1

The following error message is taken from my log message. varnishd server is getting killed for over use of memory.

We have done all the possible settings to avoid it, Still the same problem happens with multiple VMs, not just one rare incident. 

Aug 26 08:26:25 host kernel: [15332346.565597] Out of memory in UB 206: OOM killed process 24966 (varnishd) score 0 vm:520484kB, rss:203660kB, swap:12124kB
Aug 26 08:26:25 host varnishd[23199]: Child (24966) died signal=9

We are having multiple VMs, most of them are running websites. In some VMs, where the load is high, due to limited RAM, the varnishd is getting killed.

Some relavent configuration value from /etc/sysconfig/varnish

VARNISH_MAX_THREADS=1000
VARNISH_STORAGE_SIZE=256M

Server is having 2GB RAM.

What configuration needs to be done for varnishd to avoid this problem?

Mani
  • 554
  • 6
  • 11
  • 26
  • Seems like, It cant be done. But, anyhow, lets see the answer from others. Check , http://stackoverflow.com/a/16693895/372108 – Mani Aug 27 '16 at 04:55
  • You need more memory, either physical RAM or swap. Swap would might slow down things. Trying to alter the `oom_score` will only result in killing any other process which might lead to other problems, as you write that this happens all the time. Look at the memory usage of the running processes with `top` or `htop` and check if you can shutdown those processes or improve their memory usage. – Thomas Aug 27 '16 at 09:08

0 Answers0