Slowdown when create Vm Disk

[schnouk]

Hello

I have yet many experiments in vmware, so please be forgiving

I have updated a V5 HA cluster a few months ago.

Since a few days, I have a slowdown during the creation of VM.

example: to add a 300 GB drive to a VM must be 3-4 hours

I checked the hardware switch servers, bay SAN, I have no visible error.

In the vcenter, in the event of the cluster, I have the below messages that appear when the spikes of activity.

Device naa.600000e00d0000000003378800000000

performance has deteriorated. I/O latency increased

from average value of 6816 microseconds to 511861

microseconds.

warning

Can you tell me is it y of the points to be checked on my Vmware installation?

My Config:

10 active VMs

2 x hosts Vmware 5.0.0 2 x XEON 5620 64 GB RAM

Fujitsu ethernus DX60 S2 (2 x Ctrl ISCSI)

2 x switch Giga cisco SG300

[Jayden56]

Hi

Welcome to the communities.

First thing we need to check  hard disk health status.

"You must not lose faith in  humanity".

[maishsk]

When you create the disk - are you creating it as thin/thick/eager-zeroed?

What is the latency you are seeing on the disks when you create the VM?

[schnouk]

Hi,

No, I checked all disks, and I have no error.

I nevertheless opened a ticket from the manufacturer for verification

[schnouk]

Hi,

When you create the disk eager-zeroed

latency you are seeing on the disks when you create the VM

I see messages below on the datastore on which I am the creation of VM disk

Device naa.600000e00d0000000003378800000000

performance has deteriorated. I/O latency increased

from average value of 6816 microseconds to 511861

microseconds.

warning

If there is a deletion of snapshot on an another datastore after a backup with VDR so I have the same message on datastore.

during the creation of the VM disk all is slowdown.

example: for view the HA node logs or Vm logs need to wait more than 5 minutes

I have no test to move VM via vmotion, because my system is in production.

I'll look tomorrow if my vcenter appliance is on the same datastore that I created

I'll still try to copy a large file on the VM or I creates the disk and see the time it take

I make a return tomorow

Thank you for your support

[a_p_]

Just a a few thoughts. I upgraded an environment with a DX80 to vSphere 5.1 some time ago and after a couple of calls, we got an updated firmware for the storage system. The DX80 - and I assume it's the same with the DX60 - did not support "Round-Robin" path policy, so make sure you read the documentation and ensure the proper path-policy is configured for the LUNs. With Eager-Zeroed disks, the disks will be filled with zeroes at creation time. If VAAI is not supported (or does not work), zeroing puts a high load on the iSCSI network, because the zeroing has to be done by the ESXi host, rather than being offloaded to the storage system.

André

[schnouk]

Thanks André,

I just had the fijitsu response.

I must Disable the VAAI options on the vSphere 5 host

disable delay ack

and update de BIOS to V20L14

[schnouk]

Thanks to Jayden56 and maishsk for your help