|
Jim,
NetView on windows gives you an indication of a backed up ping queue or a backed up SNMP queue. If he is seeing this netmon SNMP queue backed up, then it will directly impact demandpolls. A demandpoll is nothing other than additional SNMP requests to machines which gets put on that queue. Do those get put on the end of the queue, sounds like it.
It is possible that Windows is slow with /etc/hosts...but that it does not generate high CPU while reading it. That could be a Unix / Windows difference.
Another item that could be impacting you.
The other question...do you only have one interface for each host in your /etc/hosts or DNS? If so, were those interfaces in your seed file for discovery? If not, how do you know that netmon discovered each device through that specific interface? Netmon reads ARP tables and will talk forever afterwards (via SNMP) to the device with the very first interface found on that device. It will not change to another interface until that interface (or device) is removed from the database. So, you could have 10.1.1.1 for router1 in your DNS, but if 10.2.2.2 was the first interface discovered for router1 that would be the interface that NetView is using to send snmp requests. So you can have DNS entries that NetView doesn't use if NetView discovered a different interface first. Usually people that just want to discover network devices only will have a 'database' of their network and put those 'managment interfaces' in both their seed file and also in their DNS. Then you delete the device, and rediscover it via the seed file (or do a complete rediscovery).
Many people when they first start out do a number of complete network discoveries while they are getting their configuration correct.
Stephen Hochstetler shochste@us.ibm.com
International Technical Support Organization at IBM
Office - 512-838-6198 (t/l 678) FAX - 512-838-6931
http://www.redbooks.ibm.com
|