VHDX Corruption with 3PAR 7200 - ODX?

[slink]

I have been getting frequent corruption on my Hyper-V VMs and it's really troubling me. So far I've been lucky and nothing too critical has been affected but occasionally I reboot a VM and I get this error:



Or I find a VM switched off and I get this scary error when I try and start it:



SCOM is also flagging corruption on some of the VMs. I have to restore from a snapshot and invariably run chkdsk.

I have a feeling it could be ODX as there have been reports of it causing corruption in the past but I have installed the hotfixes on all my Hyper-V hosts months ago but I am still seeing the corruption happening.

http://support.microsoft.com/kb/2870270/en-us
http://support.microsoft.com/kb/2796995

Has anybody else been experiencing this? I am considering turning ODX off, it seems very intermittent anyway, sometimes I can copy a file and it zooms across other times it appears to be copying at normal speed.

[DemonRoss]

Hi,

did you ever get to the bottom of this problem? I'm having very similar issues, which is obviously very worrying!

Any info you can pass on would be much appreciated.

Thanks.

[slink]

ODX. It's a non-feature. Causes problems with snapshot backups too. Disable it and forget it ever existed so you don't feel you're missing out. HP's implementation of it sucks. EMC has no such problems. Shame as I much prefer 3PAR for everything else.

Problems appeared to stem from Live Storage Migration. Either the VM would be corrupted or the CSV from which it was migrated from would start reporting corruption errors. I battled with it for a while, then I planned some downtime, turned off VMs on problem CSVs, did offline storage migrations to free up each CSV in turn and ran the CSV specific chkdsk commands on each one to make sure they were all OK (found and fixed some errors). Disabled ODX on each host and rebooted them. Powered up VMs and ran chkdsk on all of them to make sure they were OK.

Problems didn't reoccurr for some weeks so I figured I would try ODX again so re-enabled it on all hosts. Few days went by with no problems and then I got a corruption again so fixed that VM and disabled ODX again and I've left it like that ever since. Not worth the bother,

I'd been having issues with DPM backing up using 3PAR hardware VSS (doesn't work) and software VSS (took cluster offline) and after disabling ODX the software VSS backups started working properly as well.

There's various threads and posts dotted around on technet and elsewhere detailing problems, some by me, some by others and the solution for all has been to disable ODX.

[spencer.ryan]

I just saw this in the support portal (you may need to log in to see it)


SUPPORT COMMUNICATION -

Document ID: c04205854

Version: 1
CUSTOMER BULLETIN: HP 3PAR StoreServ Storage - HP 3PAR OS 3.1.2 can Improperly Zero Blocks Beyond the Requested Range When Using Windows Server 2012 ODX Commands



https://h20565.www2.hp.com/portal/site/ ... cachetoken

[apol]

There's a Patch for ODX related problems with 3.1.2:

http://flemmingriis.com/got-hp-3par-tim ... odx-blues/

Quote:
For 3.1.2 MU2 and 3.1.2 EMU2, apply Patch 11 followed by Patch 36.

InformOS 3.1.3 contains even more ODX-related fixes, but is a "controlled release" at the moment.