topic Re: EVA reports more allocated space than available space in Disk Enclosures

EVA reports more allocated space than available space

jason andrade — Wed, 27 Oct 2010 23:45:36 GMT

Hi,

A number of our EVAs report more allocated space than 'actual' space. Oddly this seems to only have happened after an upgrade to CV9.3 (i say oddly as i can't see how this can be related).

It doesn't matter if you have protection on None, Single or Double.

It doesn't matter if you add a disk (or indeed, where possible remove a disk from the disk group).

It doesn't 'fix' if you create vdisks or delete vdisks.

It doesn't 'fix' by letting levelling operations run over the DG.

An example or two:

EVA8000. XCS 6.22

Disk group has 8 * 300G disks. Protection level is Single. Actual Vdisks created on DG uses 211G.

Summary
Total: 1627 GB
Allocated: 1844 GB
(Alarm level exceeded)
Allocation level: Not available

Maximum Vdisk size
Vraid0: 1294 GB
Vraid1: 647 GB
Vraid5: 1035 GB

EVA8100.

Disk group has 8 * 300G disks. Protection level is Single. Actual Vdisks created on DG uses 1000G.

Summary
Total: 2201 GB
Allocated: 17179868654 GB
Allocation level: Not available

Maximum Vdisk size
Vraid0: 1026 GB
Vraid1: 513 GB
Vraid5: 821 GB

I've been told a controller resync (or fast reboot) may be required but i haven't yet found the available time to schedule this.

Anyone else had this issue and/or have advice/thoughts ?

regards,

-jason

Re: EVA reports more allocated space than available space

Rajiv Kulkarni — Thu, 28 Oct 2010 00:37:18 GMT

Hi Jason,

This is a classic issue on CV EVA and we often encounter this issue, this has been brought to notice to the Command view EVA team. As of now resync is the only option to get CV EVA to show the actual values. Please schedule a resync during low IO.

Thanks,
Rajiv

Re: EVA reports more allocated space than available space

jason andrade — Thu, 28 Oct 2010 01:49:28 GMT

Thanks Rajiv,

So following on from that, anymore care to take a stab at explaining the difference between a resync operation and a controller reboot (fast boot?)

I'm in the middle of an IT freeze and while i have lots of periods of low I/O, i have to jump a lot of hoops to get _anything_ that may affect services approved right now.

I've done a controller reboot to fix another issue (on another EVA) but i've not performed a resync operation myself (still new to EVA) and i'm trying to understand the implications.

e.g

More or less risk than a reboot ?
Faster or slower than a reboot ?
Do you have to resync each controller or just one 'resync' operation ?
Do you interrupt I/O to each controller ? For < ??s ?

regards,

-jason

Re: EVA reports more allocated space than available space

Rajiv Kulkarni — Thu, 28 Oct 2010 02:30:54 GMT

Hi Jason,

Controller resync and Controller fast boot are synonymous, one resync command would fast boot both controllers one by one. This is not data intrusive (provided Multipath is set accordingly) however the resync time depends on the total capacity used on the array. If the resync time exceeds 60 seconds then the hosts start complaining.

There is a formula to calculate this resync time and you can find it in the XCS upgrading guide.

Here's the link to the XCS 6.22 upgrade guide which talks about resync and the time taken to resync:

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c02104123/c02104123.pdf

Hope this helps.

Thanks & BR,

Rajiv