• 0
captainfred

DSM on ESXi 6.5 and using RDMs

Question

Hello,

 

Those of you who are running DSM on ESXi 6.5 and using RDMs, which mode did you use for your disks (Dependent, Independent - persistent or Independent - non persistent) and which disk compatibility (physical or virtual)? and why?

 

I'd like to use Dependent so I can create a snapshot before an update for rollback, but does this work when they are large RDM drives? Anyone tried?

 

Thanks

Edited by captainfred

Share this post


Link to post
Share on other sites

10 answers to this question

Recommended Posts

  • 0

What is your objective in using ESXi to provide a client to host DSM?  Get access to DSM features for utility? Improve hardware compatibility?  Performance?  Reduce the number of physical boxes? Where is your storage - is it drives physically connected to the ESXi platform, or SAN?  You should be able to answer these questions to decide how to approach your DSM storage strategy.

 

For me, I am using ESXi because it is the most effective way to get directly-connected NVMe SSD's to be functionally supported on DSM.  If it were not for that, I would run DSM baremetal for the best performance.  Initially, I used VMDK's to assign storage to DSM from NVMe based storage pools.  Later I discovered that the NVMe drives could be passed via physical RDM and presented to DSM via SCSI translation.  This offers spectacular performance by exposing the NVMe SSD hardware to DSM as much as possible.  Similarly, hardware that can be natively supported by DSM (10Gbe NIC, SATA controller) are passed through to the VM for best support and performance.

 

Is it even possible to do snapshots with RDM?  I thought that Dependent mode is only available with a VMDK.  Even with a VMDK, I think you would have to have a lot of free space in your storage pool.  This might be feasible in a large VM environment with an external SAN, where the objective of hosting DSM is not to maximize storage available.  Again, in my use case of using ESXi specifically to host DSM, I want all my directly-connected storage allocated to it.

 

My strategy to test upgrades with ESXi is to maintain a small storage pool separate from DSM (could be the same pool running ESXi itself) and build a second test VM with the same attributes as your production VM.  That way it's easy to copy off and/or burn down and rebuild without affecting production.  I've never considered using a VMWare snapshot to resolve this due to the limited size of the storage pool, but if I had a dedicated SAN behind ESXi, that might be more feasible as a test approach.

 

Edited by flyride

Share this post


Link to post
Share on other sites
  • 0

But how does creating another VM to test DSM updates give you a backout because the VM only has the loader in it really, the actual changes are in a partition on the physical drives with the data in other partitions?  unless you are using VMDKs on the physical disks?  

Share this post


Link to post
Share on other sites
  • 0

My test VM is VMDK which I think is more than adequate for minor updates.

 

For a major update (such as 6.1 to 6.2), I would idle my primary NAS instance, and use one or more of its storage bays (and spare drives) for a temporary RDM configuration.

 

Share this post


Link to post
Share on other sites
  • 0

Does your HDs start on Disk 2 (no Disk 1 because thats the boot loader and therefore hidden) with it running on ESXi ?

 

This is same for me on 2 separate machines - passthrough and using RDMs.

 

Capture.JPG.1b1c012aaa2660415cc26cf98c3d9967.JPG

Share this post


Link to post
Share on other sites
  • 0

No, this is derivative of how your controllers are set up in your VM.  Make sure drive numbering starts with 0:0 and not 0:1, and also make sure you don't have an extra controller definition with no drives attached to it.

Share this post


Link to post
Share on other sites
  • 0

Yeah I think I had a SCSI controller there before which I didn't need and wasn't doing anything but removed it, I bet DSM needs to do a reinstall to redetect the drives somehow? any ideas how to get it to do that? 

 

Although it doesn't really matter i suppose?

 

Thanks

Share this post


Link to post
Share on other sites
  • 0
On 9/5/2018 at 12:00 PM, captainfred said:

Does your HDs start on Disk 2 (no Disk 1 because thats the boot loader and therefore hidden) with it running on ESXi ?

 

This is same for me on 2 separate machines - passthrough and using RDMs.

 

Capture.JPG.1b1c012aaa2660415cc26cf98c3d9967.JPG

 

Can anyone solve this?

If I leave the Controller 0: 0 the boot and 0: 1 the storage disk, XPenology will not find the disk ...

 

But if I set the storage to 1: 0 for example, then he finds the disk and installs it.

However, inside the system it is detected from Disk 2 ...

 

It also appears as eSATA ... as if it were a USB drive.

 

In File Station, two units appear as "satashare"

 

How to solve?

 

I'm using MicroServer Gen 8 standard config, ESXi 6.0
DSM_DS3615xs_6.2.1-24922u2

Share this post


Link to post
Share on other sites
  • 0
On 5/3/2020 at 1:43 AM, flyride said:

Use FixSynoboot.sh

 

Would this fix the numbering of the hard disks even though I'm not running 6.2.3 ?

Share this post


Link to post
Share on other sites
  • 0

Synoboot problems and the exposure of the boot loader device can happen with earlier versions, but does not very frequently.  It won't hurt to install the script and see if it helps you.

 

However, hard disk numbering and position is controlled by DiskIdxMap and SATAPortMap which is a configuration in the loader.  Since everyone's hardware configuration is different, it's an analysis and application based on your observations of your own system.  You might want to read up on that if the script doesn't solve (or completely solve) your problem.

 

  • Thanks 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.