I've been slowly converting my Zenoss 1.x systems to 2.0.4. Some of the 1.x
systems had been running with multiple local performance monitors created by
the trick I first learned from Alan Sanabria, wherein we make copies of the
conf, bin and ZenRRD files.
As a test of the performance of the 2.x systems, I moved everything back to
the single 'localhost' performance monitor before upgrading. At the moment I'm
trying to understand the difference in behavior between two systems with
similar hardware.
The well-behaving system is actually smaller:
i386-fedora-core5, GenuineIntel Intel(R) Xeon(TM) CPU 3.00GHz, 4xCPU,
3000MHz, 5057MB
It's got over 45,000 RRD files and it's only just now beginning to run out of
RAM and drop samples.
On the other hand, the bigger system is updating only about 11,000 RRD files,
but
is dropping samples:
x86_64-redhat-rhel4as, GenuineIntel Intel(R) Xeon(TM) CPU 3.00GHz, 16xCPU,
3002MHz, 9991MB
I'm wondering if my problem is that the "bigger" system is monitoring chaotic
user
labs where systems go up and down all the time, and I'm getting too many SNMP
errors?
2007-08-27 16:08:01 INFO zen.zenperfsnmp: Count 1314 good 1175 bad 135 time
61.058222
vs
2007-08-27 16:14:19 INFO zen.zenperfsnmp: Count 680 good 679 bad 1 time
9.046172
At any rate, my first attempt at creating multiple local monitors under 2.0.4
isn't
working out so well; the zenperfsnmp processes seem to be interfering with each
other somehow, and when I ran my "zenperfsnmp_01 run -v10" the log messages
went to
a file instead of my terminal.
I realize that most people probably won't operate with this many errors and so
probably
will never run into this problem, but it's causing me a lot of headaches :(
--
David Carmean Network Appliance, Inc
Infosystems Architect, 495 E. Java Drive
Java (Sunnyvale) Engineering Lab Services Sunnyvale, CA 94089
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
