After further review, we want to move in the direction of always
quoting GUC names in error messages, rather than the previous (PG16)
wildly mixed practice or the intermittent (mid-PG17) idea of doing
this depending on how possibly confusing the GUC name is.
This commit applies appropriate quotes to (almost?) all mentions of
GUC names in error messages. It partially supersedes a243569bf6 and
8d9978a717, which had moved things a bit in the opposite direction
but which then were abandoned in a partial state.
Author: Peter Smith <smithpb2250@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/CAHut%2BPv-kSN8SkxSdoHano_wPubqcg5789ejhCDZAcLFceBR-w%40mail.gmail.com
Run pgindent, pgperltidy, and reformat-dat-files.
The pgindent part of this is pretty small, consisting mainly of
fixing up self-inflicted formatting damage from patches that
hadn't bothered to add their new typedefs to typedefs.list.
In order to keep it from making anything worse, I manually added
a dozen or so typedefs that appeared in the existing typedefs.list
but not in the buildfarm's list. Perhaps we should formalize that,
or better find a way to get those typedefs into the automatic list.
pgperltidy is as opinionated as always, and reformat-dat-files too.
Presently, when this function is called for an unlogged sequence on
a standby server, it will error out with a message like
ERROR: could not open file "base/5/16388": No such file or directory
Since the pg_sequences system view uses pg_sequence_last_value(),
it can error similarly. To fix, modify the function to return NULL
for unlogged sequences on standby servers. Since this bug is
present on all versions since v15, this approach is preferable to
making the ERROR nicer because we need to repair the pg_sequences
view without modifying its definition on released versions. For
consistency, this commit also modifies the function to return NULL
for other sessions' temporary sequences. The pg_sequences view
already appropriately filters out such sequences, so there's no bug
there, but we might as well offer some defense in case someone
invokes this function directly.
Unlogged sequences were first introduced in v15, but temporary
sequences are much older, so while the fix for unlogged sequences
is only back-patched to v15, the temporary sequence portion is
back-patched to all supported versions.
We could also remove the privilege check in the pg_sequences view
definition in v18 if we modify this function to return NULL for
sequences for which the current user lacks privileges, but that is
left as a future exercise for when v18 development begins.
Reviewed-by: Tom Lane, Michael Paquier
Discussion: https://postgr.es/m/20240501005730.GA594666%40nathanxps13
Backpatch-through: 12
Do not allow ALTER SUBSCRIPTION ... SET (failover = on|off) in a
transaction block as the changed failover option of the slot can't be
rolled back. For the same reason, we refrain from altering the replication
slot's failover property if the subscription is created with a valid
slot_name and create_slot=false.
Reprted-by: Kuroda Hayato
Author: Hou Zhijie
Reviewed-by: Shveta Malik, Bertrand Drouvot, Kuroda Hayato
Discussion: https://postgr.es/m/OS0PR01MB57165542B09DFA4943830BF294082@OS0PR01MB5716.jpnprd01.prod.outlook.com
GetPageWithFreeSpace() callers assume the returned block exists in the
main fork, failing with "could not read block" errors if that doesn't
hold. Make that assumption reliable now. It hadn't been guaranteed,
due to the weak WAL and data ordering of participating components. Most
operations on the fsm fork are not WAL-logged. Relation extension is
not WAL-logged. Hence, an fsm-fork block on disk can reference a
main-fork block that no WAL record has initialized. That could happen
after an OS crash, a replica promote, or a PITR restore. wal_log_hints
makes the trouble easier to hit; a replica promote or PITR ending just
after a relevant fsm-fork FPI_FOR_HINT may yield this broken state. The
v16 RelationAddBlocks() mechanism also makes the trouble easier to hit,
since it bulk-extends even without extension lock waiters. Commit
917dc7d239 stopped trouble around
truncation, but vectors involving PageIsNew() pages remained.
