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

[Varnish]

#729: Varnish loosing all its cache content
----------------------+-----------------------------------------------------
 Reporter:  fabrice   |       Owner:  phk
     Type:  defect    |      Status:  new
 Priority:  high      |   Milestone:     
Component:  varnishd  |     Version:  2.0
 Severity:  critical  |    Keywords:     
----------------------+-----------------------------------------------------
 We are using Varnish (on 2 servers) to cache our images coming from Apache
 (2 servers as well). The problem we have is that couples of times a day,
 Varnish looses all it cache content.
 It is quite random and luckily, the 2 Varnish servers did not do that yet
 at the same time (which would kill our Apache backend). In any case,
 varnishstat shows about 90% hit ratio and all of the sudden drops to 0 and
 builds back its cache.

 The only thing in the logs that looks a bit strange are messages like
 this:
 Jun 23 11:19:15 cache1 varnishd[17821]: Child (5617) not responding to
 ping, killing it.
 Jun 23 11:19:39 cache1 varnishd[17821]: Child (5617) died signal=3

 But even if those messages are strange (or maybe not) they do not relate
 in time with the issue we are having.

 The daemon settings are:
 -f /opt/blocket/conf/varnish.vcl -s malloc,63G -u varnish -g varnish -p
 thread_pools 4 -p thread_pool_min 32 -p listen_depth 4096 -p lru_interval
 600 -h classic,400009 -p obj_workspace 8192 -p log_hashstring off -p
 shm_workspace 16384 -p srcaddr_ttl 0 -p ping_interval 2 -p default_grace
 3600 -p pipe_timeout 10 -p sess_timeout 5 -p send_timeout 10

 Our VCL is attached (with some offuscated data)

 Maybe this is something very stupid on our side but any idea is very
 welcome to help us here.
 Thanks,
 Fabrice.

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