[Varnish] #639: Assert on writing to CLI when child process dies
Varnish
varnish-bugs at projects.linpro.no
Mon Feb 1 11:22:47 CET 2010
#639: Assert on writing to CLI when child process dies
------------------------------+---------------------------------------------
Reporter: tgr | Owner: phk
Type: defect | Status: new
Priority: normal | Milestone:
Component: varnishd | Version: 2.0
Severity: normal | Resolution:
Keywords: 2.0.6 CLI assert |
------------------------------+---------------------------------------------
Old description:
> varnishd 2.0.6 on Debian Lenny amd64.
>
> When a child dies:
>
> Jan 29 10:51:10 webcache4 varnishd[24741]: Child (24742) not responding
> to ping, killing it.
> Jan 29 10:51:12 webcache4 varnishd[24741]: Child (24742) not responding
> to ping, killing it.
>
> and there's a CLI command being written to the child, an xxxassert() is
> raised:
>
> #3 0x000000000042ef64 in mgt_cli_vlu (priv=0x7fb112813c00,
> p=0x7fb1128d3000 "debug.health") at mgt_cli.c:270
> 270 xxxassert(i == strlen(p));
>
> The master process should handle the unfinished write(), rather than
> aborting.
>
> As a dirty interim fix, it *should* be OK to ignore the unfinished write,
> like in the attached patch.
New description:
varnishd 2.0.6 on Debian Lenny amd64.
When a child dies:
{{{
Jan 29 10:51:10 webcache4 varnishd[24741]: Child (24742) not responding to
ping, killing it.
Jan 29 10:51:12 webcache4 varnishd[24741]: Child (24742) not responding to
ping, killing it.
}}}
and there's a CLI command being written to the child, an xxxassert() is
raised:
{{{
#3 0x000000000042ef64 in mgt_cli_vlu (priv=0x7fb112813c00,
p=0x7fb1128d3000 "debug.health") at mgt_cli.c:270
270 xxxassert(i == strlen(p));
}}}
The master process should handle the unfinished write(), rather than
aborting.
As a dirty interim fix, it *should* be OK to ignore the unfinished write,
like in the attached patch.
Comment (by phk):
fixed formatting
--
Ticket URL: <http://www.varnish-cache.org/ticket/639#comment:1>
Varnish <http://varnish.projects.linpro.no/>
The Varnish HTTP Accelerator
More information about the varnish-bugs
mailing list