ctl_cyrusdb segmentation fault every 30+90min after checkpointing cyrus databases

[Matthias Busch]

Hey,

(kolab 3.3 on debian 7.6)

i noticed segmentation faults in my syslog for ctl_cyrusdb
grepping for them I noticed they happen, happen again 30mins later
and again 2hrs after the first one (or 1.5 after the second)

here a copy/paste from the syslog with grep for the fault

Sep 20 12:14:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.005s pid:4357 signaled to death by signal 11 (Segmentation fault)
Sep 20 12:44:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.005s pid:4421 signaled to death by signal 11 (Segmentation fault)
Sep 20 14:14:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.007s pid:4599 signaled to death by signal 11 (Segmentation fault)
Sep 20 14:44:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.006s pid:4691 signaled to death by signal 11 (Segmentation fault)
Sep 20 16:14:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.005s pid:4900 signaled to death by signal 11 (Segmentation fault)
Sep 20 16:44:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.007s pid:4958 signaled to death by signal 11 (Segmentation fault)
Sep 20 18:14:53 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.005s pid:5574 signaled to death by signal 11 (Segmentation fault)

I then looked in the syslog what is occuring at those times. i found checkpointing immediately proceeds the segfault.

so I grepped for it:

Sep 20 12:14:51 h2343028 ctl_cyrusdb[4357]: checkpointing cyrus databases
Sep 20 12:44:51 h2343028 ctl_cyrusdb[4421]: checkpointing cyrus databases
Sep 20 13:14:51 h2343028 ctl_cyrusdb[4474]: checkpointing cyrus databases
Sep 20 13:14:51 h2343028 ctl_cyrusdb[4474]: done checkpointing cyrus databases
Sep 20 13:44:51 h2343028 ctl_cyrusdb[4532]: checkpointing cyrus databases
Sep 20 13:45:04 h2343028 ctl_cyrusdb[4532]: done checkpointing cyrus databases
Sep 20 14:14:51 h2343028 ctl_cyrusdb[4599]: checkpointing cyrus databases
Sep 20 14:44:51 h2343028 ctl_cyrusdb[4691]: checkpointing cyrus databases
Sep 20 15:14:51 h2343028 ctl_cyrusdb[4768]: checkpointing cyrus databases
Sep 20 15:14:51 h2343028 ctl_cyrusdb[4768]: done checkpointing cyrus databases
Sep 20 15:44:51 h2343028 ctl_cyrusdb[4838]: checkpointing cyrus databases
Sep 20 15:44:51 h2343028 ctl_cyrusdb[4838]: done checkpointing cyrus databases
Sep 20 16:14:51 h2343028 ctl_cyrusdb[4900]: checkpointing cyrus databases
Sep 20 16:44:51 h2343028 ctl_cyrusdb[4958]: checkpointing cyrus databases
Sep 20 17:14:51 h2343028 ctl_cyrusdb[5022]: checkpointing cyrus databases
Sep 20 17:14:51 h2343028 ctl_cyrusdb[5022]: done checkpointing cyrus databases
Sep 20 17:44:51 h2343028 ctl_cyrusdb[5183]: checkpointing cyrus databases
Sep 20 17:44:51 h2343028 ctl_cyrusdb[5183]: done checkpointing cyrus databases
Sep 20 18:14:53 h2343028 ctl_cyrusdb[5574]: checkpointing cyrus databases

I found that the checks happen every 30minutes

when they complete I get:

Sep 20 17:14:51 h2343028 ctl_cyrusdb[5022]: checkpointing cyrus databases
Sep 20 17:14:51 h2343028 ctl_cyrusdb[5022]: done checkpointing cyrus databases

but as we can see above, every 30+90 minutes they fail. there I get:

Sep 20 14:14:51 h2343028 ctl_cyrusdb[4599]: checkpointing cyrus databases
Sep 20 14:14:51 h2343028 master[3073]: process type:EVENT name:checkpoint path:/usr/lib/cyrus-imapd/ctl_cyrusdb age:0.007s pid:4599 signaled to death by signal 11 (Segmentation fault)


sooo... I dont know WHY the segfault occurs.
I dont know why it occurs at predictable time intervalls
and I dont know what is also happening at those intervalls that could cause that?

do you guys have any ideas?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kolab.org/pipermail/users/attachments/20140920/d9afb5e7/attachment.html>