@olivierlambert Thanks Olivier. I will update XO and retry the restore. I'll let you know the outcome.

@olivierlambert Sussed it! The remote storage must not contain any encrypted data using the old algorithm! That's where I went wrong I think!

Feature Request: Would it be possible to specify a delay between "Number of retries if VM backup fails"? Last night there was an internet outage that affected many of my clients s3 backups. I noticed that the retries I had specified all happened immediately after the internet failure with no delay between them. Because the internet was not restored they all failed too. It would be good to set a delay between retries to allow time for the internet to be restored. Today I manually restarted all the failed backups.
Thanks.

I will backup again without deleting the snapshot and then attempt a differential restore.
I will update this thread when done.

Thank you,
I will consider that as a way forward.

I started with XCP 1.0 a few years ago then 1.1 and 1.6 'till xenserver 6.2 became open source. Been with xenserver ever since. I run a small (one man band) company providing IT and sysadmin support to fourteen small to medium sized companies. All are virtualized on various versions of xenserver. All are using standalone servers. Some more recent ones use bcache local storage with redundant mdadm array as backing device and ssd as caching device. /dev/bcache0 then becomes local repository so all vms in repo benefit from ssd caching. Citrix pissed me off with the release of xs7.3 so I am really pleased to find this fork. Having started with XCP 1.0 it feels like coming home.
I came into IT nearly 20 years ago having spent the previous 30 years in electronic engineering. I am self taught. I have to say I found electronics easier...

@dvdhngs
Run lvscan to see if there are any active logical volumes. If there are run vgchange <group name> -an to deactivate all logical volumes. Now try to stop the array.

@frank-s Hello again. My issue is resolved. I was using used disks which had previously been used with mdadm. Although I applied "mdadm --zero-superblock" to every partition and the device itself before deleting the partition tables and dd zeroed the first 50 Gigs of the drives, the installer didn't like it. Maybe I should also have zeroed the end of the drive??? Anyway I zeroed out the entire drive (both of them) with dd and the installer worked like a charm. I also tested upgrade which also worked flawlessly.
Thank you for the great work you are doing. Over the years I have spent many many hours loading xenserver with bootable raid. This is going to make life so much easier...

@Danp Thanks for pointing me in the right direction. I was following an older howto where --global was not specified. That was my first problem - now fixed!
Second problem was that I had installed into the root directory of Centos7 so the paths for creating the links were different. Once I had found the correct paths and created the links all was good.
Thanks again.
Frank.

@florent Now this is just from memory, and I didn't take notes, but I think when I did that it restored everything to the same SR. I'll test it tomorrow and let you know the outcome.

@lsouai-vates Thank you. I was able to work around the issue using the snapshot so don't need any help right now. Please keep me posted on the bug fix.
Thanks.

@olivierlambert Thanks. I'll watch for the outcome. Appreciate all that you guys do.

FYI: I have tried this on a completely different installation with a different client - same error, but without "For each VDI, select an SR" restore is OK
Thanks.

Please note that if I don't use "For each VDI, select an SR" the backup restores normally. This is true whether or not differential restore is turned on.

@olivierlambert I have not done this with the vm I needed to restore this morning. It's very large and I don't want a massive daya copy during working hours. I tried what you asked with a small vm and I did not turn on differential restore. I got the same error. Nothing appears to be logged for the time the error happened.

@olivierlambert I have updated XO Xen Orchestra, commit fe92e Master, commit d8370. I tried to restore a small vm with only one vdi. I can confirm that this error still happens but only if I use "For each VDI, select an SR". Otherwise it restores OK but, of course, any additional vdis would be restored to the same SR. I still wonder if I'm doing something wrong. Here is a screenshot of my restore dialog:

@olivierlambert Thanks Olivier. I will update XO and retry the restore. I'll let you know the outcome.

Hopefully the times when restore is needed is hardly ever or even never. This morning, however, following a major power outage that took out an entire street, on reboot one of the vms was damaged and wouldn't boot. This vm has 2 vdis but each vdi is on a different SR. I attempted a differential restore specifying the correct SR for each vdi but the restore failed to start. It failed with the error: "Choose a SR and a Backup". I had chosen a backup from 18:00Hrs last night and I selected "For each VDI, select an SR" specifying the correct SR for each of the two vdis.
Because it did not work, and I was under pressure, I chose instead to create a new vm from the snapshot associated the same backup. It created the new vm instantly (fast clone) and it booted OK. So the client is happy, but I am left wondering if future restores will be reliable. Was my procedure correct? Can anyone throw any light on this for me? XCP-NG 8.2.1 is fully patched and XO (from the sources) is relatively recent but not completely up to date. Xen Orchestra, commit 6ecab Master, commit d499f.
Thanks.

@florent Thank you.