Hi,
It seems that 026_overwrite_contrecord.pl test often fails under valgrind. I
first thought that related failures on skink were due to me migrating the
animal to a new host (and then running out of space due to a mistake in ccache
config). But it happened again after I fixed those, and I just reproduced the
issue locally.
It's a bit odd that we didn't start to see these failures immediately, but
only in the last few days. I'd blame skink being migrated to a new home,
except that I can see the issue locally.
FWIW, the way skink runs all postgres instances through valgrind is by
replacing the postgres binary. Here's my local version of that:
andres@awork3:~/build/postgres/dev-assert/vpath$ cat
tmp_install/home/andres/build/postgres/dev-assert/install/bin/postgres
#!/bin/bash
exec /usr/bin/valgrind \
--quiet \
--error-exitcode=128 \
--suppressions=/home/andres/src/postgresql/src/tools/valgrind.supp \
--trace-children=yes --track-origins=yes --read-var-info=no \
--leak-check=no \
--run-libc-freeres=no \
--vgdb=no \
--error-markers=VALGRINDERROR-BEGIN,VALGRINDERROR-END \
/home/andres/build/postgres/dev-assert/vpath/tmp_install/home/andres/build/postgres/dev-assert/install/bin/postgres.orig
\
"$@"
make -C src/test/recovery/ check PROVE_FLAGS='-v' PROVE_TESTS='t/026_overwrite_contrecord.pl' NO_TEMP_INSTALL=1
...
not ok 1 - 000000010000000000000002 differs from 000000010000000000000002
# Failed test '000000010000000000000002 differs from 000000010000000000000002'
Regards,
Andres