appvolumes won't attach error due to error on appvolumes manager

[sjesse]

Has anyone seen this error before,

Failed to reconfigure VM "rcl-w7-libp03" (5034c321-cc84-a169-ddfd-cc62f1926413):
Fault: Unspecified (InvalidDeviceSpec)

I can't seem to get appstack to attach regularlly. This happened after I tried to migrate storage arrays

[sjesse]

For the future f anyone find this , at least for me it was a sign the appvolumes was stll looking in the wrong datastore, even though it was marked as non attachable and the volume was unmounted in vmware. Reenabling the storage and marking it as attachable, and rescanning every datastore seemed to resolve this, now I need to figure out why I can't decommission my old storage.

[Ray_handels · Accepted solution]

Hey sjesse,

This has been discussed in one of the topics on the forum, i believe the question was related to moving Appvolumes to a new storage.

What we needed to do was first create the new storage, add it to vcenter and import it in Appvolumes (just just re scan at storages, it will automatically add it). Add this new storage to the storage group so it will sync appstacks automatically and then remove the storage you won't be using anymore. You can do the re scan for Appstacks after that and it will mark the appstack as missing on that old storage.

Then you would need to alter the database, go to the snapvol_files table and remove the appstacks from the specific datastore you removed. In this table only the ID of the datastore is shown, make sure to write that down before removing the datastore.

Reasoning behind this is because if it would automatically remove these records in a worst case scenario this could happen if a datastore for whatever reason was not available it would remove all references to this datastore, that's not something you would like to do.

[sjesse]

Hi Ray,

I can't seem to find any references to moving storages other then a vsan to a traditional one, and all results seem to not mention the snapvol_files table. That's a good tip though I didn't think of that. What I did was what you said I

1.)Created a storage group(Marking automatically import and replicate)

2.)Let all volumes replicate

3.)Manually searched each datastore for anything not copied(template files, writables) and migrated those

4.)Marked all datastores as unattached

5.)Unmounted each datastore

I didn't get things working into I did the manual reimports and rescans of the new and old datastores, so I'll have to try that next time.