Hey guys,
we've been done some semi-regular testing of check_postgres against 8.5/9.0.
Most everything has worked so far, but we found a breakage in "same_schema"
check in alpha4.
rtm:check_postgres robert$ ./check_postgres.pl --action=same_schema --
host=mdb1.int.c.net --host2=mdb1.dev.c.net -u reconnoiter --dbname=reconnoiter
--port=5432 --port2=5490
ERROR: ERROR: column "tgisconstraint" does not exist
LINE 1: ...d) JOIN pg_proc p ON (p.oid = tgfoid) WHERE NOT tgisconstr...
The problem stems from this change:
http://git.postgresql.org/gitweb?p=postgresql.git;a=commit;h=2f54037dc8e957cb843587b27c7cd4f586a41fd7
quite simply, the column definitions of pg_trigger have been changed. the code
is around line 4692:
4688 if (! exists $filter{notriggers}) {
4689 $SQL = q{SELECT tgname, quote_ident(relname),
proname, proargtypes FROM pg_trigger }
4690 . q{ JOIN pg_class c ON (c.oid = tgrelid)
}
4691 . q{ JOIN pg_proc p ON (p.oid = tgfoid) }
4692 . q{ WHERE NOT tgisconstraint}; ##
constraints checked separately
4693 $info = run_command($SQL, { dbuser =>
$opt{dbuser}[$x-1], dbnumber => $x } );
would seem like a quick if based on version for that where clause should fix
it, but figured i'd just poke the list rather than a patch in case you want it
done differently.
--
Robert Treat
Conjecture: http://www.xzilla.net
Consulting: http://www.omniti.com