Provide a way to run the parallel regression tests with a user-specified

limit on the number of simultaneous connections.  Andrew Dunstan, with
review by Tom Lane.
This commit is contained in:
Tom Lane 2003-11-02 21:56:15 +00:00
parent 4593531205
commit eeaf5f04ae
3 changed files with 80 additions and 35 deletions

View File

@ -1,4 +1,4 @@
<!-- $Header: /cvsroot/pgsql/doc/src/sgml/regress.sgml,v 1.34 2003/09/12 22:17:23 tgl Exp $ -->
<!-- $Header: /cvsroot/pgsql/doc/src/sgml/regress.sgml,v 1.35 2003/11/02 21:56:14 tgl Exp $ -->
<chapter id="regress">
<title id="regress-title">Regression Tests</title>
@ -52,18 +52,18 @@ gmake check
<screen>
<computeroutput>
======================
All 77 tests passed.
All 93 tests passed.
======================
</computeroutput>
</screen>
or otherwise a note about what tests failed. See <xref
or otherwise a note about which tests failed. See <xref
linkend="regress-evaluation"> below for more.
</para>
<note>
<para>
Because this test method runs a temporary server, it will not work
when you are the root user (the server will not start as root).
when you are the root user (since the server will not start as root).
If you already did the build as root, you do not have to start all
over. Instead, make the regression test directory writable by
some other user, log in as that user, and restart the tests.
@ -88,14 +88,18 @@ gmake check
<para>
The parallel regression test starts quite a few processes under your
user ID. Presently, the maximum concurrency is twenty parallel test
scripts, which means sixty processes: there's a server process, a <application>psql</>,
and usually a shell parent process for the <application>psql</> for each test script.
scripts, which means sixty processes: there's a server process, a
<application>psql</>, and usually a shell parent process for the
<application>psql</> for each test script.
So if your system enforces a per-user limit on the number of processes,
make sure this limit is at least seventy-five or so, else you may get
random-seeming failures in the parallel test. If you are not in
a position to raise the limit, you can edit the file
<filename>src/test/regress/parallel_schedule</> to split the
larger concurrent test sets into more manageable groups.
a position to raise the limit, you can cut down the degree of parallelism
by setting the <literal>MAX_CONNECTIONS</> parameter. For example,
<screen>
gmake MAX_CONNECTIONS=10 check
</screen>
runs no more than ten tests concurrently.
</para>
</tip>
@ -109,8 +113,8 @@ gmake check
<screen>
gmake SHELL=/bin/ksh check
</screen>
If no non-broken shell is available, you can alter the parallel test
schedule as suggested above.
If no non-broken shell is available, you may be able to work around the
problem by limiting the number of connections, as shown above.
</para>
</tip>
@ -177,7 +181,7 @@ gmake installcheck
<para>
If you run the tests against an already-installed server that was
initialized with a collation order locale different than C then
initialized with a collation-order locale other than C, then
there may be differences due to sort order and follow-up
failures. The regression test suite is set up to handle this
problem by providing alternative result files that together are
@ -199,7 +203,7 @@ gmake installcheck
<literal><replaceable>testname</>_<replaceable>digit</>.out</>.
The actual digit is not significant. Remember that the
regression test driver will consider all such files to be equally
valid test results. If the test results are platform-dependent,
valid test results. If the test results are platform-specific,
the technique described in <xref linkend="regress-platform">
should be used instead.
</para>
@ -245,7 +249,7 @@ env TZ=PST8PDT date
</screen>
The command above should have returned the current system time in
the <literal>PST8PDT</literal> time zone. If the <literal>PST8PDT</literal> time zone is not available,
then your system may have returned the time in GMT. If the
then your system may have returned the time in UTC. If the
<literal>PST8PDT</literal> time zone is missing, you can set the time zone
rules explicitly:
<programlisting>
@ -364,7 +368,7 @@ diff results/random.out expected/random.out
which comparison file to use. So, to eliminate bogus test
<quote>failures</quote> for a particular platform, you must choose
or make a variant result file, and then add a line to the mapping
file, which is <filename>resultmap</filename>.
file, which is <filename>src/test/regress/resultmap</filename>.
</para>
<para>

View File

@ -7,7 +7,7 @@
#
#
# IDENTIFICATION
# $Header: /cvsroot/pgsql/src/test/regress/GNUmakefile,v 1.42 2003/03/21 17:18:34 petere Exp $
# $Header: /cvsroot/pgsql/src/test/regress/GNUmakefile,v 1.43 2003/11/02 21:56:15 tgl Exp $
#
#-------------------------------------------------------------------------
@ -25,6 +25,12 @@ SHLIB_LINK = $(BE_DLLLIBS)
# default encoding
MULTIBYTE = SQL_ASCII
# maximum simultaneous connections for parallel tests
MAXCONNOPT :=
ifdef MAX_CONNECTIONS
MAXCONNOPT += --max-connections=$(MAX_CONNECTIONS)
endif
##
## Prepare for tests
@ -113,7 +119,7 @@ all-spi:
##
check: all
$(SHELL) ./pg_regress --temp-install --top-builddir=$(top_builddir) --schedule=$(srcdir)/parallel_schedule --multibyte=$(MULTIBYTE)
$(SHELL) ./pg_regress --temp-install --top-builddir=$(top_builddir) --schedule=$(srcdir)/parallel_schedule --multibyte=$(MULTIBYTE) $(MAXCONNOPT)
installcheck: all
$(SHELL) ./pg_regress --schedule=$(srcdir)/serial_schedule --multibyte=$(MULTIBYTE)
@ -128,7 +134,7 @@ bigtest:
$(SHELL) ./pg_regress --schedule=$(srcdir)/serial_schedule --multibyte=$(MULTIBYTE) numeric_big
bigcheck:
$(SHELL) ./pg_regress --temp-install --top-builddir=$(top_builddir) --schedule=$(srcdir)/parallel_schedule --multibyte=$(MULTIBYTE) numeric_big
$(SHELL) ./pg_regress --temp-install --top-builddir=$(top_builddir) --schedule=$(srcdir)/parallel_schedule --multibyte=$(MULTIBYTE) $(MAXCONNOPT) numeric_big
##

