Re: failures on barnacle (CLOBBER_CACHE_RECURSIVELY) because of memory leaks

"Tomas Vondra" <tv(at)fuzzy(dot)cz> writes:
> So after 83 days, the regression tests on barnacle completed,

Hah, that's perseverance!

> and it
> smells like another memory leak in CacheMemoryContext, similar to those
> fixed in 078b2ed on May 18.

Ugh, will look.

> See this:
> http://pgbuildfarm.org/cgi-bin/show_log.pl?nm=barnacle&dt=2014-05-19%2011%3A25%3A03

Initially I was more worried about the crashes, but it looks like that's
probably just a side-effect of the leak:

[537a053d.1f4b:2] LOG: server process (PID 32110) was terminated by signal 9: Killed
[537a053d.1f4b:3] DETAIL: Failed process was running: select pg_get_viewdef('vv3', true);
[537a053d.1f4b:4] LOG: terminating any other active server processes
[537a053d.1f52:2] WARNING: terminating connection because of crash of another server process
...
[537a053d.1f4b:7] LOG: server process (PID 3882) was terminated by signal 9: Killed
[537a053d.1f4b:8] DETAIL: Failed process was running: SELECT gid FROM pg_prepared_xacts;
[537a053d.1f4b:9] LOG: terminating any other active server processes
...

Evidently the OOM killer is at large on this machine.

regards, tom lane