load increased after upgrade from 0.8.7.g to 0.8.8.a

[enriquet]

I am having a significant load increase after upgrading. I am logging in apparently normal mode.
I dont know exactly what to look, but in "technical support" information have the following output:

Cacti Version 0.8.8a
Cacti OS unix
SNMP Version NET-SNMP version: 5.4.3
RRDTool Version RRDTool 1.4.x
Hosts 36
Graphs 1434
Data Sources Script/Command: 69
SNMP: 33
SNMP Query: 1336
Total: 1438
Poller Information
Interval 300
Type cmd.php
Items Action[0]: 2666
Action[1]: 69
Total: 2735
Concurrent Processes 1
Max Threads 1
PHP Servers 1
Script Timeout 25
Max OID 10
Last Run Statistics Time:22.1572 Method:cmd.php Processes:1 Threads:N/A Hosts:37 HostsPerProcess:37 DataSources:2735 RRDsProcessed:1418
PHP Information
PHP Version 5.3.3-7+squeeze13
PHP OS Linux
PHP uname Linux localhost 2.6.32-5-amd64 #1 SMP Sat May 5 01:12:59 UTC 2012 x86_64
PHP SNMP Installed
max_execution_time 30
memory_limit 128M

I have 30 devices, most of them switches, and 1434 graphs.

Any help or recomendation of what to look would be welcome

[paulgevers]

Start by following the debugging page

[enriquet]

Useful page, but mostly oriented to resolve issues regarding system operation. I didnt get useful tips there.
My system is working ok, all the graphs, and plugins are working, but load used to be around 1 when i had 8.7.g and now with cacti last version is around 9

[gandalf]

Which load value is this? A screenshot of the load graph might help. For 088a, we enhanced especially the reindexing stuff, which might increas this part of the processing depending on the reindexing option used for your Data Queries. When using default "Uptime Goes Backwards", no change should be encountered#R.

[enriquet]

I am attaching a pdf file with the graph load vs time.
These are the ones generated by template unix load, and from what i can see it uses a perl script (loadavg_multi.pl) which gets the values using uptime command.
I read something about indexing db to increase performance, but i am not a dbase expert, and also had the idea that last version already had improvements in this aspect, so i dont know what else i can look.

[gandalf]

I'm not aware of any dramatical change in cmd.php apart from the one described. Are you able to run spine as a comparison?
Did you make sure, that only a single poller is running?
R.

[enriquet]

I will try spine but i have to install it. I did not know too mucho about it, but it seems much better. I ll post the results after that.
In the meanwhile, i only see one cacti file in /etc/cron.d
*/5 * * * * www-data php /usr/share/cacti/site/poller.php >/dev/null 2>/var/log/cacti/poller-error.log

It seems to be the only place from where the poller is executed. I have checked everything under /etc/cron*
Is there any place else where i should search?

Another thing i did is increased maximum concurrent poller processes to 4 (i have 2 cores)
Seems to improve a little, what do you think about this setting?

[enriquet]

I installed spine. A little improvement but still load is much greater than before the upgrade.

[gandalf]

It seems to be the only place from where the poller is executed. I have checked everything under /etc/cron*
Is there any place else where i should search?

crontab -l -u <user> might show additional entries; see 2nd link of my sig
R.

[enriquet]

Thanks a lot for all the tips. I ll do that and if i find something post it here. Anyway i am running with spine now and the system, despite the load seems stable.