Cacti on DEBIAN - DEBIAN freeze after 5 or 6 days.

[zoummuoz]

Hi,
Apologies for my english...
I'm new cacti user and i start to thanks a lot the developpers projet for this full fonctionnal and large tool.

I've install cacti last version on DEBIAN 4.
I've 120 computers, 20 servers, 10 switch, 5 routers too manage.

All works good during 3-6 days ans then the system start to freeze. Sometime i can log in ssh to reboot, other time not. The apache server seems two work, mysql too, but there is not record in mysql ...
I've change some poller settings to see if it's better :
cron : 5 min
Maximum Concurrent Poller Processe : 50
Maximum Threads per Process : 20
Number of PHP Script Servers : 5
Script and Script Server Timeout Value : 25
The Maximum SNMP OID's Per SNMP Get Request : 10

I've Egroupware working on this machine (for testing).
The CPU seem to bee good and the memory too.

Can you propose me somes experiences infos ?.

Thank's for all.

[sneakymonkey]

Check your CACTI.log file size. if its over 2Gbytes then you will have problems. \www\htdocs\cacti\log

Code: Select all

> cacti.log 

should truncate it to nothing. hope this helps!

[sneakymonkey]

also, check your logging. if its left in DEBUG mode expect your cacti.log file to grow. change the setting to errors only or something under Settings.

[zoummuoz]

On all poller query i've this log.
My file is 2Mo. I've clean it.

09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[16] DS[862] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[16] DS[859] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[16] DS[858] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[16] DS[861] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[16] DS[860] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[18] DS[1072] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[18] DS[1044] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:05 PM - CMDPHP: Poller[0] Host[18] DS[1044] WARNING: Result from SNMP not valid. Partial Result:
09/29/2008 03:25:06 PM - SYSTEM STATS: Time:5.2560 Method:cmd.php Processes:20 Threads:N/A Hosts:22 HostsPerProcess:2 DataSources:372 RRDsProcessed:199

Many thanks for your answers.

[sneakymonkey]

Try the following;

http://forums.cacti.net/viewtopic.php?t ... c&start=15

Code: Select all

mysql -u root -p cactidb

then

Code: Select all

select count(*) from poller_output;

if theres a large amount then truncate the table.

Code: Select all

truncate table poller_output; 

see if that helps. it did for me!

[Exo7]

Maximum Concurrent Poller Processe : 50
Maximum Threads per Process : 20

that's quite high value
try Poller Processess = 2 * number of cpu
threads per Process = doesn't matter if you don't use cactid (spine), but 2-4 are quite common.

[zoummuoz]

It stay online during 10 days and it die. I reboot it etc ...
Joint : screen capture of my graph.
Joint : screen capture of my poller config
joint : my.cnf

thanks for all
Merci pour votre intérêt et votre temps.