Start a Conversation

Unsolved

This post is more than 5 years old

68853

July 23rd, 2014 08:00

Expand Volumes on PS6100E

I have a PS6100E with 48TB that has almost 30TB reserved.

I am running SAN HQ and EqualLogic Group Manager but not VSM. I have installed VSM but I cannot seem to login. So I want to do the expansion of volumes using GM and the vSphere Client Datastore Volume Properties.

I know I can go into the Group Manager, Select a Volume, Modify Settings, Go to the Space Tab, and increase the Volume Size. Then I can go to the Datastore in the vSphere Client and increase its capacity. I just want to be certain this is the optimum way to do this and if there are any other steps I should be doing.

Should I power down all the VMs on one volume and then migrate them to another before expanding and increasing the datastores? I know the documentation says I can do this on the fly but I'm cautious.


Thanks in advance,

Dave

23 Posts

July 23rd, 2014 09:00

Don,

Thanks for the quick reply.

I am running ESXi v5.1 and VMFS v5.54. I currently am allocating 20% of the volume for reserve space but I think that's too high and want to reduce it to 10% like you.

Good suggestion on the new volume(s) as I would indeed be adding VMs to the datastore.

The volumes aren't being replicated but I am replicating some individual VMs.

The Storage Array Firmware is 7.0.1 so I should be good there.

Also, I have 4.0 of VSM and, you're right, this might be a Java issue. I'll see if I can resolve that.

Thanks much,

Dave

5 Practitioner

 • 

274.2K Posts

July 23rd, 2014 09:00

Hello Dave,

You left out some details which would help answer this.   Version of ESX, formatted VMFS v3 or V5?

I'll asssume ESXi v5.x and formatted VMFS v5.   Whenever doing a resize I always create a snapshot first.  Here's where VSM is really beneficial in this case.  It will ask VCS to quiet the datastore first, then create the EQL snapshot for much better consistency.  Remember to allocate some snapshot reserve too.  I like to have at least 10% of the volume allocated for reserve.  

If you do it manually,  after you resize it via the EQL GUI, you must do a rescan of the iSCSI adapter(s) first.  Otherwise ESX won't see the new space.   When you do the resize operation I prefer to do it when the IO is quieter.  Once the new space is seen you can use the extend filesystem option.   If you are in a cluster, you might need to rescan all the nodes so they pick up the new space as well.

However, consider adding a new volume instead.  Especially if you have only a few megasized volumes now.  More, smaller volumes tend to work better than few huge volumes.  Especially as the number of VMs per Datstore goes up.

If the volumes aren't being replicated, you might want to try to run space reclaim on them too.  Free up some space that way.  Reclaim requires ESXI v5, VMFS v5, (not upgraded from V3) and EQL FW 6.0.x or greater.

Other notes:

1.)  For ESXi you must run EQL FW 6.0.6-H2 or greater.  Otherwise you risk VMFS datastore corruption.

2.)  Configure ESXi per the EQL best practices guide

    en.community.dell.com/.../download.aspx

I would suggest you open a support case for the VSM problem.  Common causes are Java versions.

Also, when you log into VSM, you are actually logging into vCenter.   With version 3.5.x and 4.0 of VSM there's no need to log into the appliance itself.

Regards,

5 Practitioner

 • 

274.2K Posts

July 23rd, 2014 09:00

Re: Snapshot reserve.  That would be  a MIN setting just for the resize.  Are the volumes thin provisioned?  Since that also thin provisions the reserve too.  It would only be 20% of the allocated portion of the volume.  

You can also try to reclaim deleted space.    this KB explains how to do that for ESXi v5.0/5.1. Vmware changed the process in 5.5.

kb.vmware.com/.../search.do

Don't try to reclaim more space than what ESXI shows as available.   I.e. if it says 500GB free don't try to reclaim 600GB.   You provide a % of the volume to reclaim.  So running 300G reclaims multiple times works better.  Reclaim in ESX creates a "baloon file" first, then deletes it, then runs UNMAP against those blocks.

