[Bucardo-general] perl SV memory leak?
Mitchell Perilstein
Mitchell.Perilstein at trueposition.com
Thu Jul 11 20:42:21 UTC 2013
Hi all,
We're running 4.99.7 (PG 9.1.5, solaris) under fair amount of load and
have been seeing some unreasonable memory growth in some KIDs and
CTLs. I slayed a CTL today (manually with a sigusr1) that had grown to
1.1 Gb--the MCP happily created a new CTL and its child KID, and they
picked up where they left off, great.
If I poke through the pid dir and examine each process, a few of its
remaining unslayed friends are still growing (rss and sz are in kb).
PID PPID RSS SZ %MEM STIME PROC
3420 4712 15208 2436 0.1 15:52:56
bucardo.ctl.sync.lg_sync.pid
4730 4712 15344 2436 0.1 Jun_27
bucardo.ctl.sync.o1_sync.pid
4742 4712 15264 2436 0.1 Jun_27
bucardo.ctl.sync.o2_sync.pid
4736 4712 15264 2436 0.1 Jun_27
bucardo.ctl.sync.sg_sync.pid
3424 3420 15584 2436 0.1 15:52:56
bucardo.kid.sync.lg_sync.pid
4738 4730 721800 152452 4.6 Jun_27
bucardo.kid.sync.o1_sync.pid -- this one, 700 mb rss
4748 4742 24088 3460 0.2 Jun_27
bucardo.kid.sync.o2_sync.pid
4743 4736 202584 37252 1.3 Jun_27
bucardo.kid.sync.sg_sync.pid -- and this one, 202 mb rss
4712 1 16104 2436 0.1 Jun_27 bucardo.mcp.pid
4722 4712 14872 2436 0.1 Jun_27 bucardo.vac.pid
Observations:
a. It's not continual growth; it looks spurty; so I'm trying to figure
out what event might correlate to a growth period.
b. I've started using Devel::Leak but it's nothing concrete yet.
c. Finally, wondering if this DBI leak has ever been encountered here?
http://stackoverflow.com/questions/13338308/perl-dbi-memory-leak
Thanks!
Confidentiality Notice: This e-mail (including any attachments) is intended only for the recipients named above. It may contain confidential or privileged information and should not be read, copied or otherwise used by any other person. If you are not a named recipient, please notify the sender of that fact and delete the e-mail from your system.
More information about the Bucardo-general
mailing list