don't interrupt select()
on some platforms, and even when they do, there's race conditions.
On Unix, latches use the so called self-pipe trick under the covers to
implement the sleep until the latch is set, without race conditions. On
Windows, Windows events are used.
Use the new latch abstraction to sleep in walsender, so that as soon as
a transaction finishes, walsender is woken up to immediately send the WAL
to the standby. This reduces the latency between master and standby, which
is good.
Preliminary work by Fujii Masao. The latch implementation is by me, with
helpful comments from many people.
There is no reason that proc.c should have to get involved in this dirty hack
for letting the postmaster know which children are walsenders. Revert that
file to the way it was, and confine the kluge to pmsignal.c and postmaster.c.
This allows us to reliably remove all leftover temporary relation
files on cluster startup without reference to system catalogs or WAL;
therefore, we no longer include temporary relations in XLOG_XACT_COMMIT
and XLOG_XACT_ABORT WAL records.
Since these changes require including a backend ID in each
SharedInvalSmgrMsg, the size of the SharedInvalidationMessage.id
field has been reduced from two bytes to one, and the maximum number
of connections has been reduced from INT_MAX / 4 to 2^23-1. It would
be possible to remove these restrictions by increasing the size of
SharedInvalidationMessage by 4 bytes, but right now that doesn't seem
like a good trade-off.
Review by Jaime Casanova and Tom Lane.
max_standby_streaming_delay, and revise the implementation to avoid assuming
that timestamps found in WAL records can meaningfully be compared to clock
time on the standby server. Instead, the delay limits are compared to the
elapsed time since we last obtained a new WAL segment from archive or since
we were last "caught up" to WAL data arriving via streaming replication.
This avoids problems with clock skew between primary and standby, as well
as other corner cases that the original coding would misbehave in, such
as the primary server having significant idle time between transactions.
Per my complaint some time ago and considerable ensuing discussion.
Do some desultory editing on the hot standby documentation, too.
During Hot Standby we need to check for buffer pin deadlocks when the
Startup process begins to wait, in case it never wakes up again. We
previously made the deadlock check immediately on the basis it was
cheap, though clearer thinking and prima facie evidence shows that
was too simple. Refactor existing code to make it easy to add in
deferral of deadlock check until deadlock_timeout allowing a good
reduction in deadlock checks since far few buffer pins are held for
that duration. It's worth doing anyway, though major goal is to
prevent further reports of context switching with high numbers of
users on occasional tests.
of requirements and documentation on LogStandbySnapshot(). Fixes
two minor bugs reported by Tom Lane that would lead to an incorrect
snapshot after transaction wraparound. Also fix two other problems
discovered that would give incorrect snapshots in certain cases.
ProcArrayApplyRecoveryInfo() substantially rewritten. Some minor
refactoring of xact_redo_apply() and ExpireTreeKnownAssignedTransactionIds().
In addition, add support for a "payload" string to be passed along with
each notify event.
This implementation should be significantly more efficient than the old one,
and is also more compatible with Hot Standby usage. There is not yet any
facility for HS slaves to receive notifications generated on the master,
although such a thing is possible in future.
Joachim Wieland, reviewed by Jeff Davis; also hacked on by me.
all the data and using posix_fadvise to nudge the OS into flushing it
earlier. This also hopefully makes CREATE DATABASE avoid spamming the
cache.
Tests show a big speedup on Linux at least on some filesystems.
Idea and patch from Andres Freund.
where a database has a non-default tablespaceid. Pass thru MyDatabaseId
and MyDatabaseTableSpace to allow file path to be re-created in
standby and correct invalidation to take place in all cases.
Update and rework xact_commit_desc() debug messages.
Bug report from Tom by code inspection. Fix by me.
process. If startup waits on a buffer pin we send a request to all
backends to cancel themselves if they are holding the buffer pin
required and they are also waiting on a lock. If not, startup waits
until max_standby_delay before cancelling any backend waiting for
the requested buffer pin.
Move rd_targblock, rd_fsm_nblocks, and rd_vm_nblocks from relcache to the smgr
relation entries, so that they will get reset to InvalidBlockNumber whenever
an smgr-level flush happens. Because we now send smgr invalidation messages
immediately (not at end of transaction) when a relation truncation occurs,
this ensures that other backends will reset their values before they next
access the relation. We no longer need the unreliable assumption that a
VACUUM that's doing a truncation will hold its AccessExclusive lock until
commit --- in fact, we can intentionally release that lock as soon as we've
completed the truncation. This patch therefore reverts (most of) Alvaro's
patch of 2009-11-10, as well as my marginal hacking on it yesterday. We can
also get rid of assorted no-longer-needed relcache flushes, which are far more
expensive than an smgr flush because they kill a lot more state.
