VMware Cloud Community
JayMiller29
Contributor
Contributor

Datastore free space sizes flapping in vCenter

I have a very odd issue myself or my co-workers have never seen before.  I've been working w/ VMware support on this for a couple months now, to no avail.  I'm curious if anyone in the community has ever run across this, because it's quite hard to get to the root of. 

Issue:  We have several storage DRS clusters in which the datastores will flap back & forth regarding their available free space.  The swings are quite large...in the terabytes.  This causes us VM deployment failures, and disk management (think disk expansion) task failures.  When we SSH into the hosts and look at the amount of free space a host sees for a datastore that is flapping, it shows what we (myself and VMware support) believe is the correct value, where vCenter sees a different value.  This behavior is observed on many hosts, it's not an isolated instance.  From everything I can tell, it's vCenter that is reporting the wrong values for free space in a good number (but not all) datastores. 

vCenter:  6.5 U2

hosts:  6.5 U2

Storage Array:  multiple vendors, not one particular vendor array

sDRS settings:  We've tweaked these settings every which way we can think of to try to determine the root cause, all to no avail.  We've even disabled sDRS, it did not make a difference.

Anyone ever see an issue like this?  We're quite stumped, but need to get to some sort of resolution.

Thanks in advance,

Jay

Reply
0 Kudos
2 Replies
a_p_
Leadership
Leadership

I've seen this a couple of times too. In all cases it occurred after increasing datastores.

It seems as if one (or more) of the hosts that have that datastore presented, do not properly update/rescan the datastores/HBAs, and so the reported size in vCenter switches back, and forth depending on the host from which vCenter last synced the datastore sizes. It also doesn't always help to rescan the datastores/HBAs not only from vCenter Server, but also from the hosts directly. The only reliable way that I found so far to get the sizes properly, was to login to each host, and run vmkfstools -V from the command line.

André

Reply
0 Kudos
JayMiller29
Contributor
Contributor

Hi André,

Thanks for the response.  I've never experienced the problem you mention, but I've read all about it while dealing w/ this issue.  I wish our issue were that simple, however, these datastores are their original size, never having been extended.  I'm still digging into trying to figure out the root cause, as this is super bizarre....however, I know that datastore extensions are not the root cause. 

Thanks again,


Jay

Reply
0 Kudos