10-12-2009 08:26 AM
We recently licensed our existing SAN switches (5100s) to act as Integrated Routers - the switches are part of an existing backbone fabric, connected as edges to 48000 directors . FOS is 6.2.0e
The various unix hosts attached to these switches (dual fabric) suffered failure when only one fabric 5100 had an "fcrdisable" command run - the paths were checked prior to starting work and found to be working.
A veritas VCS cluster host appeared to suffer from a hung vxconfigd - and a reboot was required.
Why should running fcrdisable on one fabric cause so much collateral damage?
Has anyone else experienced this sort of thing when implementing FCR on an existing SAN switch such as a 5100?
10-12-2009 10:41 AM
I have found this info as below. Perhaps this will help you.
- Integrated Routing EX_Ports are only supported in the base switch on a switch with VF enabled.
- Integrated Routing and TopTalkers are not concurrently supported in FOS v6.2.
To use Integrated Routing, be sure to first disable TopTalkers prior to configuring EX_Ports.
10-13-2009 01:26 AM
Thanks for your response, as always.
I don't believe this is the issue as FCR is indeed now working correctly - ie we are able to see storage in our edge fabric from the backbone fabric (ie via the 5100 switch acting as an integrated router). viz:
BWDSW129:user> lsanzoneshow -s
Fabric ID: 101 Zone Name: LSAN_derka_D_A_EVA8400_Barn2_D
Fabric ID: 127 Zone Name: LSAN_derka_D_A_EVA8400_Barn2_D
The issue seems (to me) to be down to the fact that the 5100 was an existing SAN switch (ie not a standalone router such as a 7500) - but we configured FCR exactly as in the 6.2.x FOS admin guide - in other words it was the other F-port attached hosts on the 5100 that were badly impacted.
It was the transition to getting this working - and the fact that dual attached hosts were impacted in this way (lost paths which didn't recover) that is the issue.
Could it be an 8Gb issue?