decoding a panic

[Carlos Abalde]

> On 21 Nov 2014, at 20:03, Jason Price <japrice at gmail.com> wrote:
> 
> Recently, my varnish boxes have been seeing panics.  I don't know why.  vmod redis, mhash, and digest are in play (with the worst suspect being vmod_redis).
> 
> Last panic at: Fri, 21 Nov 2014 17:16:55 GMT                                                                                                                              [92/197]
> Assert error in VRT_SetHdr(), cache_vrt.c line 225:
>   Condition((sp)->magic == 0x2c2f9c5a) not true.
> ...
> 
> Any pointers on where to start looking?

Hi Jason,

That could be related with some VMOD corrupting memory contents, but I don’t think that panic message could be useful to find the responsible. If the panic is easy to reproduce, you could use a simple trial and error approach in order to try to identify the origin of the problem.

What Redis VMOD are you using? If using https://github.com/carlosabalde/libvmod-redis <https://github.com/carlosabalde/libvmod-redis>, what version of the VMOD, Varnish, hiredis & Redis are you using? During the last 24 hours I’ve running a benchmark of the VMOD (Varnish 3.0.6, hiredis 0.11.0, Redis 2.8.17, ~ 2,500 HTTP reqs/sec, ~ 24,000 Redis ops/sec) while monitoring syslog, uptime, etc. and I didn’t get any Varnish crashes. If you’re using this VMOD and you think the problem could be related with it, it could be useful if you could share your VCL.

Best,

—
Carlos Abalde.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.varnish-cache.org/lists/pipermail/varnish-misc/attachments/20141124/920e8713/attachment.html>