[peruser] [renton at 1gb.ru: broken chroot env]

Alexey Vlasov renton at 1gb.ru
Wed Sep 29 04:30:00 MDT 2010


On Wed, Sep 29, 2010 at 09:08:36AM +0200, Taavi Sannik wrote:
> 
> Is this with rc1 or rc2?

It seems to me that versions aren't the reason, it has always been so in
each version.
Now I use rc1.

> I have only seen this kind of behavior if you gracefully restart the
> server and look at the server-status at the same moment (with keepalive
> enabled), then the server-status displays this corrupted child table
> (probably because of the pool has been thrown away).
> 
> Can you confirm this happens after a graceful restart or randomly?

I found only one regularity, the child table gets corrupted only at
certain configuration of the virtual hosts. I refresh httpd.conf, do
graceful - the table is corrupted, graceful once more - the table is ok
(but can be corrupted again), then I do kill -9 - corrupted (but can
help randomly). So I don't know how it randomly got fixed. But if I take
the same config file of the virtual hosts at which I receive the
corrupted surrounding and transfer it on another Apache, the table will
be OK. And what is more surprising to me, in an hour this very config on
the same Apache will give the normal working table.

The only thing that definitely helps is to add or delete (or to mix)
some virtual hosts in httpd.conf. We likely would repair it ourselves,
but it is still not clear to us how to approach the decision as it is
impossible to transfer the un-working configuration on another Apache or
even to play it on operating one in some time interval.

-- 
BRGDS. Alexey Vlasov.


More information about the Peruser mailing list