This implementation adds a RelationGetNumberOfBlocks() call when the
cached relation size doesn't confirm a block exists. We've been unable
to identify a benchmark that slows materially, but this may show up as
additional time in lseek(). An alternative without that overhead would
be a new ReadBufferMode such that ReadBufferExtended() returns NULL
after a 0-byte read, with all other errors handled normally. However,
each GetFreeIndexPage() caller would then need code for the return-NULL
case. Back-patch to v14, due to earlier versions not caching relation
size and the absence of a pre-v16 problem report.
Ronan Dunklau. Reported by Ronan Dunklau.
Discussion: https://postgr.es/m/1878547.tdWV9SEqCh%40aivenlaptop
Wait for the standby to catch up before syncing the slots with
pg_sync_replication_slots(), otherwise, the logical slot could be ahead
and the sync would fail.
The other way to fix the test is to change it to use slotsync worker and
poll for the sync to get finished but the current approach is better as
this is a predictable way to write the test.
Per buildfarm
Author: Hou Zhijie
Reviewed-by: Amit Kapila
Discussion: https://postgr.es/m/OS0PR01MB571665359F2F5DCD3ADABC9F94002@OS0PR01MB5716.jpnprd01.prod.outlook.com
It is possible that even if the primary waits for the subscriber to catch
up and then disables the subscription, the XLOG_RUNNING_XACTS record gets
inserted between the two steps by bgwriter and walsender processes it.
This can move the restart_lsn of the corresponding slot in an
unpredictable way which further leads to slot sync failure.
To ensure predictable behaviour, we drop the subscription and manually
create the slot before the test. The other idea we discussed to write a
predictable test is to use injection points to control the bgwriter
logging XLOG_RUNNING_XACTS but that needs more analysis. We can add a
separate test using injection points.
Per buildfarm
Author: Hou Zhijie
Reviewed-by: Amit Kapila, Shveta Malik
Discussion: https://postgr.es/m/CAA4eK1JD8h_XLRsK_o_Xh=5MhTzm+6d4Cb4_uPgFJ2wSQDah=g@mail.gmail.com
This commit does two things:
1) Maintains inactive_since for sync slots whenever the slot is released
just like any other regular slot.
2) Ensures the value is set to the current timestamp during the promotion
of standby to help correctly interpret the time after promotion. We don't
want the slots to appear inactive for a long time after promotion if they
haven't been synchronized recently. This would also avoid the invalidation
of such slots immediately after promotion if tomorrow we have a feature
that invalidates slots based on their inactivity time. Whoever acquires
the slot i.e. makes the slot active will reset it to NULL.
Author: Bharath Rupireddy
Reviewed-by: Bertrand Drouvot, Amit Kapila, Shveta Malik, Masahiko Sawada
Discussion: https://postgr.es/m/CAA4eK1KrPGwfZV9LYGidjxHeW+rxJ=E2ThjXvwRGLO=iLNuo=Q@mail.gmail.com
Discussion: https://postgr.es/m/CALj2ACW4aUe-_uFQOjdWCEN-xXoLGhmvRFnL8SNw_TZ5nJe+aw@mail.gmail.com
Discussion: https://postgr.es/m/CA+Tgmob_Ta-t2ty8QrKHBGnNLrf4ZYcwhGHGFsuUoFrAEDw4sA@mail.gmail.com
We were directly copying the LSN locations while syncing the slots on the
standby. Now, it is possible that at some particular restart_lsn there are
some running xacts, which means if we start reading the WAL from that
location after promotion, we won't reach a consistent snapshot state at
that point. However, on the primary, we would have already been in a
consistent snapshot state at that restart_lsn so we would have just
serialized the existing snapshot.
To avoid this problem we will use the advance_slot functionality unless
the snapshot already exists at the synced restart_lsn location. This will
help us to ensure that snapbuilder/slot statuses are updated properly
without generating any changes. Note that the synced slot will remain as
RS_TEMPORARY till the decoding from corresponding restart_lsn can reach a
consistent snapshot state after which they will be marked as
RS_PERSISTENT.
Per buildfarm
Author: Hou Zhijie
Reviewed-by: Bertrand Drouvot, Shveta Malik, Bharath Rupireddy, Amit Kapila
Discussion: https://postgr.es/m/OS0PR01MB5716B3942AE49F3F725ACA92943B2@OS0PR01MB5716.jpnprd01.prod.outlook.com
pg_wal_replay_wait() is to be used on standby and specifies waiting for
the specific WAL location to be replayed before starting the transaction.
