[UCIMC-Tech] http://lists.chambana.net wedged again...

[Stuart Levy]

Yep - that makes sense.  I'm just writing as a frustrated list-admin, 
without the power to give the thing a kick (i.e. I can't test what's 
suggested below), who sometimes finds important messages held for 
moderation which I can't pass through to my e-mail lists because the 
whole web server has gone unresponsive.

The answer so far has been to ask Chris/Josh to kick the web server, and 
they've been very good about doing it, but we can't expect them to be on 
call around the clock.  If any useful info (ps alx, netstat -nt, tail 
/var/log/httpd/access_log, whatever) can be gathered automatically when 
the server gets stuck that would help diagnose this ongoing problem, it 
would be easy enough to gather it too.


On 7/2/12 6:12 PM, Daniel Lewart wrote:
> Stuart, et al,
>
> On Sun, Jul 1, 2012 at 11:43 AM, Stuart Levy <salevy at illinois.edu> wrote:
>
>> The web server for http://lists.chambana.net/ is wedged again.  This is
>> happening several nights a week - is the manual remedy just to do something
>> simple, like "apachectl restart" ??   If so, could periodic restarting be
>> stuffed in a cron job?
>>
>> One way to do it: use wget to probe the server, and only restart if it's not
>> answering
>> (exit status 4 => network error, in wget 1.12 or later):
>>
>>     wget --timeout=15 --tries=3 -O /dev/null  --quiet
>> http://lists.chambana.net/nonesuch
>>     if [ $? = 4 ]; then
>>          apachectl restart ; # or whatever
>>     fi
> First I think it is important to understand why it is hanging.
>
> Then, if your cron job is used, it should send email and perhaps log.
>
> Thanks!
> Dan
>