ALUA follow_over HP EVA 8400 and ESXi5

[ascheale]

Hi,

we currently have an issue with an EVA 8400 and ESXi5 configured with VMW_SATP_ALUA and RR.

After the first scan of the FC Controllers the optimal paths are discovered - this is OK.

But, when the SAN admin changes the owning controller the esxi host does not automatically switch over to the new optimal paths.

So all IO is going through the mirror cache of the EVA.

Only after a rescan esxi finds the new optimal paths ? Is this how it should be ?

My understanding of SCSI-3 is that the host MUST switchover... even a simple linux system does so.

Please give me some help about this issue

THX

Alex

[Josh26 · Accepted solution]

Alex wrote:

Hi,

Only after a rescan esxi finds the new optimal paths ? Is this how it should be ?

THX

Alex

ESXi does rescan automatically periodically.

If a rescan fixes the issue, you will probably find that sitting back for a while and doing nothing will also fix it.

[ascheale]

OK,

With ALUA you have more active paths than those that are used. At the moment the controller switches the paths there are no dead paths as with an active/standby storage. So now the ESXi goes through the non optima paths.

When does ESXi does its rescan ? Can you define the time ?

Alex