Recompute virtual disk digest storm issue

[adunbrook]

In our environment a few weeks ago we started getting "recompute virtual disk digest" events vCenter every ~60sec and at first we just put a blackout window during business hours so the events are not spamming the event log.

Now we want to figure out if this is normal, I found a KB talking about this issue but none of the solutions/causes apply to us. VMware Knowledge Base

Our environment has:

- no view composer

- full clones

- none of the VM's are missing or have missing digest files.

- 2 pools of linux VDIs

These messages only happen when one pool (and not the others) have Storage Acceleration turned on. So far the solutions I've seen online are delete and rebuild the pool but that cannot be done as we have users dependent on this pool. Is this normal behavior or has a digest become dirty/corrupted?

[u777]

try disabling View Storage Accelerator

[adunbrook]

I have disabled it, but is this a proper solution? is acceleration even needed for our setup?

[u777]

the storm stopped? turn the accelerator back on, if the storm is back, then look for a desktop that is constantly updated.

[adunbrook]

The storm did stop and then when we turned the accelerator back on it returned. We have about 50 VDI what is the easiest way to find out which desktop is constantly being updated? because the task log only shows the pool name and not a specific VM, and the logs for Horizon don't show anything.

[pslaughter]

Have you tried looking at it through ESXTOP? You can see down the VM level. We had several issues with Network storage latency, I used the ESXTOP to pin down the what each LUN was doing and which VM's were causing the primary issue. The link below will help explain what everything is. Under ESXTOP if you press v you can see each of the VM's with read and write latency.

http://www.running-system.com/wp-content/uploads/2015/04/ESXTOP_vSphere6.pdf

[sjesse]

Whats the cache regeneration time set to, it should be on the same page you set the blackout on.

[adunbrook]

It is set to 7 days

[JoaquinSS]

Did you ever found the solution? We are also having same issue and even after disabling the option on view the storm of task still showing up on vcenter. Does the machines that were provisioned with this option enable need to be powered down for the new option to take effect ?