This option is useful when the user makes some data changes on primary and
needs a guarantee to see these changes on standby.
The queue of waiters is stored in the shared memory array sorted by LSN.
During replay of WAL waiters whose LSNs are already replayed are deleted from
the shared memory array and woken up by setting of their latches.
pg_wal_replay_wait() needs to wait without any snapshot held. Otherwise,
the snapshot could prevent the replay of WAL records implying a kind of
self-deadlock. This is why it is only possible to implement
pg_wal_replay_wait() as a procedure working in a non-atomic context,
not a function.
Catversion is bumped.
Discussion: https://postgr.es/m/eb12f9b03851bb2583adab5df9579b4b%40postgrespro.ru
Author: Kartyshov Ivan, Alexander Korotkov
Reviewed-by: Michael Paquier, Peter Eisentraut, Dilip Kumar, Amit Kapila
Reviewed-by: Alexander Lakhin, Bharath Rupireddy, Euler Taveira
This commit adds a new property called last_inactive_time for slots. It is
set to 0 whenever a slot is made active/acquired and set to the current
timestamp whenever the slot is inactive/released or restored from the disk.
Note that we don't set the last_inactive_time for the slots currently being
synced from the primary to the standby because such slots are typically
inactive as decoding is not allowed on those.
The 'last_inactive_time' will be useful on production servers to debug and
analyze inactive replication slots. It will also help to know the lifetime
of a replication slot - one can know how long a streaming standby, logical
subscriber, or replication slot consumer is down.
The 'last_inactive_time' will also be useful to implement inactive
timeout-based replication slot invalidation in a future commit.
Author: Bharath Rupireddy
Reviewed-by: Bertrand Drouvot, Amit Kapila, Shveta Malik
Discussion: https://www.postgresql.org/message-id/CALj2ACW4aUe-_uFQOjdWCEN-xXoLGhmvRFnL8SNw_TZ5nJe+aw@mail.gmail.com
Till now, the reason for replication slot invalidation is not tracked
directly in pg_replication_slots. A recent commit 007693f2a3 added
'conflict_reason' to show the reasons for slot conflict/invalidation, but
only for logical slots.
This commit adds a new column 'invalidation_reason' to show invalidation
reasons for both physical and logical slots. And, this commit also turns
'conflict_reason' text column to 'conflicting' boolean column (effectively
reverting commit 007693f2a3). The 'conflicting' column is true for
invalidation reasons 'rows_removed' and 'wal_level_insufficient' because
those make the slot conflict with recovery. When 'conflicting' is true,
one can now look at the new 'invalidation_reason' column for the reason
for the logical slot's conflict with recovery.
The new 'invalidation_reason' column will also be useful to track other
invalidation reasons in the future commit.
Author: Bharath Rupireddy
Reviewed-by: Bertrand Drouvot, Amit Kapila, Shveta Malik
Discussion: https://www.postgresql.org/message-id/ZfR7HuzFEswakt/a%40ip-10-97-1-34.eu-west-3.compute.internal
Discussion: https://www.postgresql.org/message-id/CALj2ACW4aUe-_uFQOjdWCEN-xXoLGhmvRFnL8SNw_TZ5nJe+aw@mail.gmail.com
There are currently no tests for the low-level backup method where
pg_backup_start() and pg_backup_stop() are involved while taking a
file-system backup. The tests introduced in this commit rely on a
background psql process to make sure that the backup is taken while the
session doing the SQL start and stop calls remains alive.
Two cases are checked here with the backup taken:
- Recovery without a backup_label, leading to a corrupted state.
- Recovery with a backup_label, with a consistent state reached.
Both cases cross-check some patterns in the logs generated when running
recovery.
Compared to the first attempt in 99b4a63bef, this includes a couple of
fixes making the CI stable (5 runs succeeded here):
- Add the file to the list of tests in meson.build.
