MarkMappingOffline hangs until timeout

[christab81]

In version 1.3.3 MarkMappingOffline just hangs until timeout exception occurs. When downgrading to 1.3.2 it works as expected.

Here is a stacktrace dump:
at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.StoreOperation.Do()
at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.BaseShardMapper.Update[TMapping,TUpdate,TStatus](TMapping currentMapping, TUpdate update, Func4 constructMapping, Func2 statusAsInt, Func2 intAsStatus, Guid lockOwnerId) at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.ListShardMapper1.Update(PointMapping1 currentMapping, PointMappingUpdate update, Guid lockOwnerId) at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.BaseShardMapper.SetStatus[TMapping,TUpdate,TStatus](TMapping mapping, TStatus status, Func2 getStatus, Func2 createUpdate, Func4 runUpdate, Guid lockOwnerId)
at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.ListShardMapper1.MarkMappingOffline(PointMapping1 mapping, Guid lockOwnerId)
at Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.ListShardMap1.MarkMappingOffline(PointMapping1 mapping)

[ms92ita]

I've encountered the same issue and I have resolved with the downgrade of version. But in my case I needed version 1.3.0 to resolve the timeout.
With the version 2.0.0 on preview, I've encountered a Null Reference Exception in MarkMappingOffline with this stacktrace:

in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.StoreOperationFactory.CreateUpdateMappingOperation(StoreOperationCode operationCode, ShardMapManager shardMapManager, Guid operationId, StoreOperationState undoStartState, XElement root, Guid originalShardVersionRemoves, Guid originalShardVersionAdds) in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.StoreOperationFactory.FromLogEntry(ShardMapManager shardMapManager, IStoreLogEntry so) in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.StoreOperation.Do() in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.BaseShardMapper.Update[TMapping,TUpdate,TStatus](TMapping currentMapping, TUpdate update, Func'4 constructMapping, Func'2 statusAsInt, Func'2 intAsStatus, Guid lockOwnerId, MappingOptions options) in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.ListShardMapper'1.Update(PointMapping'1 currentMapping, PointMappingUpdate update, Guid lockOwnerId, MappingOptions options) in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.BaseShardMapper.SetStatus[TMapping,TUpdate,TStatus](TMapping mapping, TStatus status, Func'2 getStatus, Func'2 createUpdate, Func'5 runUpdate, Guid lockOwnerId, MappingOptions options) in Microsoft.Azure.SqlDatabase..ElasticScale.ShardManagement.ListShardMapper'1.MarkMappingOffline(PointMapping'1 mapping, Guid lockOwnerId, MappingOptions options) in Microsoft.Azure.SqlDatabase.ElasticScale.ShardManagement.ListShardMap'1.MarkMappingOffline(PointMapping'1 mapping, MappingOptions options)

[marklar423]

Downgrading to version 1.3.0 didn't work for me on Azure, but downgrading from 1.3.3 to 1.3.1 did work.

On version 1.3.3, I also get a timeout error when calling MarkMappingOffline

[nicksolsen]

Version 1.3.1 worked for me.

[coldwell]

I also experienced timeouts on 1.3.3. Downgrading to 1.3.2 worked as expected.

[brainded]

Has this been resolved in later builds? I am still on 1.3.2.

[jaredmoo]

Sorry, no investigation or fix yet.

[brainded]

According to AppInsights the particular culprit is the following stored proc causing 100% of my timeouts: __ShardManagement.spKillSessionsForShardMappingLocal

[jaredmoo]

That makes sense. Killing sessions can be slow if you have large transactions that are slow to roll back.

[brainded]

Is there something I could do to preemptively guard against that? Something I could potentially call to tell me that its not likely to work?

[jaredmoo]

I believe so, SQL has DMV's that you can query to get information about active sessions/transactions/etc.

[devscalemat]

Is there any update/advice for this issue?
It tried to upgrade/downgrade client version. I have almost full elastic pool right now (97 dbs) and it would be nice to remove not needed ones since some tenants are no longer using the app.
I'm getting timeouts every time i try to set mapping to offline

[scottpantall]

I came across this issue using v2.3.0. I was able to use this method by including the MappingOptions parameter. mapping = ShardMap.MarkMappingOffline(mapping, MappingOptions.None);.