IRF and FCoE

[Convergo]

Hi,

we are building an IRF with two switches (5900CP). When rebooting the subordinate all traffic (LAN and FCoE) is still available. We test this with a network file copy to a FCoE LUN :)

However when we reboot the master member within the IRF only (LAN) traffic is failing over, and there is a disruption in FCoE traffic resulting in breaking the connection to FCoE  LUNs. After rebooting the master the LUN is connected after 3 minutes but only with two paths. For the other two paths we have to disable and enable the ports to revive the 4 paths again.

We configured MAD BFD, and using an Emulex CNA adapter with 2 ports. Also read someting in the release notes (http://h20565.www2.hpe.com/hpsc/doc/public/display?sp4ts.oid=7283310&docLocale=en_US&docId=emr_na-c02018537) regarding enabling PXE on all NIC ports in order to detect LUNS.

Is this expected behaviour or should LAN & FCoE survive a master reboot within IRF ?

Help is appreciated !

[VoIP-Buddy]

Convergo,

We need more information.  How is your IRF configured?

Thanks!

David

[Convergo]

Hi,

IRF is not a suitable solution withen having 2 switches. You should opt for 4 switches if redundancy and NSPOF is a requirement.

http://h20565.www2.hpe.com/hpsc/doc/public/display?sp4ts.oid=5221896&docLocale=en_US&docId=emr_na-c04216045

This document describes the scenario's from page 14.

We will now leave the IRF concept, and install two switches connected to each other based on the QSFP+ links formerly used for IRF. For FCoE it is then possible to create two domains and seperate the VSANs.

So definately there are some pro's and con's using two switches in IRF when using FCoE and LAN together. Next question is what are the con's not using IRF :)