VM with so much Snapshot's Vmdk Solution please ?

[water270]

Hello,

I cannot backup one vm cince a long time. I am in VMWare ESX 4.0

I just see that i have much snasphots due bakup failed. I am very afraid of that.

What good solution did you preconise me to delete and to consolidate with safety of course...

Thank you very much.

N.

[RyanH84]

Hi,

I would be careful here as the old consolidation process in VMware 4.0 can cause disk space growth issues, what version are you on exactly?  4.0 Update 2, ETC? What size is the datastore where the VM resides and how much free space is there left?

As they are only snapshots for backup (which has clearly not been removing them properly) I would check BackupExec to start with and see if there are any errors relating to cleanup.

After this, I'd go through the snapshots one at a time starting at the newest - and delete them one at a time to be safe. Some of those snapshots go back a LONG way so slow and steady is definitely a good idea.

[water270]

ESX version 4.0.0 Build 164009 is the version.

Free space is 1 T

[RyanH84]

Ok that is an old version of ESX! I'd definitely look at upgrading if at all possible in the future!

Looking at your snapshots, it started in early 2014 and then there was one at later 2014 (November time). Then there is a big jump to March 2015 and then another one in June. These snapshot delta disks are fairly large so they will take a long time to delete/consolidate.

Here is what I would do.

1) Stop the backups on this box temporarily to ensure nothing interfears with the VM while you are removing snapshots.

2) Remove the most recent snapshot first. (04-07-2015).

3) Repeat the deletions one at a time, moving up the chain.

4) As you get closer to the base, the delete might take a while. As all changes will be written into the disk parent above this and they are larger disks. Just let the process complete.

5) Once finished and there are no snapshots, enable your backup again and perhaps run a clean backup with no snapshots to ensure the system backups up and then removes the snapshot as it should do.

It is possible to use the "delete all" feature, but personally, I wouldn't do this. There will be a fair few writes to the parent snapshot disk as this happens from each child. Slow and steady wins the race!

If you have any questions, just let me know.

[water270]

Yes we know we have to upgrade...

We have 2 others host with 5.1 ESXi

Ok. Yes i think of that , but before i thought to do a full backup of the entire folder on another folder on the storage.

Because i don't want to have a crash of this VM it is server where SQL Server is working ? Yes he can arrive ? during  the deletion the last one after the last one ..

The users will be impacted or not ?

Before to process  i will do too  a backup manually, all that is important of the sql server before the deletion

[RyanH84]

Your plan of taking a manual backup first of your SQL is also a good idea.

It should not impact users or operation of the VM. It will cause some performance hit on your storage as the disk changes are going to be written into each parent snapshot as you go and delete/consolidate. It shouldn't be too severe, especially if you do one at a time.

[water270]

Did you think i have needed to do a full backup of the entire folder on another folder on the storage to have all the snapshots in backup if the task crash the server ?

What are my  % chances that vm can crash ?

[RyanH84]

Given your backup solution has caused this mess, I would perform a manual backup of all your SQL data using SQL Management studio (.bak files) and place them somewhere that is not located on that server. Then start removing snapshots.

If you take your time and remove the snapshots one at a time, there is very little chance your VM will crash. I wouldn't like to give you a figure.

If you are really 100% not confident/happy with this, I'd recommend raising a support case to VMware directly.

[water270]

Yes i am. I will.

Tks.

[a_p_]

Just a quick question. How much free disk space do you have on this (or other) datastores? A usually riskless option in a case like yours is to clone the virtual disk, rather than trying to delete the snapshots. This way the original virtual disk file won't get touched and you can always revert to the current state in case something doesn't work as expected. If this is an option for you, please let me know, and we will figure out a way how to proceed.

André

[water270]

Hello,

Yes i have free space , 1 Tera.. no problem for that..

I tried to copy all the vmdk in a new folder for backup before delete snapshot one by one but for the last i can't  paste for the last,  Option is grised ..

Any idea ?

[RyanH84]

Hi,

Copying all the snapshot VMDK's is quite a messy way of making a backup.

When you delete the VM snapshots, I hope you are using the GUI menu as you have shown in your picture ()

As A.P correctly pointed out, cloning the VM could be the best option at this stage.

The issue being that if it is a live SQL server, I'd suggest stopping services or powering off the box and then cloning to ensure a clean copy.

Is this a possibility for you? You have plenty of space so if you can arrange downtime a power off and clone could be the best option!

[water270]

Ok. How i can process because in don't see this option ?

In fact the clone will take the last snapshot as i see .

Yes it can be more easy .. but how ot process ? The clone can be the backup ? is th vm crash ?

[RyanH84]

Ok, following on from AP's advice, this is what I would do:

1) Power off your SQL Server in VMware.

2) Right click the VM and select "CLONE".

3) When prompted, give the VM a name and select the right inventory location.

4) Select the right cluster/hosts

5) Select the datastore (with 1TB space free)

6) Do not customise the VM and then let the clone run/complete.

7) At this stage you will have your new VM (clone) and the old VM (Original) powered off.

😎 Power on the clone and it will act as though it is the Original VM.

9) Check services and ensure that everything runs correctly.

10) After time, delete the Old Original VM.

This new clone, will have no snapshots and will be a copy of the server at the point in time that it is now! The clone process consolidates all the snapshot disks.

If the clone fails, you can stop the clone and just power up the original server again.

Does this make sense?

[water270]

Yes i understand all that and thanks i will prefer this option.

I have to power off the VM when the users don't work.. :smileygrin: perhaps this night.

How many times it will do for that do you think ?

[RyanH84]

Hi,

I think overnight is a good option, so users are not affected.

Looking at the VM you have around 120GB of VMDK files. The clone will depend on a number of things, how busy the environment is at night (any backups running?) and the capability of your storage array. I would have thought you are expecting a few hours although I really can't say for sure!

[water270]

I don't see the option clone ? Where i is ? the vm is off .. please help ...me

[RyanH84 · Accepted solution]

To clone a VM you need to have vCenter connectivity to your hosts/cluster.

Do you have vCenter?

[water270]

Yes correct.

I must connect to the Vcenter.

The problem is that my vcenter have 3 hosts also and i cannot add a fourth in the same ..

i will found a solution.

[RyanH84]

Is that a licensing constraint? What license do you have Essentials/Plus?