Azure MySQL Connector Suddenly Exhibiting Erroneous Behavior

[mackemlad]

Hello, all - Let me start by saying we are not beginners.  We have around 60 flows that have been working as expected for around 6 months until sometime between 16:49-15:02UTC today (Jul 11, 2024).

After that time we started exhibiting the following behaviors using the Azure MySQL connector:

1) Insert Row to Azure MySQL returns a failure, despite the record actually being inserted into the table.

{
  "status": 400,
  "message": "Failed to create item\r\nclientRequestId: 64388c82-xxxx-yyyy-zzzz-c0f22766ddeb",
  "error": {
    "message": "Failed to create item"
  },
  "source": "azuremysql-scus.xxxxxxxxx.p.azurewebsites.net"
}

2) Delete Row of an Azure MySQL record by row ID fails with an error.  The record is not deleted.

{
  "status": 404,
  "message": "Item not found with primary key '_'\r\nclientRequestId: 0004d8df-xxxx-yyyy-zzzz-f0dec27e30ec",
  "error": {
    "message": "Item not found with primary key '_'"
  },
  "source": "azuremysql-scus.xxxxxxxxx.p.azurewebsites.net"
}

3) Update Row of an Azure MySQL record by row ID fails with an error.  The record is not updated.

{
  "status": 404,
  "message": "Failed to Update Item with id:_\r\nclientRequestId: dde0f153-xxxx-yyyy-zzzz-63049f946cac",
  "error": {
    "message": "Failed to Update Item with id:_"
  },
  "source": "azuremysql-scus.xxxxxxxxx.p.azurewebsites.net"
}

4) Get rows from an Azure MySQL table works without issue.

5) Execute a SQL Statement (including insert and update) works without issue.

We are experiencing this issue in all environments (Dev, Validation, Test, Production) across two tenants AS WELL AS in a brand new environment containing a brand new flow that attempts to update a record in a brand new Azure MySQL database, all created specifically to try to isolate the problem in as simple of a test scenario as possible.

The best we can fathom is that the connector itself has been updated behind the scenes to negative effect and, if that is so, we hope that Microsoft is aware of the issue and is feverishly trying to recover from it.

In the real world, though, we are concerned that this is a regression, either planned or otherwise, with which we are going to have to live for some non-trivial time.

Is anyone experiencing similar issues, and have you identified a reasonable workaround that does not involve re-implementing all of our flows?

Thanks in advance for any guidance.

[Plucas]

We're also having the same issue. My logic app began failing on 7/11/2024 sometime between 10:47am CST and 1:25pm CST. Commenting in hopes of someone identifying more of the root cause behind this.