Azure: Performing the role of the work roller through "recycling"

Question

I am currently working on an Azure project that works 100% locally with emulator resources. Now I'm trying to deploy a working role, but I have a problem that I'm not sure how to troubleshoot.

After deploying a working role on my Azure portal, two instances continuously go through "recycling".

I can try RDP in the role, but I only have about a minute to look around until the connection closes, I guess due to processing.

After some searches, it does not seem to be a very common problem. Is there something trivial, I forget what might cause this problem? How could you solve the problem? Thank you for your time:)

Answer 1

From Avvash above:

Yes. this means that some problems in your working role code cause the host process to crash your working role. If you look at your error stack, you should see a function or link from your code that generates this error. If you need help, open a free Azure-enabled incident for Windows Azure Support and they will help you.

Answer 2

If there is no link, you can fix this problem with:

Unzip your CSPKG file, and then unzip the .CSSX file again (just rename CSSX to zip) and match all the links and static content. There you can map what is in the virtual machine. Also in the 2-minute windows, when you are RDP, try looking for an application event log for an exception and get it, because this will be the key to finding the root cause.

IF you can see the exception in the event log and look for the exception, you can find where it was created. You can also use Intellitrace, which may require redeploying the application.

There are also ways to copy WinDBG and locks for a specific process that you can debug. I'm not sure how much you want to try, but just copy WinDBG to VM and using it will be enough (not sure how much experience you have with WinDBG, although how much time you would like to spend.)

Answer 3

In addition, this problem was repeated repeatedly. The following is a sequence of steps for debugging repeated clips of clips :

Debug Azure Role Repeat Operations

• Enable Remote Access to Your Role - RDP Login
• Check eventvwr.msc (Windows Logs → Applications, Applications and Service Logs → Windows Azure)
• View the azure text file C:\logs on C:\logs and c:\resources
• View custom logs in volume E: or F: for any custom startup run log

• Launch AzureTools and attach it to the startup processes (download WinDBG , use Utils-> Attach Debugger, select the process - WaWorkerHost / WaIISHost, etc.), use G to continue and view the debugger output for failure crashes.

  Install Azure Debugging Tools through Powershell

  PS> md c: \ tools; Import module bit-bit; Start-BitsTransfer http://dsazure.blob.core.windows.net/azuretools/AzureTools.exe c: \ tools \ AzureTools.exe; C: \ Tools \ AzureTools.exe

If all of the above items do not work - try using other tools in the AzureTools cluster - for example, fusion magazine , etc., this approach will work!

WinDBG Output Result - Inability to Find an Assembly (WaIISHost)

Answer 4

The most likely reason is that you are missing an assembly. One tactic to catch this is to wrap any launch processing in a master attempt / catch that management is logging an error in Azure storage.

If you added any links, make sure that they are set to copylocal = true and that any external assets that were included in your service package should also be included.

Answer 5

Just a suggestion: Also check the installed (if any) and any other links that you use are 64-bit. The basics of VM have a 64-bit OS. I once encountered such a problem due to problems with 32/64 bits.

Answer 6

Are your work roles out of the work cycle? Local recycling is very fast, and you may not notice it, but the turnaround time in the cloud can be long.

Answer 7

If the problem is caused by the start batch file, I stopped the loop by editing the batch file in the instance to include "exit / b 0" at the beginning. This will tell Azure that the launch was successful, and then you have all the time you need to diagnose problems without killing the VM.