EVA 5000 disk position replacement

[strajanoski]

on our EVA 5000 we need to reorganize the disks in the diskshelfs. Is it possible due to the virtulaization simply to shut down the system, make the necesary disk reorganization (change the location of the disks in the bayes) and then just start the system again. As far as i know every disk has its own WWN, so its grouping information is not dependant on the disk slot position in the shelf anymore. All disks belong anyhow to the same fibrechanel loop.

[Uwe Zessin]

The EVA puts meta-data on the disk drives, so it can pick up the whole configuration even after the replacement of both controllers.
I've done it once on an EVA4000, but as far as I can tell, that this has always been the case.

> All disks belong anyhow to the same fibrechanel loop.

No, the EVA5000 has two dual-loops.

[strajanoski]

Yes this is true, but the second loop is for redundecy only

[Uwe Zessin · Accepted solution]

OK. In other words: a HSV110 controller as used in the EVA5000 has four (4) connectors for disk drive connectivity.
That makes two (2) dual(2) loops: 2*2=4.

Otherwise you could not put 240 disk drives at the back-end as there are only 126 AL_PAs for NL_Ports in an FC_AL.

[strajanoski]

Does this mean, that when the EVA boots, needs to check all the drives connected and find the configuration information stored somewhere. What I have read is that RSS takes care for the disk group information and EVA tries to organize the disk drives in so called RSS groups, (6 to 11 disks), but I don't know if there is some kind of master EVA configuration and where it is stored.
Anyhow swaping disk slot position shouldn't cause any troubles and is a safe operation. (All our data is stored on this EVA 5000)

[Mark...]

Hi,
The system configuration is a bit more complicated perhaps than you may think ? 3 levels of metadata are held.
Anyway, the first level of controller information is held on what are know as quorum disks. The controllers need to read this at power up to establish the name, wwid, etc of the controllers/configuration to start with.
After this they go onto the second level of metadata which holds disk group information bit map of data and so on.
The third level holds some event log information SMA/SMS specific info, licence info etc.

As you stated, it is possible to power everything, do a disk shuffle and as long as you maintain a VALID configuration then all should be OK. However the system may now not be optimal for maximum data recovery ref RSS's etc but this also depends on disk shelves and other things.

There are other posts that deal with this if you require information.

EVA only use 120 out of 126 possible id's. There is then an A & B side. Loose the A side, in total for whatever reason, then all the disks should still be available on th eB side. It is not just for failover.

Bottom line, consider what you are doing before you move disks around & understand the possible "issues".

Mark...