View File

@ -1,5 +1,5 @@
#! /bin/sh
# $Header: /cvsroot/pgsql/src/test/regress/Attic/pg_regress.sh,v 1.35 2003/08/07 14:36:31 tgl Exp $
# $Header: /cvsroot/pgsql/src/test/regress/Attic/pg_regress.sh,v 1.36 2003/11/02 21:56:15 tgl Exp $
me=`basename $0`
: ${TMPDIR=/tmp}
@ -13,6 +13,8 @@ Usage: $me [options...] [extra tests...]
Options:
--debug turn on debug mode in programs that are run
--inputdir=DIR take input files from DIR (default \`.')
--max-connections=N maximum number of concurrent connections
(default is 0 meaning unlimited)
--multibyte=ENCODING use ENCODING as the multibyte encoding, and
also run a test by the same name
--outputdir=DIR place output files in DIR (default \`.')
@ -92,6 +94,7 @@ unset multibyte
dbname=regression
hostname=localhost
maxconnections=0
: ${GMAKE='@GMAKE@'}
@ -124,6 +127,9 @@ do
--temp-install=*)
temp_install=`expr "x$1" : "x--temp-install=\(.*\)"`
shift;;
--max-connections=*)
maxconnections=`expr "x$1" : "x--max-connections=\(.*\)"`
shift;;
--outputdir=*)
outputdir=`expr "x$1" : "x--outputdir=\(.*\)"`
shift;;
@ -156,6 +162,26 @@ do
esac
done
# ----------
# warn of Cygwin likely failure if maxconnections = 0
# and we are running parallel tests
# ----------
case $host_platform in
*-*-cygwin*)
case "$schedule" in
*parallel_schedule*)
if [ $maxconnections -eq 0 ] ; then
echo Using unlimited parallel connections is likely to fail or hang on Cygwin.
echo Try \"$me --max-connections=n\" or \"gmake MAX_CONNECTIONS=n check\"
echo with n = 5 or 10 if this happens.
echo
fi
;;
esac
;;
esac
# ----------
# When on QNX or BeOS, don't use Unix sockets.
@ -248,15 +274,15 @@ if [ -f "$inputdir/resultmap" ]
then
while read LINE
do
HOSTPAT=`expr "$LINE" : '.*/\(.*\)='`
if [ `expr "$host_platform:$compiler" : "$HOSTPAT"` -ne 0 ]
then
# remove hostnamepattern from line so that there are no shell
# wildcards in SUBSTLIST; else later 'for' could expand them!
TESTNAME=`expr "$LINE" : '\(.*\)/'`
SUBST=`echo "$LINE" | sed 's/^.*=//'`
echo "$TESTNAME=$SUBST" >> $TMPFILE
fi
HOSTPAT=`expr "$LINE" : '.*/\(.*\)='`
if [ `expr "$host_platform:$compiler" : "$HOSTPAT"` -ne 0 ]
then
# remove hostnamepattern from line so that there are no shell
# wildcards in SUBSTLIST; else later 'for' could expand them!
TESTNAME=`expr "$LINE" : '\(.*\)/'`
SUBST=`echo "$LINE" | sed 's/^.*=//'`
echo "$TESTNAME=$SUBST" >> $TMPFILE
fi
done <"$inputdir/resultmap"
fi
SUBSTLIST=`cat $TMPFILE`
@ -309,10 +335,11 @@ then
# Windows needs shared libraries in PATH. (Only those linked into
# executables, not dlopen'ed ones)
# ----------
case $host_platform in *-*-cygwin*)
PATH=$libdir:$PATH
export PATH
;;
case $host_platform in
*-*-cygwin*)
PATH=$libdir:$PATH
export PATH
;;
esac
if [ -d "$temp_install" ]; then
@ -556,11 +583,19 @@ do
else
# Start a parallel group
$ECHO_N "parallel group ($# tests): $ECHO_C"
if [ $maxconnections -gt 0 ] ; then
connnum=0
test $# -gt $maxconnections && $ECHO_N "(in groups of $maxconnections) $ECHO_C"
fi
for name do
(
$PSQL -d "$dbname" <"$inputdir/sql/$name.sql" >"$outputdir/results/$name.out" 2>&1
$PSQL -d "$dbname" <"$inputdir/sql/$name.sql" >"$outputdir/results/$name.out" 2>&1
$ECHO_N " $name$ECHO_C"
) &
if [ $maxconnections -gt 0 ] ; then
connnum=`expr \( $connnum + 1 \) % $maxconnections`
test $connnum -eq 0 && wait
fi
done
wait
echo
@ -593,8 +628,8 @@ do
fi
done
# If there are multiple equally valid result file, loop to get the right one.
# If none match, diff against the closet one.
# If there are multiple equally valid result files, loop to get the right one.
# If none match, diff against the closest one.
bestfile=
bestdiff=