- Race condition with the first WAL segment that we expect in the
primary's archives, by adding a poll on pg_stat_archiver. The second
segment with the checkpoint record is archived thanks to pg_backup_stop
waiting for it.
- Fix failure of test where the backup_label does not exist. The
cluster inherits the configuration of the first node; it was attempting
to store segments in the first node's archives, triggering failures with
copy on Windows.
- Fix failure of test on Windows because of incorrect parsing of the
backup_file in the success case. The data of the backup_label file is
retrieved from the output pg_backup_stop() from a BackgroundPsql written
directly to the backup's data folder. This would include CRLFs (\r\n),
causing the startup process to fail at the beginning of recovery when
parsing the backup_label because only LFs (\n) are allowed.
Author: David Steele
Discussion: https://postgr.es/m/f20fcc82-dadb-478d-beb4-1e2ffb0ace76@pgmasters.net
This reverts commit 99b4a63bef. The test is proving to be unstable,
so revert it for now.
One of the failures seen involves the cluster started without the
backup_label, where the archives of the primary are overwritten, causing
recovery failures on Windows. This is simple to fix, but there is
another issue that's creeping behind in the form of an "invalid data in
file" ERROR when parsing the backup_label for the second recovery case,
as an effect of the backup_label data written after retrieving its
contents from pg_backup_stop().
_
Per buildfarm member sidewinder.
There are currently no tests for the low-level backup method where
pg_backup_start() and pg_backup_stop() are involved while taking a
file-system backup. The tests introduced in this commit rely on a
background psql process to make sure that the backup is taken while the
session doing the SQL start and stop calls remains alive.
Two cases are checked here with the backup taken:
- Recovery without a backup_label, leading to a corrupted state.
- Recovery with a backup_label, with a consistent state reached.
Both cases cross-check some patterns in the logs generated when running
recovery.
Author: David Steele
Discussion: https://postgr.es/m/f20fcc82-dadb-478d-beb4-1e2ffb0ace76@pgmasters.net
The test ensures that all the WAL on the publisher is sent to the
subscriber before shutdown by comparing the confirmed_flush_lsn of the
associated slot with the shutdown_checkpoint WAL location. But if the
shutdown_checkpoint location falls into a new page in the WAL then the
check won't work. So, ensure that the shutdown_checkpoint WAL record
doesn't fall into a new page.
Reported-by: Bharath Rupireddy
Author: Bharath Rupireddy
Reviewed-by: Vignesh C, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/CALj2ACVLzH5CN-h9=S26mdRHPuJ9yDLUw70yh4JOiPw03WL0CQ@mail.gmail.com
The code path in charge of invalidating a replication slot includes a
call to pgstat_drop_replslot(), which would result in removing the
statistics of the slot once invalidated. However, there is no need to
remove the statistics of an invalidated slot as one could still be
interested in looking at them to understand the activity of the slot
until its actual removal.
The initial design of the feature committed in be87200efd used the
approach to drop the slots, which is likely why the statistics were
still removed during the invalidation.
Another problem with this operation is that it was done without holding
ReplicationSlotAllocationLock, leaving it unprotected on concurrent
activity. This part is arguably a bug, but that's a limited problem in
practice so no backpatch is done.
In passing, this commit adds a test to check this behavior. The only
remaining code path where slot statistics are dropped now related to the
slot getting dropped.
Author: Bertrand Drouvot
Discussion: https://postgr.es/m/ZermH08Eq6YydHpO@ip-10-97-1-34.eu-west-3.compute.internal
This patch provides a way to ensure that physical standbys that are
potential failover candidates have received and flushed changes before
the primary server making them visible to subscribers. Doing so guarantees
that the promoted standby server is not lagging behind the subscribers
when a failover is necessary.
The logical walsender now guarantees that all local changes are sent and
flushed to the standby servers corresponding to the replication slots
specified in 'standby_slot_names' before sending those changes to the
subscriber.
Additionally, the SQL functions pg_logical_slot_get_changes,
pg_logical_slot_peek_changes and pg_replication_slot_advance are modified
to ensure that they process changes for failover slots only after physical
slots specified in 'standby_slot_names' have confirmed WAL receipt for those.
