Some problems - help needed - stuck and unhappy

[fredpt]

Hello fellow cactiers!

Let me begin by a huge thanks for all involved in this comunity so that others are able to implement such a wonderfull software.
I have been following this project for some time now and always tried to read, read and read so that i could solve my problems. I have done that until now but i am now facing a wall where some issues i can't overcome.
I ask if you guys can spare some of your precious time to help me solve the issues i face.

My setup is:
VM in esxi5.5 running CactiEZ latest version.
Cacti NPC plugin 2.0.4 / ndoutils 1.7b
Nagios 3.4

After many weeks trying and failling i have managed to find a working configuration for most of the stuff but also some other issues i haven't been able to tackle.
The problem that i would like to start with is:
- poor performance from NPC page in cacti. Takes a long time to refresh and sometimes it just doesn't show info and i have to reload page. I have already ran those sql commands to add the ***long records in database;
- in npc page i can't see any host in host list but i can see the hosts in nagios and in database (npc_host);
- on the syslog i have an entry saying that the poller is running too often.

I have been learning Linux as i go along with cacti so please bare with my lack of experience but i also have a huge commitment.

Help!

[BSOD2600]

Sounds like your main problems are with the NPC plugin and thus, you're going to need to seek direct support from its author, as its not a official Cacti plugin.

regarding poller running too often, whats the exact message?
Do you have multiple poller entries in your cron per chance?

[fredpt]

Hello and thanks for reply.
Been having a LOT of work and my mind was kept elsewhere

For what i can see by cron logs (cause editing /etc/crontab) shows me an empty file (WTF?!?), the poller.php is running each minute (as configured in cacti). The cron entry is 5 minutes (i believe these are default values).
Funny is that today when i was checking logs i don't see the entry anymore. And i don't remember changing anything.

Anyway thanks for the kind reply. Regarding NPC i was hoping that someone had the same kind of issue. In my case i was able to improve the speed a lot by making sql accept 10k connections

My best
Fred

[BSOD2600]

Empty cron? uhh well since the poller is running, you need to find where/how this is being invoked on your system

Also, you do realize you cannot simply change the poller to run every 1 min (from the default 5), without performing additional configuration steps with regards to RRA definitions, template adjustments, etc. have you done this already?

Are you using Spine and Boost plugin to help the installation scale? There is a thread somewhere on here with some pretty impressive stats on how large Cacti installations can become...

[cigamit]

Don't edit that file, run this command.

Code: Select all

crontab -l

[fredpt]

Ok just learned that one about listing crontab

Result:
*/1 * * * * php /var/www/html/poller.php > /dev/null 2>&1
0 1 * * * nice -n 15 /var/www/backup.sh
0 2 * * * echo 3 > /proc/sys/vm/drop_caches

Regarding the poller, i use spine and haven't used boot plugin yet.
Just installed boost and going to read any documentation available.
In the timming i was under the impression that defaults were 1 and 5 minutes. Was it 5 and 5?
Sorry i didn't realize i had to make changes when timming of poller was altered.
Guess i focused too much on getting the system running and missed important info.

========================================================================

My present setup is monitoring our office. We have about 10 esxi 5.5/5.0 running about 30 vm's total, 3 w2k8 servers and several windows 2k3 servers.
Some linux machines like untangle, blueonyx and alike.

Everything is working nice and stable. Sometimes some windows server stops reporting data and i have to restart the snmp service on that server and stuff flows again.
The npc plugin is working fast and nice now, just would like to figure out why the hosts tab in npc doesnt show info even that the sql table has it.

Cigamit: thanks a lot for cactiez! This made possible for people like me to implement a live monitoring system that i'm proud of and own it to you all that develop and maintain this wonder.

[fredpt]

Ok now it's back....

08/15/2014 12:53:01 AM - POLLER: Poller[0] NOTE: Cron is configured to run too often! The Poller Interval is '60' seconds, with a minimum Cron period of '60' seconds, but only 60 seconds have passed since the poller last ran.
08/15/2014 12:53:01 AM - POLLER: Poller[0] NOTE: Poller Int: '60', Cron Int: '300', Time Since Last: '60', Max Runtime '298', Poller Runs: '5'

[cigamit]

Poller Int: '60', Cron Int: '300'

Just to double check, what is your Poller and Cron Interval set to under Settings?

[fredpt]

Well, now i realize what hapened lol
The esxi that was hosting cacti vm had a hardware failure and when back up the vm didnt go up.
After many tries i went for a snapshot and that brought the old poller 1 minute setting...
<--- FAIL!

Sorry for wasting your time.

[fredpt]

Now i get confused.

I installed a new server with cactiez cd.
Updated os and software to latest versions.
When i look @ pooler config i see a 1 minute setup for poller and cron.
I didn't change it so i assume is correct settings, then i get this in log:

08/20/2014 06:49:02 AM - POLLER: Poller[0] WARNING: Cron is out of sync with the Poller Interval! The Poller Interval is '60' seconds, with a maximum of a '300' second Cron, but 121 seconds have passed since the last poll!

Crontab:
*/1 * * * * php /var/www/html/poller.php > /dev/null 2>&1
0 1 * * * nice -n 15 /var/www/backup.sh
0 2 * * * echo 3 > /proc/sys/vm/drop_caches

[HBomb]

Old topic but still.. This just means your poller couldnt poll all the devices in time probably updating the RRDs took too long, consider using the Boost plugin in that case.

Also remove this drop_caches command from the crontab this is most harmful I can't believe they've left that in to be honest. What will happen at 2AM is you'll lose all the cached metadata about your (probably hundreds by now) RRDs and every update will require actual disk read and writes crippling your performance. Took me ages to realise why I had gaps at 2AM while all my backups were later (as I use the other cron file for everything other than the poller). Can you imagine what this does to a cacti with 30 000 .rrds? Disk IO got reduced to 100s of kilobytes a second for the rrdtool process instead of the usual 10-20 MB/s
</rant>

Now i get confused.

I installed a new server with cactiez cd.
Updated os and software to latest versions.
When i look @ pooler config i see a 1 minute setup for poller and cron.
I didn't change it so i assume is correct settings, then i get this in log:

08/20/2014 06:49:02 AM - POLLER: Poller[0] WARNING: Cron is out of sync with the Poller Interval! The Poller Interval is '60' seconds, with a maximum of a '300' second Cron, but 121 seconds have passed since the last poll!

Crontab:
*/1 * * * * php /var/www/html/poller.php > /dev/null 2>&1
0 1 * * * nice -n 15 /var/www/backup.sh
0 2 * * * echo 3 > /proc/sys/vm/drop_caches