[CosmosDB] Review Scaling and RuntimeScaleMonitoring Settings

[FinVamp1]

3The potential exists for the Runtime Scale Monitoring decide to Scale In and out depending on the work length and Physical Partition count.

This came in as a CRI 405625406.

Repro steps

Provide the steps required to reproduce the problem

1. Create an App on Elastic Premium
2. Create a CosmosDB instance.
3. Connect these to the same VNet.
4. Create a CosmosDB trigger and point it to the collection from Step 2.
5. Enable Runtime Scale Monitoring
6. Add more than 1000 documents to the collection above.
7. The Trigger on one instance will process the back log but the Scale Monitor will try to Scale In but also log the error that the remaining work is too high.

Events logged when three instances.
Number of instances (3) is too high relative to number of partitions for collection ({collectionName}, 1).   Events that are logged with one instances.
Remaining work for collection ({collectionName}, 9543) is too high relative to the number of instances (1).

Expected behavior

To allow more concurrency to allow the backlog to be processed in an accelerated manner.

Actual behavior

The work will only be

Known workarounds

The backlog will be processed on a single instance.

Related information

Provide any related information

• Package version 4.0.0
• Links to source https://github.com/Azure/azure-webjobs-sdk-extensions/tree/dev/src/WebJobs.Extensions.CosmosDB

[FinVamp1]

HI, I'm also in the Functions team as a a software engineer and this came in as a customer issue. The answer you provided doesn't mean that there isn't a need for a fix related to this.