qmail-local.mov segfault errors

vsystems · Unread post by **vsystems** » Tue Jul 10, 2007 9:15 am

Hi,

I'm getting quite a few of the following errors in my messages logs;

Jul 10 15:36:23 server kernel: qmail-local.mov[30798]: segfault at 00000000bffffd94 rip 00000000004088f7 rsp 0000007fbffff7b8 error 4

I didn't initially believe this to be a problem until one of my hosting clients reported a failed send mail notification reporting the following error;

Aack, child crashed. (#4.3.0)

The server is a Plesk 8.1.1 on CentOS v4.4 (64bit).
psa-qmail-1.03-cos4.build81070322.16

Unfortunately I'm a somewhat new to qmail / linux, and wasn't able to find anything doing a google search, so I'm hoping there might be some gurus out there that could point me in the right direction on how I can work out what's causing this and how (potentially) I can fix this...

Thanks in advance!

Troy McClure · Unread post by **Troy McClure** » Mon Sep 17, 2007 4:36 pm

Did you ever find out anything about this? I get these every once in a while too and have also gotten the Aack, child crashed. I am on CentOS 5 64-bit, Plesk 8.2.

Unread post by **scott** » Mon Sep 17, 2007 8:18 pm

try setting "noacpi" in your kernel boot line.

example from grub.conf:

kernel /vmlinuz-2.6.22.4-65.fc7 ro root=/dev/WinonaGroup00/LogVol00 rhgb quiet selinux=0 noacpi

Troy McClure · Unread post by **Troy McClure** » Tue Sep 18, 2007 9:18 am

Thanks Scott, I will try that.

Troy McClure · Thu Oct 25, 2007 11:54 am

Well, I still get these sometimes. I have now installed your asl testing kernel (2.6.22.9-2.art) and am still getting this. I just had one happen that I could get more information about so here goes. The recipients mailbox filled up during the delivery I think. A user sent 3 emails with pictures in them and the first one came back with aack crash, then the next two came back with recipients mailbox full. I have also found out recently that hard disk quotas are not supported right now on my server. My /etc/fstab has this line

/dev/md0 / ext3 defaults 1 1

when I think it should have this line

/dev/md0 / ext3 defaults,usrquota 1 1

Could that cause this to happen and is it ok to go ahead and edit that line and reboot?

Unread post by **scott** » Thu Oct 25, 2007 1:46 pm

Maybe its the size of the message causing that? Or did you rule that out already

Yep its ok to turn quotas on like that.

Troy McClure · Unread post by **Troy McClure** » Fri Oct 26, 2007 3:08 pm

I don't think it is the size. These were only 2mb each and I know other messages much bigger go through fine. I only get these every once in a while (like 1 a week). Do you think quotas could have anything to do with this? I am going to enable quotas tonight, so I guess I will find out soon enough.

Unread post by **scott** » Fri Oct 26, 2007 4:10 pm

no I dont think so. Size or maybe encoding would be the first things I'd look into.