VSAN testing with 6.7.0 version

[ManivelR]

Hi Guys,

I have some doubts.Pls help on this case.

VSAN testing with 6.7.0 version

We are currently doing testing on VSAN 6.7.0.

Our setup is as follows:-

3 ESXi servers with 6.7.0

7 test VMs are running including VCSA 6.7.0

VSAN version is 6.7.0

2* 2TB SSD disk/per ESXi host(each ESXi host has one disk group and 1* 2TB is going for cache tier and another 1* 2TB is going for capacity tier).

In storage policy:- FTT is set as 1 and selected RAID 1.

Whenever i try to test by taking maintenance of one ESXi host(enter maintenance),i see some warning message(35 objects will become non-compliant with storage policy),however the VM storage policy(compliance status) is mostly compliant. Why im receiving this warning message? please help.

Note:- Eventhough im getting some warning message,ESXi maintenance is getting successful.

Regards,

Manivel R

[depping]

there's a good post explaining all of this here:

vSAN Maintenance Mode - RAID-1 and RAID-5 using "Ensure Accessibility" - Virtual Blocks

The summary is: as you are only moving the components at risk, the data stored on the host that goes into maintenance mode is unavailable. This means that the objects that are comprised of these components are now no longer in compliance with the policy.

[ManivelR]

Thanks so much Depping for sharing the article.I will take a look very shortly.

Suppose if im using 4 ESXi host(VSAN cluster) with FTT-1/RAID 1,then there will not be any warning(non-compliant) with respect to storage policy.

Please correct me if im wrong.

Thanks,

Manivel R

[TheBobkin]

Hello Manivel,

"Suppose if im using 4 ESXi host(VSAN cluster) with FTT-1/RAID 1,then there will not be any warning(non-compliant) with respect to storage policy."

No, this will be the same regardless of how many hosts are in the cluster as you are placing it in Maintenance Mode with 'Ensure Accessibility' - when you are choosing this option you are essentially telling vSAN to run off a reduced number of replicas (reduced to 1 from 2 in your case) temporarily while you do maintenance on a host.

If you want retain FTT=1/Storage Policy Compliance of VMs during maintenance windows then you will a) have to use Full Data Evacuation option and b) have enough available nodes and space to do this (minimum 4 nodes for FTT=1, RAID1).

Bob

[ManivelR]

Thanks very much Bob for reply. I can understand the concept little bit and currently referring some articles.

vSAN Maintenance Mode - RAID-1 and RAID-5 using "Ensure Accessibility" - Virtual Blocks

Lets say i have 4 ESXi hosts with FTT-1/RAID 1

2* 2TB SSD disk/per ESXi host(each ESXi host has one disk group and 1* 2TB is going for cache tier and another 1* 2TB is going for capacity tier).

While conducting any maintenance (using Ensure accessibility option) on one ESXi host(lets say 3 hours),we can use"Ensure accessibility" option.In this case,VMs will continue to run without any impact from other ESXi hosts( i mean object replicas and witness component will continue to run from any other ESXi hosts)Is this right statement?

Thanks,

Manivel R

[TheBobkin · Accepted solution]

Hello Manivel,

"each ESXi host has one disk group and 1* 2TB is going for cache tier and another 1* 2TB is going for capacity tier"

On a side note:

1:1 Cache:Capacity ratio is very likely overkill - depending on your workload something nearer 600GB for the cache-tier devices will likely more suffice (and if it is a heavy and/or write-intensive workload you will want to be using multiple Disk-Groups per node with more than one Capacity-tier device per Disk-Group).

Designing vSAN Disk groups - All Flash Cache Ratio Update - Virtual Blocks

"While conducting any maintenance (using Ensure accessibility option) on one ESXi host(lets say 3 hours),we can use"Ensure accessibility" option."

When you place a host in MM with EA option, it doesn't immediately start rebuilding/moving the data but this will start the repair timer countdown - with default clom repair delay timer (60 minutes) it will start rebuilding after 1 hour, this can of course be changed to a higher value (needs to be applied to all hosts in the cluster):

VMware Knowledge Base

"In this case,VMs will continue to run without any impact from other ESXi hosts( i mean object replicas and witness component will continue to run from any other ESXi hosts)Is this right statement?"

The VMs will remain accessible, but you need to understand that until the other data-replicas come back (and resync the delta changes that occurred in their absence), the non-compliant Objects are actually FTT=0 and thus can be permanently lost from a single physical device (e.g. cache or capacity SSD), thus always ensure you have current backups when using MM with Ensure Accessibility.

Bob

[ManivelR]

Thanks Bob for the detailed explanation.

Regards,

Manivel R