Need to rescan twice during the real failover

[jpraveleau]

Hi,

I currently try SRM and the hp SRA with two EVA4xxx.

The only problem that I have is concerning the need of make the rescan twice !! Effectively, I use qlogic hba, and the rescan need to be twice. SRM doesn't make two rescan and when I run a real failover, the luns in the recovery site are not see by the recovery esx servers. I need to rescan a second time and the script continue without errors. What is the solution ?

jp

[Michelle_Laveri]

A pal of mine is HP/VMware EVA guy who has been working with SRM (like me) since the beta...

I've emailed him and ask him - if he has seen this problem - and join in this thread

Regards

Mike

[Smoggy]

you should not need to rescan the HBA's twice to pick up device changes.

are you using the latest firmware on the qla cards?

if you change the state of the diskgroup/DR group in HP CA manually outside of SRM do you still need to rescan the HBA's twice to see any change? i don't have a EVA to test this on but this sounds odd. If the manual test also requires two rescans I would also engage HP to verify your setup.

what ESX build are you using here?

cheers

Lee Dilworth

[jpraveleau]

Hi,

- are you using the latest firmware on the qla cards?

Yes, I use the last release (QMH2462 with the 1.73af firmware - 16 jun 2008).

- if you change the state of the diskgroup/DR group in HP CA manually outside of SRM do you still need to rescan the HBA's twice to see any change? i don't have a EVA to test this on but this sounds odd. If the manual test also requires two rescans I would also engage HP to verify your setup.

Yes, I always had this problem. With and without SRM...

[AlistairS]

Hi

First, thanks to Mike for the email invite to this thread!

As Lee quite rightly points out, if two rescans are required when changing the DR group, it doesn't sound like SRM is root cause. It would be great to know a little more about your configuration. Can you share with us your ESX build numbers, EVA vdisk information in DR group and an insight to your FC zoning?

Thanks,

Alistair

[jpraveleau]

Hi,

It seems that the problem is due to the esx advanced parameters (of my secondery site) : Scsi.ScsiRescanAllHba that was set to 0 instead of 1.

I continue the eval.

jp

[Michelle_Laveri]

Hi,

It seems that the problem is due to the esx advanced parameters (of my secondery site) : Scsi.ScsiRescanAllHba that was set to 0 instead of 1.

I continue the eval.

jp

So did changing this setting fix your problem?

Was this regarded as an "error" in the HP SRA, or a "by design" issue. I notice the release notes suggest cases for adjusting this parameter...

Regards

Mike