[svlug] REPOST: Apache comes to a crawl...at decreasing intervals

[Deirdre Saoirse]

On Mon, 17 Jul 2000, Jasmine Sante wrote:

> I'm running Apache version 1.3.9 on red hat linux version 6.0 (kernal
> 2.29), with no special configuration that I'm aware of.  The site has
> periodically been unreachable, beginning about a month ago.

When you say "unreachable" do you mean not responding to ping? Or just not
responding to an http request? What kind of volume of http requests are
you getting per day?

When this happens, what's the load average of the machine? Are any
partitions full? [answered, nevermind] Is the network choked?

Are you running any other processes that are http-related (specifically
Zope, etc.?)?

I'd suggest putting something on it to monitor it like big brother.

> The first time the site became unreachable, I rebooted the server and
> restarted Apache and all was well for about 2 weeks.  The second time,
> only a week had passed; the third, only a few days had passed.  Most
> recently, the site was unreachable on Thursday, then on Saturday then
> this morning (Monday).  I find that "ps" then shows no httpd processes
> running.  I've examined the obvious logfiles in /var/log, but see no
> obvious cause.
> 
> When I simply restart Apache, my sites are still not accessible *but*
> if I restart the server and then restart Apache, everything comes up
> golden (for a while anyway).
> 
> The only significant change made to Apache right before all this
> started happening was installing Chili!ASP (I shutdown the chili!asp
> server on Sat and then still experienced a slowdown on Mon).

Obvious solution is uninstalling or disabling chiliasp and seeing if that
changes it.

As a guess, based on a similar problem with Zope: it is spawning threads
and not killing them and then the web server stops responding when the max
# of threads is reached.

>              total       used       free     shared    buffers     cached
> Mem:        127792      61176      66616      85544       5616      27252
> -/+ buffers/cache:      28308      99484
> Swap:       254000          0     254000

Is this *after* the problem but *before* the reboot?

-- 
_Deirdre   *   http://www.sfknit.org   *   http://www.deirdre.net
"Trust me, I'm a science fiction writer" -- Larry Niven @ Conolulu