Can you send (to the list) your cron entry for
the observium poller please?
I had a similar issue that turned out to be caused by a
single device taking way too long to complete the poll
(lots of interfaces - not an issue with the poller itself)
and changed my poller-wrapper.py settings to the
following:
0,5,10,15,20,25,30,35,40,45,50,55 * * * * root
/opt/observium/poller-wrapper.py 32 >> /dev/null
2>&1
Check your config and see if you only have '1' after
'poller-wrapper.py' - it may be worth adjusting as per the
notes on
http://www.observium.org/wiki/Performance_tuning.
Adam
________________________________________
From: observium [
observium-bounces@observium.org]
on behalf of ecaroh [
lists@datenarbeiter.de]
Sent: Thursday, 11 December 2014 5:41 AM
To: Observium Network Observation System
Subject: Re: [Observium] gaps in graphs
Hi,
no, unfortunately not. I can't explain it with disk io.
There are to less machines monitored and the vz/kvm host
is not overloaded. The only thing i can suspect is, that
observium is running as a vz container and this causes
problems.
Before anybody moans, I know that firewalls and monitoring
system should stand alone. But cost for power and
administrational overhead would be to much for my intended
monitoring.
From other monitoring software i know that this can happen
if the collector needs to long for a polling cycle. But i
must admit, that i read about "multi" polling in
observium. I have to check whether this is active. Another
solution would be, to set up a physical test server.
I did not checked in deep for logs and other stuff,
because my main job actually is CEO of a business center
and not it staff as a couple of years before. But i will
do it when i have some free time.
At the moment i have " Total time for all devices:
59.73s" for polling. This result does not point to
polling problems.
If i find the solution, i will not keep it for me and post
it here.
--
ecaroh
PS: I run smokeping on the same machine with integration
in observium. There are no gaps.
Am 10.12.2014 um 14:28 schrieb Ole Hansen <
oha@netic.dk>:
DId you find a solution
for this? I'm seeing the same problem - and I'm
collecting snmp from the same devices from another
server (Zenoss distributed collector) on the same
network with no drops in the graphs.
--
Ole Hansen
On 01/12/2014, at
15.33, ecaroh <lists@datenarbeiter.de>
wrote:
Hello,
i got sporadic dropouts over all my graphs. If this
happen most of my graphs have dropouts.
The poller runs not longer than a minute.
`Total time for all devices: 41.07s 35.17s'
Memory usage 200 of 1000 MB RAM. All devices are in
the same physical gigabit network. Ping is less than
50 ms to each device. Monitored are only 16 devices.
Observium itself runs as a openvz container
(turnkeylinux).
I searched the web but does not find anything that
matches my setup. Does anybody have a hint for me?
Thanks,
ecaroh
<observium_gaps-in-graph.png>
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
--
HTH
ecaroh
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium