A few things I have noticed after creating a performance monitor with a 60 
second cycle time.

After you move the devices to the new monitor, you will need to stop the 
zenperfsnmp process and restart it, so that zenperfsnmp will stop monitoring 
the moved devices. (You can restart all zenoss processes, but that seems to be 
overkill to me.)

When you start the new (fast) performance monitor, it will begin to monitor the 
devices you have configured. If zenperfsnmp has already created an RRD, then it 
will NOT store the results in the cycle time frame (it will only collect them 
in the new cycle time.) In other words, you will have more numbers to use in 
your average, but your graphs will not change. 

One way around this is to delete your old RRD files. When I did this, my CPU 1 
Min graph for my cicso router displays appropriate granularity, as do my Serial 
Inetrfaces. However, there seems to be an issue that arises with the 
measurement of FastEthernet devices on my Cisco router that are created with 
the new cycle time. They start recording GB of data, instead of MB of data. 
(You can view the internal cycle time of the rrd file by using "rrdtool info 
rrdfilename.rrd | grep step".) If I keep the rrd files created under the 
default zenperfsnmp process (with a 5 min cycle time), it works properly. I 
suspect that the problem is with the RPN statement on the ifInfOctets and 
ifOutOctets data sources.

------------------------
 James Roman




-------------------- m2f --------------------

Read this topic online here:
http://community.zenoss.com/forums/viewtopic.php?p=8664#8664

-------------------- m2f --------------------



_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users

Reply via email to