Hi Spacerat
I;ve got similar messages after I upgraded from 5.0.4 to 5.0.5 tonight.
I didn't get them during installation but I did get the same messages afterwards during a manual startup.
We've been running zimbra happily in an OpenVZ VPS (the opensource version of Virtuozzo) for almost two years without problems.
I only seem to get the messages when I manually start the server with 'zmcontrol start' it gives me the error as zmstats is starting.
Code:
Starting ldap...Done.
Starting logger...Done.
Starting mailbox...Done.
Starting antispam...Done.
Starting antivirus...Done.
Starting snmp...Done.
Starting spell...Done.
Starting mta...Done.
Starting stats...*** glibc detected *** free(): invalid pointer: 0xb7e76280 ***
Done.
*** glibc detected *** free(): invalid pointer: 0xb7e76280 ***
[zimbra@mail /]$ *** glibc detected *** free(): invalid pointer: 0xb7e76280 ***
It must try to start it again shortly after, because although zmcontrol hangs and I have to close it with ctrl-c, the stats do come up after a minute or so.
As per the ubuntu thread referred to by Spacerat I've tried decreasing the defualt java_heap size from 30 to 20, and I've been increasing the privvmpages allocated to the zimbra vps.
As I'v raised it, the failcnt column has decreased (that 177 in the fail cnt column has only gone up by 2 since I made my last resource increase, which seems to correspond nicely to the number glibc errors I get.) As I increase the privvmpages I've received fewer lines of the glibc error.
at the moment the resource allocation looks like this (about a 1/3 of the server):
Code:
uid resource held maxheld barrier limit failcnt
800: kmemsize 10421733 13875147 275052407 302557647 0
lockedpages 0 1103 13430 13430 0
privvmpages 832631 2267817 2307990 2407990 177
shmpages 16 1312 60436 60436 0
dummy 0 0 0 0 0
numproc 146 239 6714 6714 0
physpages 152402 211030 0 9223372036854775807 0
vmguarpages 0 0 604363 9223372036854775807 0
oomguarpages 152402 211030 604363 9223372036854775807 0
numtcpsock 60 100 6714 6714 0
numflock 49 311 1000 1100 0
numpty 1 2 512 512 0
numsiginfo 0 14 1024 1024 0
tcpsndbuf 769336 1346536 64183591 91684135 0
tcprcvbuf 1033376 1687352 64183591 91684135 0
othersockbuf 178360 270776 32091795 59592339 0
dgramrcvbuf 0 105544 32091795 32091795 0
numothersock 125 161 6714 6714 0
dcachesize 0 0 60074003 61876224 0
numfile 3390 4657 107424 107424 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
numiptent 140 140 200 200 0 I guess the two things I'd like to know are:
- Where do I look to try and narrow down exactly whats causing this? (no errors in zinbra.log, there are some stacktrace.xxxxx files in the /opt/zimbra/log directory
- Can anyone suggest any more tweaks to java/zimbra settings? (Its been working fine for ages, is there some new memory hungry feature in 5.0.5?)
Cheers.