[Varnish] #729: Varnish loosing all its cache content

[Varnish]

#729: Varnish loosing all its cache content
----------------------+-----------------------------------------------------
 Reporter:  fabrice   |        Owner:  phk       
     Type:  defect    |       Status:  closed    
 Priority:  high      |    Milestone:            
Component:  varnishd  |      Version:  2.0       
 Severity:  critical  |   Resolution:  worksforme
 Keywords:            |  
----------------------+-----------------------------------------------------
Changes (by phk):

  * status:  new => closed
  * resolution:  => worksforme


Comment:

 Hi Chris,

 Those messages (not responding to ping, killing it.) are indeed relevant:
 They tell that the master process could not get a CLI reply out of the
 child inside the timeout, and therefore have decided that it is stuck,
 killed it, and started a fresh copy.

 This usually happens because your varnish child process gets stuck in
 disk-I/O and/or paging activity.

 If the responsetime is otherwise OK for you (use varnishhist to monitor
 it) you can increase the parameter "cli_timeout" to make the master
 process more patient.

 Poul-Henning

-- 
Ticket URL: <http://varnish-cache.org/ticket/729#comment:2>
Varnish <http://varnish-cache.org/>
The Varnish HTTP Accelerator