Ticket #1240 (closed defect: invalid)

Opened 16 months ago

Last modified 16 months ago

Varnish keeps on restarting every 2 hours

Reported by: redline Owned by:
Priority: normal Milestone:
Component: build Version: 3.0.3
Severity: normal Keywords:
Cc:

Description

Hello ,

My server is running on CentOS 6.3 64bits Varnish 3.0.3 keeps on restarting every 2 hours .

I tried using malloc and file as storage , both causes the problem .

This is the error that i see in /var/log/messages :

Dec 22 15:14:52 us1 varnishd[25636]: Child (25637) died signal=15 Dec 22 15:14:52 us1 varnishd[25636]: child (8257) Started Dec 22 15:14:52 us1 varnishd[25636]: Child (8257) said Child starts Dec 22 15:14:52 us1 varnishd[25636]: Child (8257) said SMF.s0 mmap'ed 7896244224 Dec 22 17:14:04 us1 varnishd[25636]: Child (8257) died signal=15 Dec 22 17:14:04 us1 varnishd[25636]: child (27043) Started Dec 22 17:14:04 us1 varnishd[25636]: Child (27043) said Child starts Dec 22 17:14:04 us1 varnishd[25636]: Child (27043) said SMF.s0 mmap'ed 789624422 Dec 22 17:18:49 us1 varnishd[25636]: Manager got SIGINT Dec 22 17:18:51 us1 varnishd[28073]: Platform: Linux,2.6.32-220.17.1.el6.x86_64, Dec 22 17:18:51 us1 varnishd[28073]: child (28074) Started Dec 22 17:18:51 us1 varnishd[28073]: Child (28074) said Child starts Dec 22 17:18:51 us1 varnishd[28073]: Child (28074) said SMF.s0 mmap'ed 789776384

What should cause this problem ?

Thank you.

Change History

comment:1 Changed 16 months ago by redline

The error :

Dec 22 15:14:52 us1 varnishd[25636]: Child (25637) died signal=15
Dec 22 15:14:52 us1 varnishd[25636]: child (8257) Started
Dec 22 15:14:52 us1 varnishd[25636]: Child (8257) said Child starts
Dec 22 15:14:52 us1 varnishd[25636]: Child (8257) said SMF.s0 mmap'ed 7896244224
Dec 22 17:14:04 us1 varnishd[25636]: Child (8257) died signal=15
Dec 22 17:14:04 us1 varnishd[25636]: child (27043) Started
Dec 22 17:14:04 us1 varnishd[25636]: Child (27043) said Child starts
Dec 22 17:14:04 us1 varnishd[25636]: Child (27043) said SMF.s0 mmap'ed 789624422
Dec 22 17:18:49 us1 varnishd[25636]: Manager got SIGINT
Dec 22 17:18:51 us1 varnishd[28073]: Platform: Linux,2.6.32-220.17.1.el6.x86_64,
Dec 22 17:18:51 us1 varnishd[28073]: child (28074) Started
Dec 22 17:18:51 us1 varnishd[28073]: Child (28074) said Child starts
Dec 22 17:18:51 us1 varnishd[28073]: Child (28074) said SMF.s0 mmap'ed 789776384
Last edited 16 months ago by redline (previous) (diff)

comment:2 Changed 16 months ago by kristian

  • Status changed from new to closed
  • Resolution set to invalid

This could be a matter of hitting ulimit (e.g: number of file descriptors exceeded, memory usage etc). That's a reasonable explanation for getting SIGTERMs.

It's unlikely that this is an actual Varnish bug, however, since there are no known scenarios where Varnish itself would generate a SIGTERM.

If you are using the distro-specific bootup scripts, they should deal with ulimit correctly. If you've rolled your own, it's worth looking at what the distro-provided script do in that regard.

Regardless, since we do not suspect that this is a bug, I recommend that you use the "varnish-misc" mail list - it has far more subscribers than our bug tracking system too.

 https://www.varnish-cache.org/lists/mailman/listinfo

For now, I'm closing this ticket. If, after soliciting help from varnish-misc, you still suspect a bug, feel free to re-open or re-post this issue and we'll look closer at it.

Note: See TracTickets for help on using tickets.