I've observed something that I find interesting, and am wondering about,
I have TruCluster 1.5 on a version 4.0e system. There are three network
interfaces on each member. There is a disk service configured to handle
Oracle, a Webserver, and a Virtual IP address on the outside interface.
As part of the action scripts for failover, there is another script run that
does an 'ifconfig alias' on another interface that is connected to an
internal LAN. This is on tu2.
I ran Performance Manager yesterday and saw that tu2 had a pattern of
Ethernet collisions happening on the member with the diskservice. When I
ran a tool to put a load on the webserver the collision rate went up to 400
per second. This appeared to be a bad card, but when I failed over to the
second member, the collisions stopped on tu2, but then appears on tu2 of the
other member. The problem seems to go with the member running the service.
We put a sniffer on the network and ran tests but it did not show up as
causing problem in the network.
Is this normal?
Dana Huggard
Systems Support Programmer
CUE Datawest Ltd.
Phone: (604) 734-7494 ext. 3138
Email: dhuggard_at_cuedatawest.com
www.cuedatawest.com
Received on Thu Mar 02 2000 - 17:47:36 NZDT