In passing this patch fixes smgr_redo's failure to perform visibility-map
truncation, and cleans up some rather dubious assumptions in freespace.c and
visibilitymap.c about when rd_fsm_nblocks and rd_vm_nblocks can be out of
date.
of shared or nailed system catalogs. This has two key benefits:
* The new CLUSTER-based VACUUM FULL can be applied safely to all catalogs.
* We no longer have to use an unsafe reindex-in-place approach for reindexing
shared catalogs.
CLUSTER on nailed catalogs now works too, although I left it disabled on
shared catalogs because the resulting pg_index.indisclustered update would
only be visible in one database.
Since reindexing shared system catalogs is now fully transactional and
crash-safe, the former special cases in REINDEX behavior have been removed;
shared catalogs are treated the same as non-shared.
This commit does not do anything about the recently-discussed problem of
deadlocks between VACUUM FULL/CLUSTER on a system catalog and other
concurrent queries; will address that in a separate patch. As a stopgap,
parallel_schedule has been tweaked to run vacuum.sql by itself, to avoid
such failures during the regression tests.
stage of required deadlock detection to allow re-enabling max_standby_delay
setting of -1, which is now essential in the absence of improved relation-
specific conflict resoluton. Requested by Greg Stark et al.
records for heap and btree. Minor change, mostly API changes to
pass through the required values. This is a simple change though
also provides the refactoring required for further enhancements
to conflict processing using the relOid. Changes only have effect
during Hot Standby.
restore_command, if the connection to the primary server is lost. This
ensures that the standby can recover automatically, if the connection is
lost for a long time and standby falls behind so much that the required
WAL segments have been archived and deleted in the master.
This also makes standby_mode useful without streaming replication; the
server will keep retrying restore_command every few seconds until the
trigger file is found. That's the same basic functionality pg_standby
offers, but without the bells and whistles.
To implement that, refactor the ReadRecord/FetchRecord functions. The
FetchRecord() function introduced in the original streaming replication
patch is removed, and all the retry logic is now in a new function called
XLogReadPage(). XLogReadPage() is now responsible for executing
restore_command, launching walreceiver, and waiting for new WAL to arrive
from primary, as required.
This also changes the life cycle of walreceiver. When launched, it now only
tries to connect to the master once, and exits if the connection fails, or
is lost during streaming for any reason. The startup process detects the
death, and re-launches walreceiver if necessary.
woken by alarm we send SIGUSR1 to all backends requesting that they
check to see if they are blocking Startup process. If so, they throw
ERROR/FATAL as for other conflict resolutions. Deadlock stop gap
removed. max_standby_delay = -1 option removed to prevent deadlock.
binary, revert PGDLLIMPORT decoration of global variables. I'm not sure
if there's any real harm from unnecessary PGDLLIMPORTs, but these are all
internal variables that external modules really shouldn't be messing
with. ThisTimeLineID still needs PGDLLIMPORT.
Conflict reason is passed through directly to the backend, so we can
take decisions about the effect of the conflict based upon the local
state. No specific changes, as yet, though this prepares for later work.
CancelVirtualTransaction() sends signals while holding ProcArrayLock.
Introduce errdetail_abort() to give message detail explaining that the
abort was caused by conflict processing. Remove CONFLICT_MODE states
in favour of using PROCSIG_RECOVERY_CONFLICT states directly, for clarity.
This includes two new kinds of postmaster processes, walsenders and
walreceiver. Walreceiver is responsible for connecting to the primary server
and streaming WAL to disk, while walsender runs in the primary server and
streams WAL from disk to the client.
Documentation still needs work, but the basics are there. We will probably
pull the replication section to a new chapter later on, as well as the
sections describing file-based replication. But let's do that as a separate
patch, so that it's easier to see what has been added/changed. This patch
also adds a new section to the chapter about FE/BE protocol, documenting the
protocol used by walsender/walreceivxer.
Bump catalog version because of two new functions,
pg_last_xlog_receive_location() and pg_last_xlog_replay_location(), for
monitoring the progress of replication.
Fujii Masao, with additional hacking by me
of this are to centralise the conflict code to allow further change,
as well as to allow passing through the full reason for the conflict
through to the conflicting backends. Backend state alters how we
can handle different types of conflict so this is now required.
As originally suggested by Heikki, no longer optional.
Previously we only cancelled sessions that were in-transaction.
