Tom Lane wrote:
I'm noticing that buildfarm member hyena sometimes fails the parallel
regression tests, for instance
http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=hyena&dt=2006-07-19%2009:20:00
The symptom is always that one of the tests fails entirely because
psql couldn't connect:
psql: could not connect to server: Connection refused
Is the server running locally and accepting
connections on Unix domain socket "/tmp/.s.PGSQL.55678"?
It's a different test failing in each occurrence. Sometimes there are
ensuing failures in subsequent tests that expect the side-effects
of the one that failed, but there's clearly a common cause here.
AFAIK it is not possible for Postgres itself to cause a "connection
refused" failure --- that's a kernel-level errno. So what's going on
here? The only idea that comes to mind is that this version of Solaris
has some very low limit on SOMAXCONN, and when the timing is just so
it's bouncing connection requests because several of them arrive faster
than the postmaster can fork off children. Googling suggests that there
are versions of Solaris with SOMAXCONN as low as 5 :-( ... but other
pages say that the default is higher, so this theory might be wrong.
What is SOMAXCONN set to on that box, anyway? If it's tiny, I suggest
you increase SOMAXCONN to something saner, or if you can't, run "make
check" with MAX_CONNECTIONS=5 added to the make command. Does the
buildfarm script have provisions for site-local settings of this
parameter?
Yes it sure does.
This is the box that Sun donated, btw.
I get: ndd /dev/tcp tcp_conn_req_max_q => 128
Is that the Solaris equivalent of SOMAXCONN? That's low, maybe, but not
impossibly low.
I don't have root on the box, though. For now I have set MAX_CONNECTIONS
to 8, to provide a modest limit on parallelism. I will see if I can
coordinate with Robert and Josh to increase the OS limits.
Thanks for the diagnosis.
cheers
andrew
---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to [EMAIL PROTECTED] so that your
message can get through to the mailing list cleanly
