Restoring ANY database to Azure MI fails

Question

Our freshly provisioned Azure SQL managed instance fails restoring backups from URL.

According to the server log the restore itself seems to finish after a certain amount of time but errors in following processes lead to a termination of the whole process...

The error message prompted by SSMS is

Msg 22003, Level 16, State 1, Line 7
Stale/aborted version cleanup was aborted for database id '5' due to database shutdown.
Msg 3013, Level 16, State 1, Line 7
RESTORE DATABASE is terminating abnormally.

The relevant part of the server log can be found here.

I tried backup files from different SQL Server versions (2012, 2016, 2019) -- even the backup of an empty database causes this error...

Did anybody experience something similar? Google does not seem to help at all.

The command used is:

RESTORE DATABASE Test FROM  URL = N'https://STORAGEACCOUNT.blob.core.windows.net/STORAGECONTAINER/Test.bak'   

The necessary credential got created via:

CREATE CREDENTIAL [https://STORAGEACCOUNT.blob.core.windows.net/STORAGECONTAINER]
WITH IDENTITY = 'SHARED ACCESS SIGNATURE',
SECRET = 'sv=xxxxxxxxxxxxxxxxxxxxxx';  

I removed STORAGECACCOUNT and STORAGECONTAINER as well as most of the SECRET.

Answer 1

When using a managed service, I expect that getting past errors like these ones is outside of your control:

Windows Fabric partition 'c26ec868-5d64-4773-b627-dc346663bd2b' (partition ID 'D0C328AD-F880-49BE-9E9F-149444D63C2C') encountered error 'CHANGE_ROLE_FAILED' and is reporting 'transient' failure to Windows Fabric. Refer to the SQL Server error log for information about the errors that were encountered.. If this condition persists contact the system administrator.

Error: 41621 Severity: 16 State: 1.
Fabric replica publisher encountered an error (SQL Error Code: 41615 State: 3) while publishing event 'FABRIC_REPLICA_STATE_CHANGE' to subscriber of type 'SUBSCRIBER_TYPE_FABRIC_DBR' on Fabric partition 'c26ec868-5d64-4773-b627-dc346663bd2b' (partition ID: 'D0C328AD-F880-49BE-9E9F-149444D63C2C'). Refer to the SQL error code for more details. If this condition persists contact the system administrator.

Error: 41629 Severity: 16 State: 0.
Fabric Service 'c26ec868-5d64-4773-b627-dc346663bd2b' encountered a permanent error while performing a Windows Fabric operation on 'c26ec868-5d64-4773-b627-dc346663bd2b' database (ID 5). Refer to the SQL Server error log for information about the errors that were encountered. If this condition persists contact the system administrator.

[FabricDbrSubscriber::RestoreRoutine] Restore FAILED with result: 3 '[FAILED]'

In this case, the "system administrator" you need to contact is Microsoft. I'd recommend that you reach out to them via support channels.

Answer 2

Just got feedback from Microsoft, this is a known issue which is currently being fixed.

As per their recommendation creating the backups with the option CHECKSUM leads to restorable backups.

BACKUP DATABASE [DBNAME] TO URL = N'https://<storage_account>.blob.core.windows.net/<backup_container>/on_prem_db_full.bak'
WITH CHECKSUM

Answer 3

this issue is known to MS team and they are working on this. Use below code that will initiate backup backups with checksum and will be helpful for DMS service where you can not add explicit Checksum keyword for backups.

EXEC sp_configure 'backup checksum default', 1; GO RECONFIGURE WITH OVERRIDE; GO