Simple fix is to just cancel all sessions without waiting. Doing
it this way avoids complicating common code paths, which would
not be worth the trouble to cover this rare case.
Problem report and fix by Andres Freund, edited somewhat by me
deletion, so that we attempt to unlink the correct filepath. unlink()
errors are ignorable there, so lack of a DatabasePath initialization step
did not cause visible problems until a related bug showed up on Solaris.
Code refactored from xact_redo_commit() to
ProcessCommittedInvalidationMessages() in inval.c. Recovery may replay
shared invalidation messages for many databases, so we cannot
SetDatabasePath() once as we do in normal backends. Read the databaseid
from the shared invalidation messages, then set DatabasePath
temporarily before calling RelationCacheInitFileInvalidate().
Problem report by Robert Treat, analysis and fix by me.
This is more in keeping with modern practice, and is a first step towards
porting to Win64 (which has sizeof(pointer) > sizeof(long)).
Tsutomu Yamada, Magnus Hagander, Tom Lane
Enabled by recovery_connections = on (default) and forcing archive recovery using a recovery.conf. Recovery processing now emulates the original transactions as they are replayed, providing full locking and MVCC behaviour for read only queries. Recovery must enter consistent state before connections are allowed, so there is a delay, typically short, before connections succeed. Replay of recovering transactions can conflict and in some cases deadlock with queries during recovery; these result in query cancellation after max_standby_delay seconds have expired. Infrastructure changes have minor effects on normal running, though introduce four new types of WAL record.
New test mode "make standbycheck" allows regression tests of static command behaviour on a standby server while in recovery. Typical and extreme dynamic behaviours have been checked via code inspection and manual testing. Few port specific behaviours have been utilised, though primary testing has been on Linux only so far.
This commit is the basic patch. Additional changes will follow in this release to enhance some aspects of behaviour, notably improved handling of conflicts, deadlock detection and query cancellation. Changes to VACUUM FULL are also required.
Simon Riggs, with significant and lengthy review by Heikki Linnakangas, including streamlined redesign of snapshot creation and two-phase commit.
Important contributions from Florian Pflug, Mark Kirkwood, Merlin Moncure, Greg Stark, Gianni Ciolli, Gabriele Bartolini, Hannu Krosing, Robert Haas, Tatsuo Ishii, Hiroyuki Yamada plus support and feedback from many other community members.
This patch also removes buffer-usage statistics from the track_counts
output, since this (or the global server statistics) is deemed to be a better
interface to this information.
Itagaki Takahiro, reviewed by Euler Taveira de Oliveira.
via the "flat files" facility. This requires making it enough like a backend
to be able to run transactions; it's no longer an "auxiliary process" but
more like the autovacuum worker processes. Also, its signal handling has
to be brought into line with backends/workers. In particular, since it
now has to handle procsignal.c processing, the special autovac-launcher-only
signal conditions are moved to SIGUSR2.
Alvaro, with some cleanup from Tom
(That flat file is now completely useless, but removal will come later.)
To do this, postpone client authentication into the startup transaction
that's run by InitPostgres. We still collect the startup packet and do
SSL initialization (if needed) at the same time we did before. The
AuthenticationTimeout is applied separately to startup packet collection
and the actual authentication cycle. (This is a bit annoying, since it
means a couple extra syscalls; but the signal handling requirements inside
and outside a transaction are sufficiently different that it seems best
to treat the timeouts as completely independent.)
A small security disadvantage is that if the given database name is invalid,
this will be reported to the client before any authentication happens.
We could work around that by connecting to database "postgres" instead,
but consensus seems to be that it's not worth introducing such surprising
behavior.
Processing of all command-line switches and GUC options received from the
client is now postponed until after authentication. This means that
PostAuthDelay is much less useful than it used to be --- if you need to
investigate problems during InitPostgres you'll have to set PreAuthDelay
instead. However, allowing an unauthenticated user to set any GUC options
whatever seems a bit too risky, so we'll live with that.
To make this work in the base case, pg_database now has a nailed-in-cache
relation descriptor that is initialized using hardwired knowledge in
relcache.c. This means pg_database is added to the set of relations that
need to have a Schema_pg_xxx macro maintained in pg_attribute.h. When this
path is taken, we'll have to do a seqscan of pg_database to find the row
we need.
In the normal case, we are able to do an indexscan to find the database's row
by name. This is made possible by storing a global relcache init file that
describes only the shared catalogs and their indexes (and therefore is usable
by all backends in any database). A new backend loads this cache file,
finds its database OID after an indexscan on pg_database, and then loads
the local relcache init file for that database.
