We use the VIP swap technique of rolling out new versions of our apps. These apps use an Azure SQL Database. Starting sometime early this year we started seeing SQL Database errors during the VIP swap. The problem is very consistent, if we have database activity during the VIP Swap it is very likely to generate one or more of these errors. As you can see below in our custom formatted version of the error the actual SQL Error number is 121 and the text is 'The semaphore timeout period has expired':

SQLError Number:121, Message:A transport-level error has occurred when receiving results from the server. (provider: TCP Provider, error: 0 - The semaphore timeout period has expired.) 

We do utilize the Enterprise library for transient error handling and I have seen these errors from time to time and random points, but the fact that this is very consistent with a VIP swap bothers me. Since both deployments exist before and after the VIP swap why would there be any interruption in SQL Database connectivity? Our apps consist of a Web App, a few Web Services, and a few Worker Roles. I have seen this error from all of them at different times during VIP swaps.

Any ideas?

Hi Curious,

Thank you for your question.  

I am trying to involve someone more familiar with this topic for a further look at this issue. Sometime delay might be expected from the job transferring. Your patience is greatly appreciated.  

Thank you for your understanding and support. 

Hi,

For this issue, I would have a support case created for further research as soon as possible:

http://support.microsoft.com/default.aspx?id=fh;en-us;offerprophone

Thanks,

Cathy Miller

• Marked as answer by Curious George Wednesday, September 12, 2012 4:09 PM

Hi,

For this issue, I would have a support case created for further research as soon as possible:

http://support.microsoft.com/default.aspx?id=fh;en-us;offerprophone

Thanks,

Cathy Miller

• Marked as answer by Curious George Wednesday, September 12, 2012 4:09 PM

I'm having the exact same issue.  Is this a known bug, or am I doing something wrong?

Indeed I have received acknowledgement from Microsoft that this is a known bug (via Azure support). I also was told it was supposed to be fixed last week, but clearly this has been delayed as we are still seeing it as well.

Just an update that I was informed today that the fix was delayed. No new ETA yet. I will try to remember to update this thread when I hear and have validated it is fixed.

Just an update. This bug still exists, we see it most times when we deploy. I have not followed up with Azure support but it feels like they completely dropped the ball on this.

We get it on pretty much every VIP swap as well.

MS is failing pretty hard on the support side...

Just wanted to throw my hat into this issue as well.

Same issue when doing a VIP swap.  This is effecting our production users.

Still happening for us.  After more than a year...no progress.

Yeah it appears this has completely fallen off the radar for Microsoft. Might I suggest everyone submit a support ticket for this and maybe it will raise it to their attention as a needed fix?

• Edited by Curious George Tuesday, October 08, 2013 3:43 PM

Yeah it appears this has completely fallen off the radar for Microsoft. Might I suggest everyone submit a support ticket for this and maybe it will raise it to their attention as a needed fix?

• Edited by Curious George Tuesday, October 08, 2013 3:43 PM

Still happens so apparently this is not fixed. This is really crucial.

Over two years later this still isn't fixed.

This is happening for us as well. Very frustrating. Will be signing up for support shortly and submitting a ticket!

We're also seeing exactly this problem... three years after it seems to have been first reported!

It looks like plenty of others are seeing the same. Does anyone have a link to a Connect or Support ticket that we can add our voices to?

I just want to add to this too. I have been using azure for 1 year and have 2 cloud services and we also use sql azure databases. We have the function to disable database activity in our applications which stops all database requests. We disable database activity now before doing any VIP swap and then we enable database activity afterwards. If we do not do this we always see two of the exact same error "The semaphore timeout period has expired". I would also like a fix to this.