'rescan' EVA8000 hardware bus ? can't see any enclosures any more

[jason andrade]

Hi,

Due to some unfortunate disk shuffling, our EVA 8000 no longer sees any hardware other than the enclosure with the controllers.

All vdisks seem to be fine but we can't see any enclosures.

Anyone got tips on options for some sort of rescan ?

The EVA is cabled as a 2C18D unit in 2 racks.

The 2nd rack has 6 shelves and 12 disks in the 2 shelves (all ungrouped).

Ideally i'd love to avoid an actual power down - though is there a quick reboot option that would avoid downtime ?

Or better still, a way to get the eva to rescan/reset the bus without an outage.

Note, no cabling or power has been changed.

[AnthonySN]

could attach a screen shot ?
what is your CV version ?
any reason why the disks were shuffled?
did you try rebooting your management server?
are you able to see the number on the EMU of the disk enclosures?

[jason andrade]

I can't easily attach a screen shot but.

1) we have two cabinets (rack1/rack2)
2) i was ungrouping drives in rack2 with the idea of moving them to another array
3) i added 12 146G disks into the shelves (so there would be no empty shelves when i pulled other disks out)
4) removed all the drives in the shelves (they were 450G drives) i was migrating out
5) thud. the entire array doesn't see any more drives and has put them all into the 'unmappable hardware' folder

[jason andrade]

other answers:

1) I've rebooted the CV mgmt server, no luck
2) CV 9.2 (shortly to become 9.3)

This is what i see:

Unmappable Hardware Folder Properties
Name: Unmappable Hardware
Unmappable devices: 148

I can't see any emus or any hardware in fact except the actual controllers.

please note that in terms of 'shuffling' - disks were removed from the array (all ungrouped) and new disks (also previously ungrouped) were added.

no actual disks were 'moved' within the same array.

[Mark...]

Hi
Reading through your note, assuming I have read it correctly, I think you have four shelves with NO disks in - correct ??
You should have at least 1 disk in every shelf. If you do not then you should change your config so that you do have at least one disk in every shelf.
Mark...

[jason andrade]

hi mark,

nope, not reading it correctly.

i was aware that the controller needs drives in the shelf so it can communicate with the shelf, so the order in which i did this was:

6 shelves in rack2 which had 1-3 disks (450G) in each shelf, 12 in total.

added 146G disk to shelf
added 2nd 146G disk to shelf
removed (already ungrouped) 450G disks from shelf

the problem seems to be it is not enough to have just have drives in a shelf - they have to be grouped? or grouped and then ungrouped..

whatever, the end result is all drives (and shelves) are now unmappable and basically the EMU bus seems to be annoyed.

any ways to reset this ? (without an outage..)

[gregersenj]

Does the EMUs got their numbers correctly?

If not you got a bad CAB bus.

And that will require down time to fix.

BR
/jag

[jason andrade]

All EMUs have correct numbers.

I had a HP engineer have a look at it today and he's logged a call for me with HP to investigate and suggest solutions.

We have a maintenance window coming up in a couple of days so i'm hoping to get some sort of solution in place to be able to use the window to fix this.

The array is apparently performing ok..

[gregersenj · Accepted solution]

Then the bus is ok.

Check if you really got connection to thye EVA.
Do you have all path to the EVA in Device manager.
If not rescan.

If path ok and all EMU is ok.
Reboot the master controller, in a service window.

BR
/jag

[jason andrade]

i'll plan for a boot. in the meantime, what exactly am i looking for in terms of 'can commandview really see all paths' to the eva ?

if i can see the (both) controllers but no shelves, i have difficulty figuring out how commandview is not seeing the EVA itself properly.

regards,

-jason