Author: Hou Zhijie and Shveta Malik
Reviewed-by: Masahiko Sawada, Peter Smith, Bertrand Drouvot, Ajin Cherian, Nisha Moond, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
This reverts commit 08a52ab151, due to some sporadic instability in
the test. Getting the test right should require some redesign with a
second injection point, but let's revert it for now to avoid these
issues in the CI as a lot of patches are under discussion in this last
commit fest.
Per buildfarm members hachi and gokiburi.
Discussion: https://postgr.es/m/ZekQQHCrIqLVpGz5@paquier.xyz
This commit adds a recovery test to provide coverage for the bug fixed
in 818fefd8fd, using an injection point to wait just after the process
of an active slot is killed. The trick is to give enough time for
effective_xmin and effective_catalog_xmin to advance so as the slot
invalidation robustness can be checked since the active process is
killed without holding its slot's mutex for a short time.
Author: Bertrand Drouvot
Discussion: https://postgr.es/m/ZdyZya4YrNapWKqz@ip-10-97-1-34.eu-west-3.compute.internal
When this code was written the duplicity didn't matter, but with all the
SLRU-bank stuff we just added, it has become excessive. Turn it into a
simpler loop with no code duplication. Also add a test so that this
code becomes covered.
Discussion: https://postgr.es/m/202403041517.3a35jw53os65@alvherre.pgsql
Per a demand from the author and the reviewer of this commit, this adds
to Cluster.pm a helper routine that can be used to monitor when a
process reaches a wanted wait event. This can be used in combination
with the module injection_points for the "wait" callback, though it is
not limited to it as this monitors pg_stat_activity for a wait_event and
a backend_type.
Author: Bertrand Drouvot
Reviewed-by: Andrey Borodin
Discussion: https://postgr.es/m/ZeBB4RMPEZ06TcdY@ip-10-97-1-34.eu-west-3.compute.internal
This test serves as a way to demonstrate how to use the features
introduced in 37b369dc67, while providing coverage for 7863ee4def
that caused the startup process to throw "PANIC: could not locate a
valid checkpoint record" when starting recovery. The test checks that a
node is able to properly restart following a crash when a restart point
was finishing across a promotion, with an injection point added in the
middle of CreateRestartPoint() to stop the restartpoint in flight. Note
that this test fails when 7863ee4def is reverted.
Kyotaro Horiguchi is the original author of this test, that has been
originally posted on the thread where 7863ee4def was discussed. I
have just upgraded and polished it to rely on injection points, making
it much cheaper to reproduce the failure.
This test requires injection points to be enabled in the builds, hence
meson and ./configure need an update to pass this knowledge down to the
test. The name of the new injection point follows the same naming
convention as 6a1ea02c49. The Makefile's EXTRA_INSTALL of recovery
TAP tests is updated to include modules/injection_points.
Author: Kyotaro Horiguchi, Michael Paquier
Reviewed-by: Andrey Borodin, Bertrand Drouvot
Discussion: https://postgr.es/m/ZdLuxBk5hGpol91B@paquier.xyz
By enabling slot synchronization, all the failover logical replication
slots on the primary (assuming configurations are appropriate) are
automatically created on the physical standbys and are synced
periodically. The slot sync worker on the standby server pings the primary
server at regular intervals to get the necessary failover logical slots
information and create/update the slots locally. The slots that no longer
require synchronization are automatically dropped by the worker.
The nap time of the worker is tuned according to the activity on the
primary. The slot sync worker waits for some time before the next
synchronization, with the duration varying based on whether any slots were
updated during the last cycle.
A new parameter sync_replication_slots enables or disables this new
process.
On promotion, the slot sync worker is shut down by the startup process to
drop any temporary slots acquired by the slot sync worker and to prevent
the worker from trying to fetch the failover slots.
A functionality to allow logical walsenders to wait for the physical will
be done in a subsequent commit.
