Access the deployment error log from the management portal and Visual Studio
Since deployment error are not uncommon, and many times you cannot use Remote Desktop in order to view the Windows Azure Log in the instance which constantly keeps recycling, access to the deployment error log from the management portal could save us developers a lot of headache, and a lot of support calls for you MS guys
1. When uploading a package, analyze it and tell me in clear way if there are missing references/DLLs or anything else that will stop the role from running.
2. Better and more error messages. When something goes wrong make sure to tell what went wrong. If a role recycles it is almost impossible to troubleshoot it!
3. if it runs in the emulator it should run in Azure. Make sure applications can not access more files or features in the emulator than they can in Azure.
4. Make the emulator stable so we don't have to restart the emulator and/or visual studio all the time.
Take a look at the Azure forum and you will see a lot of people confused about why the process is stuck on 'busy' or initializing, or other strange behaviours. Maybe its missing a reference? Who knows. Its often a trial and error. How about giving some useful feedback - like the output window in VS.net - when attempting to deploy
Paul Turner commented
I think I just lost a day trying to get to the exception information I needed to solve a deployment issue. The Worker Role host doesn't deal with unhandled exceptions very gracefully at all. Whilst having the host crash isn't the worst idea, it really pains me to miss the rich .NET exception data.
I would like to see, at a minimum, any unhandled exception being written to a log (the Application log is suitable, especially as it's already exported by WAD), before the host terminates. No exception should just disappear.
Ideally, we should be able to see any exceptions encountered in deploying a role (deployment, startup scripts, role start) through the Azure dashboard. I appreciate that might be a big ask, but it would be very convenient.
James D. Schwarzmeier commented
Agree -- when a deployment or app start problem happens (including endless restart/cycling loops), it's very painful to go digging to try to find out what went wrong. Sometimes one has to look in storage logs, sometimes one has to remote desktop and dig through files local to each server instance, etc. Both the portal and tooling in Visual Studio should provide more detail about what's happening when apps don't deploy or start -- and provide an easier way to get at the full logs.
Dealing with unhandled exceptions is currently a big pain. +1 for this.