Hi Hrvoje (and others),
Today I will be looking into the load issue with zenmodeler. Watch this
ticket.
http://dev.zenoss.org/trac/ticket/754
-Eric
Hrvoje Habjanić wrote:
On Friday 29 December 2006 21:31, Erik A. Dahl wrote:
Here are somethings to try to optimize the modeler:
1. make sure you aren't trying to collect big routing tables (like
the entire internet table). If you are take out RouteMap from
zCollectorCollectPlugins in /Devices/Network/Router (if that's where
the router is stored)
Solaris have special problems whit this. Linux should be ok.
2. If you have linux or solaris boxes the software list can be very
long you might take that out by adding HRSWInstalled to
zCollectorIgnorePlugins
Yes, but, unfortunately, i need those.
3. Try switching to snmp v2 which is more efficient but sometimes
gives wacky results by setting zSnmpVer to v2
Tried that, got "wacky" results ... :-)
4. You might turn the daemon off by default and run it using cron
only once a day.
Hmm ....
As i see it, the problem is in tinysnmp. It seems he doesn't "select" for udp
packets, but instead loops in while loop!? And that's what is causing high
load ...
If i find some free time, i'll try to pinpoint real problem ... :-/
H.
-EAD
On Dec 28, 2006, at 3:14 AM, Hrvoje Habjanić wrote:
On Thursday 28 December 2006 04:20, Clayton Dillard wrote:
I'm running Zenoss on a dual core intel proc with 1GB ram and zenoss
eats into swap faster than you can say alakazam!
I also have zenoss running in vm from the vmware image that zenoss
provides and the performance on it is fine.
Every now and then I get errors in the zencommand log that talk about
losing connection to the mysql server.
Can anyone help me with this? Zenoss seems like a decent
alternative to
Sitescope and Nagios but the low reliability won't work for
production
monitoring at our company.
Similar problem here. I have 70+ devices, and i get occasionaly big
load. What
i can tell, the guilty one is zenmodeler. It just kills machine. I
got around
this by putting "parallel 2" in my zenmodeler.conf, but the load,
still, can
go to 30, and more. Good thing is that modeler is collecting logs
only twice
per day, and lasts 2 hours.
But, it is still the problem. Does anyone have suggestion for this?
H.
- Clay
Top output:
Tasks: 93 total, 1 running, 92 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.2%sy, 0.0%ni, 99.8%id, 0.0%wa, 0.0%hi, 0.0%
si,
0.0%st
Mem: 1023788k total, 1013336k used, 10452k free, 26772k
buffers
Swap: 2031608k total, 1128060k used, 903548k free, 235156k
cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+
COMMAND
2849 zenoss 25 0 1143m 361m 1360 S 0 36.2 3:10.03
python
2859 zenoss 15 0 323m 87m 1784 S 0 8.7 1:36.99
python
3037 zenoss 15 0 310m 32m 1508 S 0 3.3 0:28.19
python
2835 zenoss 15 0 413m 21m 1900 S 0 2.2 0:11.05
python2.4
3008 zenoss 15 0 279m 16m 1932 S 0 1.7 0:04.19
python
2913 root 15 0 277m 13m 1588 S 0 1.4 0:05.08
python
2985 zenoss 16 0 276m 12m 1536 S 0 1.2 0:08.09
python
2867 zenoss 15 0 279m 9732 1436 S 0 1.0 0:02.76
python
2876 zenoss 15 0 279m 7768 1352 S 0 0.8 0:02.26
python
2947 root 15 0 274m 7452 1444 S 0 0.7 0:00.75
python
2382 mysql 15 0 244m 7012 2824 S 0 0.7 0:04.43
mysqld
2840 zenoss 15 0 341m 6004 1404 S 0 0.6 0:01.13
python
3019 root 15 0 338m 5904 1404 S 0 0.6 0:01.18
python
2831 zenoss 15 0 155m 4048 1252 S 0 0.4 0:26.96
python2.4
2830 zenoss 15 0 142m 1148 760 S 0 0.1 0:00.00
python2.4
2834 zenoss 18 0 125m 1056 668 S 0 0.1 0:00.00
python2.4
2259 root 15 0 100m 944 692 S 0 0.1 0:00.17
snmpd
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