Author: Shveta Malik, Hou Zhijie based on design inputs by Masahiko Sawada and Amit Kapila
Reviewed-by: Masahiko Sawada, Bertrand Drouvot, Peter Smith, Dilip Kumar, Ajin Cherian, Nisha Moond, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
Before the previous commit, the test could hang until
LOG_SNAPSHOT_INTERVAL_MS (15s), until checkpoint_timeout (300s), or
indefinitely. An indefinite hang was awfully improbable. It entailed
the test reaching checkpoint_timeout before the
DecodingContextFindStartpoint() of a CREATE SUBSCRIPTION, yet after the
preceding WAL record. Back-patch to v16, which introduced the test.
Bertrand Drouvot, reported by Noah Misch.
Discussion: https://postgr.es/m/20240211010227.a2.nmisch@google.com
One IPC::Run::start() used an IPC::Run::timer() without checking for
expiration. The other used no timeout or timer. Back-patch to v16,
which introduced the test.
Reviewed by Bertrand Drouvot.
Discussion: https://postgr.es/m/20240211010227.a2.nmisch@google.com
The failure is that the remote slot is not synchronized after the same
slot on standby gets invalidated. The reason was that remote_slot's
restart_lsn was lagged behind the standby's oldest WAL segment. The test
didn't ensure that remote_slot's LSN was advanced to the latest position
before we tried to sync the slots via new the function
pg_sync_replication_slots().
Discussion: https://postgr.es/m/CAA4eK1JLBi3HzenB6do3_hd78kN0UDD1mz-vumWE52XHHEq5Bw@mail.gmail.com
This commit introduces a new SQL function pg_sync_replication_slots()
which is used to synchronize the logical replication slots from the
primary server to the physical standby so that logical replication can be
resumed after a failover or planned switchover.
A new 'synced' flag is introduced in pg_replication_slots view, indicating
whether the slot has been synchronized from the primary server. On a
standby, synced slots cannot be dropped or consumed, and any attempt to
perform logical decoding on them will result in an error.
The logical replication slots on the primary can be synchronized to the
hot standby by using the 'failover' parameter of
pg-create-logical-replication-slot(), or by using the 'failover' option of
CREATE SUBSCRIPTION during slot creation, and then calling
pg_sync_replication_slots() on standby. For the synchronization to work,
it is mandatory to have a physical replication slot between the primary
and the standby aka 'primary_slot_name' should be configured on the
standby, and 'hot_standby_feedback' must be enabled on the standby. It is
also necessary to specify a valid 'dbname' in the 'primary_conninfo'.
If a logical slot is invalidated on the primary, then that slot on the
standby is also invalidated.
If a logical slot on the primary is valid but is invalidated on the
standby, then that slot is dropped but will be recreated on the standby in
the next pg_sync_replication_slots() call provided the slot still exists
on the primary server. It is okay to recreate such slots as long as these
are not consumable on standby (which is the case currently). This
situation may occur due to the following reasons:
- The 'max_slot_wal_keep_size' on the standby is insufficient to retain
WAL records from the restart_lsn of the slot.
- 'primary_slot_name' is temporarily reset to null and the physical slot
is removed.
The slot synchronization status on the standby can be monitored using the
'synced' column of pg_replication_slots view.
A functionality to automatically synchronize slots by a background worker
and allow logical walsenders to wait for the physical will be done in
subsequent commits.
Author: Hou Zhijie, Shveta Malik, Ajin Cherian based on an earlier version by Peter Eisentraut
Reviewed-by: Masahiko Sawada, Bertrand Drouvot, Peter Smith, Dilip Kumar, Nisha Moond, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
This commit introduces a new subscription option named 'failover', which
provides users with the ability to set the failover property of the
replication slot on the publisher when creating or altering a
subscription.
This uses the replication commands introduced by commit 7329240437 to
enable the failover option for a logical replication slot.
If the failover option is set to true, the associated replication slots
(i.e. the main slot and the table sync slots) in the upstream database are
enabled to be synchronized to the standbys. Note that the capability to
sync the replication slots will be added in subsequent commits.
Thanks to Masahiko Sawada for the design inputs.
Author: Shveta Malik, Hou Zhijie, Ajin Cherian
Reviewed-by: Peter Smith, Bertrand Drouvot, Dilip Kumar, Masahiko Sawada, Nisha Moond, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
This commit tweaks a couple of things in 035_standby_logical_decoding to
hopefully stabilize it:
- Autovacuum is now disabled, as it could hold a global xmin with a
transaction.
- Conflicts are generated with command sequences that removed rows (on
catalogs, shared or non-shared, or just plain relations) followed by a
VACUUM. This was unstable because these did not check that the horizon
moved between the SQL commands and the VACUUM. The logic is refactored
as follows, to ensure that VACUUM removes dead rows before testing for
slot invalidation on a standby (idea suggested by Andres Freund):
-- Grab the current horizon.
-- Launch SQL commands removing rows.
-- Check that the snapshot horizon has been updated.
-- Launch VACUUM on the relation whose rows have been removed by the
first step.
Note that there are still some issues because of standby snapshot WAL
records generated by the bgwriter, but this makes the test much more
stable.
Per reports from buildfarm members dikkop and skink, with analysis and
tests from Alexander Lakhin.
While on it, fix a couple of incorrect comments.
Author: Bertrand Drouvot
Reviewed-by: Alexander Lakhin, Michael Paquier
Discussion: https://postgr.es/m/OSZPR01MB6310ED3CEDB531BCEDBC6AF2FD479@OSZPR01MB6310.jpnprd01.prod.outlook.com
Discussion: https://postgr.es/m/bf67e076-b163-9ba3-4ade-b9fc51a3a8f6@gmail.com
Backpatch-through: 16
This patch changes the existing 'conflicting' field to 'conflict_reason'
in pg_replication_slots. This new field indicates the reason for the
logical slot's conflict with recovery. It is always NULL for physical
slots, as well as for logical slots which are not invalidated. The
non-NULL values indicate that the slot is marked as invalidated. Possible
values are:
wal_removed = required WAL has been removed.
rows_removed = required rows have been removed.
wal_level_insufficient = the primary doesn't have a wal_level sufficient
to perform logical decoding.
The existing users of 'conflicting' column can get the same answer by
using 'conflict_reason' IS NOT NULL.
Author: Shveta Malik
Reviewed-by: Amit Kapila, Bertrand Drouvot, Michael Paquier
Discussion: https://postgr.es/m/ZYOE8IguqTbp-seF@paquier.xyz
Some of the TAP tests have been historically setting the environment
variable PGDATABASE to 'postgres', which is not needed because
PostgreSQL::Test::Cluster already sets it when initialized. This commit
removes these explicit setups.
Note that the dependency of cluster -a with PGDATABASE (from
1caef31d9e) is still documented.
Author: Bharath Rupireddy
Discussion: https://postgr.es/m/CALj2ACXLAz5dW3ZP+Fec8g6jQMMmDyCVT+qdbye2h7QJJmhsdw@mail.gmail.com
There are a lot of Perl scripts in the tree, mostly code generation
and TAP tests. Occasionally, these scripts produce warnings. These
are probably always mistakes on the developer side (true positives).
Typical examples are warnings from genbki.pl or related when you make
a mess in the catalog files during development, or warnings from tests
when they massage a config file that looks different on different
hosts, or mistakes during merges (e.g., duplicate subroutine
definitions), or just mistakes that weren't noticed because there is a
lot of output in a verbose build.
This changes all warnings into fatal errors, by replacing
use warnings;
by
use warnings FATAL => 'all';
in all Perl files.
Discussion: https://www.postgresql.org/message-id/flat/06f899fd-1826-05ab-42d6-adeb1fd5e200%40eisentraut.org
The code was passing a scalar argument to node->restart(), but it was
expecting a hash, which causes a warning from Perl ("Odd number of
elements in hash assignment").
But the node->restart() function doesn't take a mode argument anyway.
This was probably copied from an incorrect comment (see commit
750c59d7ec). The default restart mode is already "fast", so the test
should still be semantically correct without explicitly specifying the
mode.
