|
From: Szabo E. <sz...@on...> - 2007-02-27 16:07:27
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Matthew Hattersley wrote: > Hi guys, > > I'm having a weird one here. Processes keep dying on my DL boxes (I have > over 70 out in the wild). I have over 20 boxes running 1.2.8 and 1.2.11 but never experienced the same. I use them for openvpn termination though. [...] > __alloc_pages: 0-order allocation failed (gfp=0x1d2/0) > VM: killing process procinfo > __alloc_pages: 0-order allocation failed (gfp=0xb2/0) [...] > This is the output from dmesg. This kills a load of daemons which really > messes things up. Any ideas? maybe procinfo? it is (re)spawned by the init: # grep proc /etc/inittab 8:2345:respawn:env - TERM=linux /bin/procinfo -f -F /dev/tty8 AFAIK VM kills the process that uses the maximum RSS (the non-swapped physical memory that the process used). Look for any memleaking ever-growing process during normal operation. Good luck, Endre -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFF5FcqnhpQUdN8PPERAjiHAJ4+LQlMOAuT7BpT2nxGeDB7e9J9IQCgifi+ U4k/J9kwOKcU57YIGvGB/VY= =Eisl -----END PGP SIGNATURE----- |