Quote:
Originally Posted by Sephirot
No it did not happend again after that day. It seems to be a very rare case.
I reduced the shm_prune_period and shm_ttl from 3600s to 1800s.
eAccelerator is set to use both, shared memory and disk (default). Maybe I should set it to "shm", which causes eAccelerator to use the disk only if the shared mem is full?
|
I'm no real expert on eAccelerator sadly, so I'm not sure if this would solve the rare case. It may even be a bug with eAccel but I can't be sure on that. If you change the setting and it doesn't occur again then I'd say it's worked, but it might be a cat and mouse game with trial and error unfortunately.