Troubleshooting "not synchronized" status

[wallewek]

Thank you very much Boris.

I agree that the drive needs to be considered suspect. I will be checking it further, even though it is not currently showing any errors, and may replace it regardless. This might be a good opportunity to test StarWind image re-creation using PowerShell -- and a good reason to spend some time reviewing StarWind's PowerShell tools.

Fascinating. So you're saying StarWind _does_ use iSCSI for internal replication and synchronization, but distinguishes between internal and external access, and stops the latter when out of synch. OK, that makes sense to me.

I certainly respect and support StarWind's need to keep internal details confidential, and can only admire the wizardry occurring inside.

I have to say this, though: it was extremely unclear why my attempts to make iSCSI connections resulted in login failures. If they were being blocked by StarWind, it should have said so -- CLEARLY -- in the StarWind user interface (I know you can't change what the Windows iSCSI connection interface shows).

As for the waiting to create the target connections, I think you misunderstood what I was asking. I wasn't talking about creating a standalone image at all. I was asking about the HA imaging process, as follows:

When I create the HA image, as I understand it, the process (in brief) is to:
1. Create the initial image on the first host, along with its cache, etc.
2. Replicate that image onto the second host, including its cache there as well.
(2.5) at this point, the StarWind VSAN automatically starts synchronizing.
3. Configure iSCSI connections for both hosts.

What I noticed is that the step 2.5 can take some time, as a big thick image can take a while to fully synchronize, even if it is a new, blank partition. And while that initial synchronization is taking place, we cannot make iSCSI connections to the replica image.

The StarWind GUI and documentation give no warning or hint that this is so, leaving us to try to guess why we're getting logon errors.

Even understanding that the synchronization process does not rely on our connecting those iSCSI targets (and thus, if synch is not occurring, it's not due to that) is helpful to know.

This is based on your otherwise-excellent document "Installation and Configuration of HyperConverged 2 Nodes with Hyper-V Cluster", which _really_ should mention (and perhaps explain) this delay somewhere between steps 43 and 44.

But I really appreciate your time and patience, and helpful explanations, Boris. Thanks!

-- Ken

[Boris (staff)]

Ken,

The steps yo described are completely correct. This is how it is usually done. Yet, to minimize the waiting time for step 2.5 to complete, there is a small hack, so to say. If you need to assign a considerable amount of GB to some disk, first you create it as a small one (about 1-2 GB only), and replicate this small disk. After it is synchronized, you right-click this newly created HA image and go to Extend Size of HA Device. Then, put the size to be added to this initially small image to meet your expectations regarding its final size and enjoy. This will be expanded in a really fast manner.

As for the absence of any mention in the documentation concerning impossibility to connect the non-synchronized target, it is generally an obvious thing that a non-synchronized part of the HA device shall not get connected.

If you need more information about StarWind, you can browse through our Resource Library at https://www.starwindsoftware.com/resource-library and the Knowledge Base at https://knowledgebase.starwindsoftware.com/

Feel free to let me know if you need further assistance.

[wallewek]

Thank you Boris. That is a cute little hack, and I see how it works.

And I appreciate your links to those documentation pages. I've been through them before, but I had another look. I rather enjoyed one on iSCSI vs SMB, but now I can't find it. Oh well.

Re: your comment that

"As for the absence of any mention in the documentation concerning impossibility to connect the non-synchronized target, it is generally an obvious thing that a non-synchronized part of the HA device shall not get connected."

Let me just say that I strongly disagree. Yes, it's obvious to those who have worked with StarWind VSANs a bit, and understand how they work.

But that's not who you write detailed "Installation and Configuration" documents for, is it?

The fact that you might have to wait hours between two sequential steps in an otherwise complete, detailed document, and that if you don't, you will get error messages that don't even hint at the real cause -- no, I don't think that's obvious at all.

I could probably write on one page some key concepts about StarWind VSANs that I didn't understand when I started, and don't seem to be outlined clearly anywhere. Nothing major -- just fundamental principles.

This is a great example of why I say documentation shouldn't be written by people who know a product intimately. They simply can't see it "from the outside", which is the perspective they need in order to communicate effectively with the people they are trying to help.

But I really do appreciate your patience. Use that feedback as you will.

-- Ken

[Boris (staff)]

Ken,

We do appreciate your feedback and will take it into account when preparing new tech documents and updating the existing ones. We cherish our clients and aim at making their user experience the best one.