This change should effectively eliminate number of databases as a factor
in backend startup time, even with large numbers of databases. However,
the real reason for doing it is as a first step towards getting rid of
the flat files altogether. There are still several other sub-projects
to be tackled before that can happen.
fsync() fails, say "file" rather than "relation" when printing the filename.
This makes messages that display block numbers a bit confusing. For example,
in message 'could not read block 150000 of file "base/1234/5678.1"', 150000
is the block number from the beginning of the relation, ie. segment 0, not
150000th block within that segment. Per discussion, users aren't usually
interested in the exact location within the file, so we can live with that.
To ease constructing error messages, add FilePathName(File) function to
return the pathname of a virtual fd.
This patch gets us out from under the Unix limitation of two user-defined
signal types. We already had done something similar for signals directed to
the postmaster process; this adds multiplexing for signals directed to
backends and auxiliary processes (so long as they're connected to shared
memory).
As proof of concept, replace the former usage of SIGUSR1 and SIGUSR2
for backends with use of the multiplexing mechanism. There are still some
hard-wired definitions of SIGUSR1 and SIGUSR2 for other process types,
but getting rid of those doesn't seem interesting at the moment.
Fujii Masao
After a patch originally submitted by Nobuhiro Iwamatsu, but corrected
(I think) to match our guidelines for safe use of asm fragments.
This should be considered untested ...
during it:
When bgwriter is active, the startup process can't perform mdsync() correctly
because it won't see the fsync requests accumulated in bgwriter's private
pendingOpsTable. Therefore make bgwriter responsible for the end-of-recovery
checkpoint as well, when it's active.
When bgwriter is active (= archive recovery), the startup process must not
accumulate fsync requests to its own pendingOpsTable, since bgwriter won't
see them there when it performs restartpoints. Make startup process drop its
pendingOpsTable when bgwriter is launched to avoid that.
Update minimum recovery point one last time when leaving archive recovery.
It won't be updated by the end-of-recovery checkpoint because XLogFlush()
sees us as out of recovery already.
This fixes bug #4879 reported by Fujii Masao.
a backend has done exit(0) or exit(1) without having disengaged itself
from shared memory. We are at risk for this whenever third-party code is
loaded into a backend, since such code might not know it's supposed to go
through proc_exit() instead. Also, it is reported that under Windows
there are ways to externally kill a process that cause the status code
returned to the postmaster to be indistinguishable from a voluntary exit
(thank you, Microsoft). If this does happen then the system is probably
hosed --- for instance, the dead session might still be holding locks.
So the best recovery method is to treat this like a backend crash.
The dead man switch is armed for a particular child process when it
acquires a regular PGPROC, and disarmed when the PGPROC is released;
these should be the first and last touches of shared memory resources
in a backend, or close enough anyway. This choice means there is no
coverage for auxiliary processes, but I doubt we need that, since they
shouldn't be executing any user-provided code anyway.
This patch also improves the management of the EXEC_BACKEND
ShmemBackendArray array a bit, by reducing search costs.
Although this problem is of long standing, the lack of field complaints
seems to mean it's not critical enough to risk back-patching; at least
not till we get some more testing of this mechanism.
when we are waiting for old snapshots to go away during a concurrent index
build. In particular, this rule lets us avoid waiting for
idle-in-transaction sessions.
This logic could be improved further if we had some way to wake up when
the session we are currently waiting for goes idle-in-transaction. However
that would be a significantly more complex/invasive patch, so it'll have to
wait for some other day.
Simon Riggs, with some improvements by Tom.
As pointed out by ITAGAKI Takahiro, we split SInvalLock into two in 8.4,
so to keep the numbers of the rest of the locks unchanged from 8.3, we
don't need a placeholder.
its usual buffer cleaning duties during archive recovery, and it's responsible
for performing restartpoints.
This requires some changes in postmaster. When the startup process has done
all the initialization and is ready to start WAL redo, it signals the
postmaster to launch the background writer. The postmaster is signaled again
when the point in recovery is reached where we know that the database is in
consistent state. Postmaster isn't interested in that at the moment, but
that's the point where we could let other backends in to perform read-only
queries. The postmaster is signaled third time when the recovery has ended,
so that postmaster knows that it's safe to start accepting connections.
The startup process now traps SIGTERM, and performs a "clean" shutdown. If
you do a fast shutdown during recovery, a shutdown restartpoint is performed,
like a shutdown checkpoint, and postmaster kills the processes cleanly. You
still have to continue the recovery at next startup, though.
Currently, the background writer is only launched during archive recovery.