Re: 7.0.1.  I would upgrade to 7.0.5.  There are some important fixes there.  

re: VSM v4. Recent versions of Java require that you create an exception for self signed certificates. Also, VSM v4 isn't accessible via the VI client.  You have to use the 5.1 Web GUI, since Vmware is moving to Web GUI only with the next release of ESXi.  

5 Practitioner

 • 

274.2K Posts

July 23rd, 2014 10:00

You can thin provision EQL volumes at any time.  It's not deteced by the host.  It's only whether EQL allocates the space up front or on demand.

I don't move off VMs, I just make sure they aren't really pushing I/O at the time.  

23 Posts

July 23rd, 2014 10:00

Good info on the reclaiming deleted space. It turns out, though, for whatever reason, only 2 of the 3 volumes are thin provisioned. It appears that I can check the box for Thin Provisioned Volume on the volume not thin provisioned. I'm not sure if that's advisable or not but will check before doing so.

I would want to expand these volumes during non-production times and then can upgrade the FW to 7.0.5 at that same time.

I wasn't aware of having to use web gui for the VSM. Thanks.

Is it good practice to move my VMs off the volume before expanding? I will be adding new volumes and this isn't a concern with them but I do want to expand at least one of my current volumes.

Thanks,

dave

5 Practitioner

 • 

274.2K Posts

July 29th, 2014 11:00

That's an easy one.   Check under the advanced tab for that volume, and you'll likely find "multihost access" is disabled.   So only the first sever to connect to it will be allowed to connect.  Regardless of what the ACL says.   That's the default setting, you have to enable it when you create a volume.   This prevents accidental double mounting of volumes.  

23 Posts

July 29th, 2014 11:00

Don, related to this expanding volume inquiry, I have decided to create some new volumes as well so I have the smaller volumes rather than one or two large ones. I created a 5TB volume but notice that the iSCSI connections are not the same as the other 3 volumes originally installed. The other 3 volumes have connections for all 4 hosts (16 connections) but the new volume only has connections to one host (4 connections). Before I have vSphere recognize the new volume, I need to correct the iSCSI connections, I presume. I'm not seeing where to add the iSCSI connections in my group manager. I've read to run the iSCSI Initiator but am not sure where that is. Thanks in advance for any help.

Dave

23 Posts

July 29th, 2014 13:00

When I Modify Settings for that volume and go to the Advanced tab there isn't any "multihost access" option. There is, however, a check box that is already checked for "Allow simultaneous connections from initiators with different IQNs".

5 Practitioner

 • 

274.2K Posts

July 29th, 2014 13:00

That's the one. At the CLI, which is what I prefer, it's called multihost access.  :-)

To start new iSCSI sessions, you just need to rescan the iSCSI adapter(s).  It will discover the new volume then login to them.  You have to do this on all nodes.

Or if you are in a cluster, at the cluster name, right mouse "Rescan" option.  That will do all the nodes in that cluster.

Regards,

23 Posts

July 30th, 2014 08:00

Don, Thanks for all your help!  Yep, I did the rescan and now the connections show up in Group Manager. vSphere still doesn't see the volume though. Missing a step somewhere.

5 Practitioner

 • 

274.2K Posts

July 30th, 2014 09:00

Did you format it?    You have to use "Add Storage" under Configuration->Storage to create a new VMFS Datastore from a new EQL LUN.

When it comes time to select the LUN,  I expand the Path name to show the complete EQL volume name to make sure i'm doing the correct volume.

Regards,

4 Operator

 • 

1.7K Posts

July 30th, 2014 09:00

Just an additional information. If you add storage to a esxi node which is part of a vSphere Cluster the vCenter trigges a HBA Rescan on all members of the cluster. So at the end all Hosts see the new Datastore.

The above is only true for "Block" Storage and not for IP based File Storage like NFS.

For next time just install the Dell EQL Virtual Storage Manager (VSM) as an Appliance and you can provisioning a new EQL Volume and vSphere Datastore from the vSphere WebClient in one step driven by the Wizard.

Regards,

Joerg

23 Posts

July 30th, 2014 09:00

Yep, just finished the Add Storage part after I sent my last post. Looks like everything is working just fine. Thanks again for your help.

No Events found!

Top