An unmanaged I/O workload is detected on a SOIC-enabled datastore

[homerjnick]

I get the message in all three of my ESXi 6 servers that are Cisco Blades: "An unmanaged I/O workload is detected on a SOIC-enabled datastore"

I do not have any datastores enabled for Storage I/O Control...in fact my license will not allow it.

So why am I getting this in my events tab?

[vXav]

I get the same events.

Even if is disabled and not being used, it seems like SIOC is running anyway, it's not clear how but you're not alone in this.

Usually it's safe to ignore them.

A few things to check:

- Do you use a VM backup solution (Like Veeam)?

- Have you got a "dynamic" storage array (Like EqualLogic that moves things around, replicate ...)?

- Have you got more than one vCenter?

Check this KB : Unmanaged workload is detected on datastore running SIOC (1020651) | VMware KB

[homerjnick]

Ok...that is strange...have you logged a call with VMware about it?

I do use Veeam but I'm seeing these messages in the events tab on our ESXi servers during production time and outwith the backup window...Veeam is not running or doing anything when these are getting logged.

We use an IBM Storwize V5000 unit that has the "Easy Tier" feature licensed where blocks of data that have the heaviest IO are moved to SSD...and we replicate to another Storwize V5000 unit...but we have had our SAN in for a year now with no mention of these messages...they just now suddenly appear.

I guess Easy Tier/Replication may be the cause...but strange it would appear now.  I guess the timing of the message appearing in the ESXi servers coincides with a new firmware being applied to the V5000...so you may be right.

Still...I do not SOIC enabled so that is a bug.

[vXav · Accepted solution]

That would make sense indeed.

It's not clear what SIOC does in the background when disabled, there are a bunch of mentions online but I never found a detailed answer to it.

Here's a reddit thread by someone having the same question as you.

https://www.reddit.com/r/sysadmin/comments/4gok83/vmware_showing_errors_from_sioc_but_sioc_is_not/

The bit that caught my attention (even though he's got no proof to back that but ...)

I think it always shows these. Suspect they made it so SIOC is Always "enabled" to some degree, but possibly made the On/Off switch prevent SIOC from actually intervening during latency.

One thing's for sure though.... SIOC adds some useful performance charts, so I like having SIOC watching and reporting data, even if it is functionally disabled on the datastore by having the box unchecked, it still does (or seems to).