Discussion: https://www.postgresql.org/message-id/e3f4bf1b-63d3-408a-b07e-d35a0fdf1b98@eisentraut.org
This is a function that makes a node jump by N WAL segments, which is
something a couple of tests have been relying on for some cases related
to streaming, replication slot limits and logical decoding on standbys.
Hence, this centralizes the logic, while making it cheaper by relying on
pg_logical_emit_message() to emit WAL records before switching to a new
segment.
Author: Bharath Rupireddy
Reviewed-by: Kyotaro Horiguchi, Euler Taveira
Discussion: https://postgr.es/m/CALj2ACU3R8QFCvDewHCMKjgb2w_-CMCyd6DAK=Jb-af14da5eg@mail.gmail.com
A PostgreSQL release tarball contains a number of prebuilt files, in
particular files produced by bison, flex, perl, and well as html and
man documentation. We have done this consistent with established
practice at the time to not require these tools for building from a
tarball. Some of these tools were hard to get, or get the right
version of, from time to time, and shipping the prebuilt output was a
convenience to users.
Now this has at least two problems:
One, we have to make the build system(s) work in two modes: Building
from a git checkout and building from a tarball. This is pretty
complicated, but it works so far for autoconf/make. It does not
currently work for meson; you can currently only build with meson from
a git checkout. Making meson builds work from a tarball seems very
difficult or impossible. One particular problem is that since meson
requires a separate build directory, we cannot make the build update
files like gram.h in the source tree. So if you were to build from a
tarball and update gram.y, you will have a gram.h in the source tree
and one in the build tree, but the way things work is that the
compiler will always use the one in the source tree. So you cannot,
for example, make any gram.y changes when building from a tarball.
This seems impossible to fix in a non-horrible way.
Second, there is increased interest nowadays in precisely tracking the
origin of software. We can reasonably track contributions into the
git tree, and users can reasonably track the path from a tarball to
packages and downloads and installs. But what happens between the git
tree and the tarball is obscure and in some cases non-reproducible.
The solution for both of these issues is to get rid of the step that
adds prebuilt files to the tarball. The tarball now only contains
what is in the git tree (*). Getting the additional build
dependencies is no longer a problem nowadays, and the complications to
keep these dual build modes working are significant. And of course we
want to get the meson build system working universally.
This commit removes the make distprep target altogether. The make
dist target continues to do its job, it just doesn't call distprep
anymore.
(*) - The tarball also contains the INSTALL file that is built at make
dist time, but not by distprep. This is unchanged for now.
The make maintainer-clean target, whose job it is to remove the
prebuilt files in addition to what make distclean does, is now just an
alias to make distprep. (In practice, it is probably obsolete given
that git clean is available.)
The following programs are now hard build requirements in configure
(they were already required by meson.build):
- bison
- flex
- perl
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://www.postgresql.org/message-id/flat/e07408d9-e5f2-d9fd-5672-f53354e9305e@eisentraut.org
Historically, the statistics of the checkpointer have been always part
of pg_stat_bgwriter. This commit removes a few columns from
pg_stat_bgwriter, and introduces pg_stat_checkpointer with equivalent,
renamed columns (plus a new one for the reset timestamp):
- checkpoints_timed -> num_timed
- checkpoints_req -> num_requested
- checkpoint_write_time -> write_time
- checkpoint_sync_time -> sync_time
- buffers_checkpoint -> buffers_written
The fields of PgStat_CheckpointerStats and its SQL functions are renamed
to match with the new field names, for consistency. Note that
background writer and checkpointer have been split into two different
processes in commits 806a2aee37 and bf405ba8e4. The pgstat
structures were already split, making this change straight-forward.
Bump catalog version.
Author: Bharath Rupireddy
Reviewed-by: Bertrand Drouvot, Andres Freund, Michael Paquier
Discussion: https://postgr.es/m/CALj2ACVxX2ii=66RypXRweZe2EsBRiPMj0aHfRfHUeXJcC7kHg@mail.gmail.com
Peter Eisentraut
Tom Lane
Nathan Bossart
Amit Kapila
Noah Misch
Alexander Korotkov
Michael Paquier
Alvaro Herrera
Andrew Dunstan
Bruce Momjian