VMware Cloud Community
GTMcKaige
Contributor
Contributor
Jump to solution

Using SIOC with SRM 5

Recently I modified my datastores to use SIOC and during our monthly SRM fail-over all recovery plans failed. Testing has shown that in our environment having the SIOC box checked is causing the failures. The environment is as follows

ESXi 5 (Licensed properly for SIOC), Dell R910s, NetApp 3270 (HA pair on each end) , iSCSI sw intiator properly linked to pNICs x 2 on each ESXi host, round robing PSP which shows IO down all paths.....

In reading the requirements for SIOC I found

1) Datastores that are Storage I/O Control-enabled must be managed by a single vCenter Server system.
2) Storage I/O Control is supported on Fibre Channel-connected, iSCSI-connected, and NFS-connected
storage. Raw Device Mapping (RDM) is not supported.
3) Storage I/O Control does not support datastores with multiple extents.

Looks clean to me unless I'm misreading the single vCenter bit and this includes SRM DR sites, but its a different datastore on the other side...

I don't remember anything in the SRA documents or the SRM docs that would stop me from using SIOC, does anyone have any information on SIOC not being used in conjunction with SRM?

Tags (5)
0 Kudos
1 Solution
6 Replies
jeff9565
Enthusiast
Enthusiast
Jump to solution

Can you give the errors you are receiving?

0 Kudos
GTMcKaige
Contributor
Contributor
Jump to solution

Error - Cannot unmount datastore 'my_test_station' from host 'x.x.x.x'. The resource 'my_test_station' is in use. Cannot unmount datastore 'my_test_station' from host 'x.x.x.x'. The resource 'my_test_station' is in use.

Once SIOC is unchecked recovery works fine.

0 Kudos
mal_michael
Commander
Commander
Jump to solution

Hi,

SIOC cannot be used when doing a planned migration, because datastore with SIOC enable cannot be unmounted.

I believe it is a documentation bug if this is not documented.

Michael.

0 Kudos
GTMcKaige
Contributor
Contributor
Jump to solution

OK so after re-reading all the guides for (SRM, SIOC, NetApp SRA) and what you sent over it seems the docs aren't clear as to why Planned SRM recovery plans fail when SIOC is enabled, but I'm guesing the fact that SIOC writes to the disk every 4 seconds = busy disk for the unmount process.

Thanks for pointing me to the right resources.

Froggy

0 Kudos
locohost
Contributor
Contributor
Jump to solution

SRM tries to unmount the datastore on the protected side after it brings stuff up at the recovery site.  You cannot unmount a datastore with SIOC enabled.

It seems like SRM should be able to handle disabling SIOC first.

That said, you should be able to just manually disable storage io control on all the volumes in question prior to the planned migration.  As implied above, this should not effect a real disaster or forced takeover because SRM will not have to worry about cleaning up the protected side since your datacenter is on fire, was hit by an asteroid, underwater, or without power.

0 Kudos