We could launch it during crash recovery as well, but it seems better to keep
that codepath as simple as possible, for the sake of robustness. And it
couldn't do any restartpoints during crash recovery anyway, so it wouldn't be
that useful.
log_restartpoints is gone. Use log_checkpoints instead. This is yet to be
documented.
This whole operation is a pre-requisite for Hot Standby, but has some value of
its own whether the hot standby patch makes 8.4 or not.
Simon Riggs, with lots of modifications by me.
GUC variable effective_io_concurrency controls how many concurrent block
prefetch requests will be issued.
(The best way to handle this for plain index scans is still under debate,
so that part is not applied yet --- tgl)
Greg Stark
initialization, to give loadable modules a reasonable place to perform
creation of any shared memory areas they need. This is the logical conclusion
of our previous creation of RequestAddinShmemSpace() and RequestAddinLWLocks().
We don't need an explicit shmem_shutdown_hook, because the existing
on_shmem_exit and on_proc_exit mechanisms serve that need.
Also, adjust SubPostmasterMain so that libraries that got loaded into the
postmaster will be loaded into all child processes, not only regular backends.
This improves consistency with the non-EXEC_BACKEND behavior, and might be
necessary for functionality for some types of add-ons.
replication patch needs a signal, but we've already used SIGUSR1 and
SIGUSR2 in normal backends. This patch allows reusing SIGUSR1 for that,
and for other purposes too if the need arises.
heap page, where a set bit indicates that all tuples on the page are
visible to all transactions, and the page therefore doesn't need
vacuuming. It is stored in a new relation fork.
Lazy vacuum uses the visibility map to skip pages that don't need
vacuuming. Vacuum is also responsible for setting the bits in the map.
In the future, this can hopefully be used to implement index-only-scans,
but we can't currently guarantee that the visibility map is always 100%
up-to-date.
In addition to the visibility map, there's a new PD_ALL_VISIBLE flag on
each heap page, also indicating that all tuples on the page are visible to
all transactions. It's important that this flag is kept up-to-date. It
is also used to skip visibility tests in sequential scans, which gives a
small performance gain on seqscans.
truncations in FSM code, call FreeSpaceMapTruncateRel from smgr_redo. To
make that cleaner from modularity point of view, move the WAL-logging one
level up to RelationTruncate, and move RelationTruncate and all the
related WAL-logging to new src/backend/catalog/storage.c file. Introduce
new RelationCreateStorage and RelationDropStorage functions that are used
instead of calling smgrcreate/smgrscheduleunlink directly. Move the
pending rel deletion stuff from smgrcreate/smgrscheduleunlink to the new
functions. This leaves smgr.c as a thin wrapper around md.c; all the
transactional stuff is now in storage.c.
This will make it easier to add new forks with similar truncation logic,
like the visibility map.
(but not locked, as that would risk deadlocks). Also, make it work in a small
ring of buffers to avoid having bulk inserts trash the whole buffer arena.
Robert Haas, after an idea of Simon Riggs'.
allowed different processes to have different addresses for the shmem segment
in quite a long time, but there were still a few places left that used the
old coding convention. Clean them up to reduce confusion and improve the
compiler's ability to detect pointer type mismatches.
Kris Jurka
on non-full-page-image WAL records, and quite arbitrarily, only if there's
less than 20% free space on the page after the insert/update (not on HOT
updates, though). The 20% cutoff should avoid most of the overhead, when
replaying a bulk insertion, for example, while ensuring that pages that
are full are marked as full in the FSM.
This is mostly to avoid the nasty worst case scenario, where you replay
from a PITR archive, and the FSM information in the base backup is really
out of date. If there was a lot of pages that the outdated FSM claims to
have free space, but don't actually have any, the first unlucky inserter
after the recovery would traverse through all those pages, just to find
out that they're full. We didn't have this problem with the old FSM
implementation, because we simply threw the FSM information away on a
non-clean shutdown.
functions into one ReadBufferExtended function, that takes the strategy
and mode as argument. There's three modes, RBM_NORMAL which is the default
used by plain ReadBuffer(), RBM_ZERO, which replaces ZeroOrReadBuffer, and
a new mode RBM_ZERO_ON_ERROR, which allows callers to read corrupt pages
without throwing an error. The FSM needs the new mode to recover from
corrupt pages, which could happend if we crash after extending an FSM file,
and the new page is "torn".
Add fork number to some error messages in bufmgr.c, that still lacked it.
This basically takes some build system code that was previously labeled
"Solaris" and ties it to the compiler rather than the operating system.
Author: Julius Stroffek <Julius.Stroffek@Sun.COM>