[imp] Performance after upgrade

Stephen A. Cochran Lists stephen.a.cochran.lists at cahir.net
Tue Oct 31 09:11:41 PST 2006


After ignoring this problem for a long time, I finally tracked it  
down. Currently running Horde 3.1.1 and IMP 4.1.1.

Apache processes are thrown into a 100% CPU loop when the IMAP server  
issues an alert:

[31-Oct-2006 09:43:06] PHP Notice:  (null)(): [ALERT] You have more  
than 100000k of mail in total.  Please discard some of it. in Unknown  
on line 0

Obviously not a lot of info to track it down with, but at least I  
finally found the cause. I didn't find anything in the CHANGES file  
that looked relevant, but I'm going to update and hope that fixes the  
problem as well.

Steve Cochran



On Apr 19, 2006, at 2:25 AM, Stephen A. Cochran Lists wrote:

> I just upgraded from Horde 3.0.x/IMP 4.0.3 to Horde 3.1.1/IMP 4.1.1.
>
> Before the upgrade, the CPU was mostly idle. After the upgrade I've  
> noticed fairly poor performance. I just logged on and apache was  
> using 100% of the CPU:
>
> PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> 9655 apache    25   0 41144  21m  16m R 13.3  2.1 148:33.29 httpd
> 10389 apache    25   0 42948  22m  16m R 13.3  2.2  75:13.65 httpd
> 9648 apache    25   0 43004  18m  14m R 12.3  1.8 510:46.03 httpd
> 9642 apache    25   0 43048  24m  16m R 11.0  2.4 128:10.01 httpd
> 9641 apache    25   0 43228  24m  16m R 10.3  2.4 181:06.34 httpd
> 9643 apache    25   0 44648  21m  16m R 10.0  2.1  85:47.41 httpd
> 9645 apache    25   0 41036  20m  15m R 10.0  2.1 179:49.97 httpd
> 9646 apache    25   0 44128  22m  16m R 10.0  2.2 120:10.17 httpd
> 9647 apache    25   0 42152  18m  15m R 10.0  1.9 160:27.47 httpd
> 9657 apache    25   0 41004  20m  15m R 10.0  2.0 183:30.14 httpd
> 9687 apache    25   0 44712  23m  16m R 10.0  2.3  74:18.54 httpd
> 9690 apache    25   0 44236  24m  16m R 10.0  2.4 101:57.41 httpd
> 10317 apache    25   0 44184  24m  16m R 10.0  2.4  74:20.90 httpd
> 10388 apache    25   0 42928  19m  16m R 10.0  1.9  72:14.31 httpd
> 10484 apache    25   0 42384  19m  15m R 10.0  1.9  74:01.57 httpd
> 10485 apache    25   0 42628  22m  14m R 10.0  2.3  70:57.77 httpd
> 10536 apache    25   0 40016  17m  14m R 10.0  1.7  72:42.74 httpd
> 10555 apache    25   0 39704  17m  12m R 10.0  1.7  71:49.96 httpd
> 10586 apache    25   0 39800  19m  14m R 10.0  2.0  63:12.42 httpd
>
>
> From the horde log, the only notable thing is lots of the following  
> lines, sometimes several a second, sometimes one every 1-5 seconds.  
> Hard to tell how high the load is on this server, but it's not very  
> high. Are the preferences called on login or constantly during a  
> session?
>
> Apr 19 02:11:17 HORDE [debug] [] SQL Query by Prefs_sql::retrieve 
> (): SELECT pref_scope, pref_name, pref_value FROM horde_prefs WHERE  
> pref_uid = ? AND (pref_scope = ? OR pref_scope = 'horde') ORDER BY  
> pref_scope [on line 132 of "/var/www/html/mail/horde/lib/Horde/ 
> Prefs/sql.php"]
>
> Restarting apache seems to have cleared the CPU log for now, but  
> any idea what could cause the httpd processes to all suck so much  
> CPU time?
>
> Steve Cochran



More information about the imp mailing list