Hi Guys,

I'm running a Fedora linode with 2.4.26-linode26-1um. The uptime of the box is/was 24 days.

The machine runs a number of standard services (mail,pop,apache etc ..) and I haven't had any problems or changed any config for a while - when I logged on earlier all the services mentioned above had stopped.

Checking services logs I see :

apache : [Sun May 16 04:23:03 2004] [notice] seg fault or similar nasty error detected in the parent process
exim : 2004-05-16 04:26:43 1BPCAd-0006oU-Vf daemon: delivery process fork failed: Cannot allocate memory

In /var/log/messages :

May 16 07:44:24 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 07:48:52 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 07:48:52 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 07:48:53 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0xf0/0)
May 16 07:48:53 robl kernel: ENOMEM in do_get_write_access, retrying.
May 16 07:48:53 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 07:48:56 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 07:49:23 robl kernel: __alloc_pages: 2-order allocation failed (gfp=0x1f0/0)
May 16 07:53:24 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)
May 16 08:08:21 robl kernel: __alloc_pages: 0-order allocation failed (gfp=0x1d2/0)


Having done a little googling for the above kernel error messages I found :

http://groups.google.com/groups?hl=en&l ... 26rnum%3D8

Which seems to imply it's a kernel issue ? Anybody else have any thoughts on this ? I think I'm going to move back to the last kernel I was using as I moved to this kernel a month or so ago when I noticed it was announced.

Anybody else see the same problem ?

Rob

The thread you refer to is pretty old - they were working on 2.4.10.

It looks like you ran out of memory.

What Linode have you got and how big is your swap partition?

It's entirely possible I ran out of memory - that will be the next MRTG graph I suppose !

I've only got 64Mb Linode with 128Mb swap.

I'm a little suprised as the box wasn't doing anything exciting from my load and traffic graphs and 128Mb swap whilst not being huge isn't bad.

I'll set up some more monitoring and keep an eye on mem usage I guess.