1

I got an VPS at burst.net, very cheap one, but this doesn't matter.

The strange behavior of it is, the php-cgi process, which started using lighthttpd's spawn-cgi, dies every a few minutes. however, other processes are fine and great, even include a java process, and I'm sure there is no "out of memory" issue, so it's not killed by OOM killer.

I used strace to trace the process, and found out it was killed by SIGKILL, hence no single log was left on the disk, just dies suddenly.

Is there anyway I can find out what process/thing sent the SIGKILL to the poor php process?

Filed a ticket with the vendor, but they said they won't care.

strace -p 7176
Process 7176 attached - interrupt to quit
wait4(-1, <unfinished ...>
+++ killed by SIGKILL +++
Improve this question
3
  • What kind of virtualization is used for your VM?
    – Mircea Vutcovici
    Dec 23, 2011 at 7:38
  • @mircea-vutcovici openvz it is
    – Shawn
    Dec 23, 2011 at 18:55
  • I've the same problem. It seems burst.net is killing cron daemon regularly.
    – chris
    Sep 23, 2012 at 18:10

3 Answers 3

Reset to default
1

Finally I recompiled the php using php-fpm, that seems to be much better and steady. anyway, openvz is not as good. but you pay that little, what to expect?

Improve this answer
2
  • Don't settle for OpenVZ crap; even at the low end you can find decent Xen and KVM hosts.
    – Michael Hampton
    Oct 5, 2012 at 4:27
  • Make sure to accept this as your answer, so it's clear what solved your issue. (When you can in a couple days).
    – HopelessN00b
    Oct 5, 2012 at 9:53
0

A singnal can be send only be a process run as the same user as the destination or as root. You can use systemtap to find the culprit. See: SystemTap SigMon

Improve this answer
2
  • thanks..this is a cool however it seems I need some new kernel to get it to work. yum install systemstap will give me a new kernel and I doubt it will not work on the VPS -- using openvz. probably I should find something using Xen
    – Shawn
    Dec 23, 2011 at 18:57
  • (systemtap does not require a new kernel; people have used it fine under a xen domu.)
    – fche
    Dec 13, 2013 at 16:37
-1

After asking the burst.net support, they changed something on the server. Cron is running continuously now.

It seems like crond was killed by the host system.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .