17 Jan
2011
17 Jan
'11
4:55 p.m.
On Mon, Jan 17, 2011 at 11:32:13AM +0000, Ian Hobson wrote:
Hi all,
My fastcgi process has just stopped twice recently, and I am having
trouble finding out why.
History
Recent upgraded VPS to Ubuntu 10.4. from 8.4 - apparently without
problem.
Running nginx with php via init-fastcgi script
nginx was not changed in the upgrade, neither was the init-fastcgi
script.
Oddities noticed.
/etc/init.d/init-fastcgi stop took 30 seconds but gave no errors.
I could then start the process fine without error.
I can find nothing in the logs that is suspicious, for either time.
This morning Baiduspider was spidering a site running an old version of
Word Press at about the time the problem started, but did not
get a 504 error.
I wish I had run ps -ax to see if the processes were in the machine. I
ran it, and nginx was fine, but did not check the php side. (still
asleep).
Has anyone any ideas about what might be happening, what could cause it,
or indeed anything helpful to stopping it happening again?
I'd start by enabling debugging, if that doesn't show you where it's
breaking then I'd turn on core dumps and perhaps look at that with
pstack or gdb to see where's its faulting. That may not provide the
answer but it might give some clues.
If you have the space perhaps simpler methods might be to use strace,
follow children and write to files. But since that (mostly) just traces
syscalls (unless using other tracers) it may not show what fastcgi is
doing.
See what enabling debugging yields first, it might be simple to locate.
--
Best regards,
Ed http://www.s5h.net/