Limits to DS, graphs?

[vtmgr]

Not sure how this started, but a week or so ago, some of my graphs stopped updating. Drive got flaky, one thing led to another, and I ended up updating from 0.8.6b to 0.8.6c, php4 to 5...

The problem is that some of my graphs are updating, and some not. It seems to be the older hosts (0-49 of 57), and older graphs. (new graphs in old hosts DNW, either)

Are there limits to how many graphs, DS, etc. are allowed. I have about 850graphs, and 1150DSs.

Platform is Win2k3, IIS6, PHP5, MySQL 4.0.20.

Running Cactid 0.8.6c.

No errors reported in the log, just no data.

All subsystems seem to be fine, MySQL knows about the hosts...

Anyone else seen this? more info can be provided.

Thanks in advance.

[TheWitness]

1) Run cmd.php / cactid from command line 1 time with logging set to debug
2) Do the same with poller.php
3) Post the STDOUT from each run.

TheWitness

[vtmgr]

Attached are outputs of "cactid > cactid.out" and "php poller.php > php.out"

Debug level logging.

Thanks for your help, and all your contributions to such a great program.

[TheWitness]

You are correct, both your polling and your rrd output is clean. You next need to perform a mysql dump and e-mail it to me. I will get you an answer once I have the dump. My e-mail is listed at the bottom of this post.

TheWitness

[willie]

Hey,

Just to let you know. I run a very similar setup, IIS6, W2K3,PHP 4,mySQL 4.0.20.
I am running 1027 hosts, 3680 graphs, and I have a poller cache of 7405. I had a problem at one time but it was a corrupted rrd file. I restored my rra directory from the prior day's backup and everything went back to normal. what I learned was not to copy the rra directory while the poller was running <g>. BTW, with cactid, I am running through my hosts is 120 secs.

Regards,

Willie

[TheWitness]

GroceryNetworkGuru,

Per my e-mail, I suspect that your inordinately high snmp timeout is problematic. Try a lower timeout and a higher retry count and see how it goes.

TheWitness

[vtmgr]

Shortening the timeouts helped some.

I also changed some parameters in the my.ini, and php.ini.

Of course I didn't keep track of what they were, but I bumped the key_buffer and query_cache_size variables(my.ini) to 64M, and these variables in the php.ini:

max_execution_time = 120 ; Maximum execution time of each script, in seconds
max_input_time = 120 ; Maximum amount of time each script may spend parsing request data
memory_limit = 20M ; Maximum amount of memory a script may consume (8MB)

The clue was when I queried poller_item, it only returned about 2/3 of my hosts. So I went to see what limits I could extend.

Of course my problem now is with zombies, but that is the subject of a different post...

Thx again for help.

[TheWitness]

These settings are not necessarily good:

max_execution_time = 120 ; Maximum execution time of each script, in seconds
max_input_time = 120 ; Maximum amount of time each script may spend parsing request data
memory_limit = 20M ; Maximum amount of memory a script may consume (8MB)

Max_input_time -> What is the default. I am not too sure the behavior of this.
Max_execution_time -> Cacti is usually 296 seconds, but it is set in the scripts (or so I thought). Looks like script server may be a problem with a timeout of 120 in your configureation
Memory_limit -> With todays technology you should shoot for more like 64M.

Change the "Max_Execution_Time" and "Max_Input_Time" to 300 and see if your Zombies go away. This is likely script server related due to your settings.

TheWitness

[TheWitness]

I did some testing and the php.ini settings are Ok. Therefore, I would have to see specific information on failures from your log file.

TheWitness