Status with RequestId in North Central US seems not to be working

Sep 8, 2011 at 5:02 PM

Doing a simple Get Request to the Status url

You get: The underlying connection was closed: An unexpected error occurred on a receive.

The export action started ok, the request id was generated and in the Blob a file with 0 bytes was generated.

The database is about 30GB

Any Ideas? Thanks in advance.-

Sep 8, 2011 at 6:48 PM

Are you using the client side tools?  Then the get status URL has no effect.

If you are using the service, then you should not be hitting the service endpoints directly.  Use the tool to request the status.  There is not "one" service endpoint for status.  There are multiple (per data center in fact).  So you need to use the tool in order for it to route you correctly.

If you are trying to write your own tool look at the DatacenterMapper class to see how the requests are routed by the database name, and then how the queries are formed.


Sep 8, 2011 at 8:11 PM

Yes, I am using the clent side tools.

I think I know what was the problem, I started the export action with the credentials of a user that isn't SQL Azure Administrator, it is just a db_owner of the db.  Now checking the status with the same credentials makes the service to fail.  With the Administrator credentials, the status request works ok.

If the export action is running for about 5 hours and the bacpac file  is still in 0 bytes means that the export action has failed? Should we start the export action with the Sql Azure Administrator credentials?

Sep 12, 2011 at 3:38 PM

During export nothing is written to your blob storage until the entire operation is complete.  So the file should stay at 0 bytes until the operation is 100% done.  We don't want to write a partial file to your storage, or have users pick up the file thinking it is done and maybe get other errors.

So the entire bacpac is kept in the Windows Azure servers until the operation completes, then it is copied to your blob storage account as one of the final steps in processing (after that we flag the job as complete so you can see the status).