[horde] Sweating Bullets!! Help!

[eculp@encontacto.net]

Quoting Chris <grooveman at comcast.net>:

>
>
> Craig White wrote:
>
>> On Tue, 2006-03-28 at 19:31 -0500, Chris wrote:
>>
>>
>>> Hello everyone,
>>>
>>> I am in trouble here.  I have had horde up and running beautifully for
>>> about 3 months now.  Now, all-of-a-sudden, it has become extremely
>>> slow.  Every time a page needs to load, it takes like 6 or 7 minutes!
>>> This is bad news!
>>>
>>> I am running Openbsd 3.8.  I am using horde 3.0.9 with Imp 4.0.4, Ingo
>>> 1.0.2, Gollem 1.0.1(sql backend), Kronolith2.0.6, Turba 2.0.5 and 
>>> Nag 2.0.4.
>>>
>>> The machine is a Pentum IV 3.0ghz machine with 2 gigs of ram.  I have
>>> been watching Top for hours now, and the CPU never spikes above 5%.  The
>>> processes at the top are httpd and perl, and they are at 0%.
>>>
>>> The hard drive is running great.  fsck comes back with no errors, and
>>> throughput is smokin'.
>>>
>>> I have ruled-out imap, as I have configured a few mail clients, and it
>>> works very smoothly and quickly.  SMTP is likewise functioning normally.
>>>
>>> The problem only occurs when hitting it from the web interface.  I have
>>> tried this on a machine directly next to this one, plugged into the same
>>> switch and machines across the Internet -- no difference in
>>> performance.  All other servers on the same segment are performing as
>>> normal.  Not a network issue.
>>>
>>> I had a small db error in my horde.log, but I restored my db, the error
>>> is gone, but the problems persist.
>>>
>>> This all started today without warning.  No one has touched this box in
>>> over a month.  All log files I can think of look nominal.  No
>>> significant errors.  I cannot find anything that could account for this!
>>>
>>> I have rebooted, restarted apache a million times, restarted the db,
>>> deleted and recreated the unix socket, cleared the tmp directory and
>>> restored the database.  Nothing seems to make a difference.
>>>
>>> If I restart apache, I get about 20 seconds of normal behaviour on the
>>> web client, after that, it suddenly degrades to 6-7 minutes lag per page.
>>>
>>> HELP!  I'm totally at a loss!
>>>
>>>
>> ----
>> big log files...they can really slow things down.
>>
>> check the syslogs, apache logs, horde.log...
>>
>> serving other web pages or db's out of that server? Check the speed on
>> them.
>>
>> Craig
>>
>>
>>
> I have scrapped all the log files in /var/www/tmp, /var/log and
> /var/www/logs.  I have run mysqlcheck with the --auto-repair switch as
> well.  No improvement.
>
> There are no other databases on this server other than the horde
> database.  No other websites than the horde interface.
>
> Anything else you can think of here?

Why don't you run top in a term window and watch it while you are 
waiting your 6 or 7 minutes and see what is eating your cpu cicles.  
Apache will do this on ocasion.

ed