2

I have a fairly simple WCF Role on Azure that I am trying to deploy two instances of - the role is fairly-well united tested and I've been able to run it successfully on the local emulator with no readily apparent issues.

The role has a couple of startup tasks that run in the background as the role starts, namely the installation of a pair of Windows Services that run asynchronously in the background doing data processing.

When I deploy the Azure, the first role instance boots normally and quickly, with the Windows Services successfully installed and running (I RDPed in to verify.) The second instance permanently hangs in a "Waiting for Host" state. I've tried rebooting that role instance individually and it doesn't appear to fix the problem.

I've also tried redeploying the entire package to Azure with the same results - first role instance starts fine, the second hangs.

What can cause this problem? Where should I look to try to fix the issue?

Aaronontheweb
  • 8,224
  • 6
  • 32
  • 61
  • If you switch it down to just one instance, can you get both staging and production running at the same time? – alun Aug 14 '11 at 20:12
  • 2
    I think this is a job for support. ([http://windowsazure.com/support](http://windowsazure.com/support)) – user94559 Aug 14 '11 at 23:51
  • Alun - I have an older build of the service running two instances in production. Trying to push a newer build to staging now. So of the 4 instances I am trying to get deployed (2 production, 2 staging) I can get three of them up and running. – Aaronontheweb Aug 15 '11 at 00:33
  • Wow that's super wierd. Have to agree with @smarx, support might be your best bet. – alun Aug 15 '11 at 04:23
  • Are the roles contending for any shared resources? I'm not sure what that would be, but that's what it sounds like -- the first role starts, grabs hold of something, and the second role is waiting to grab that same resource once it frees up (which it never does). Just grasping here. Good luck! – David Hoerster Aug 15 '11 at 19:41
  • 1
    @smarx - talked to support; the issue was that my solution was deployed onto a bad node, which I didn't think was possible given what I know about the Fabric Controller. – Aaronontheweb Aug 18 '11 at 13:10

1 Answers1

1

So ultimately I resolved this issue by turning to Windows Azure support - turns out I was being consistently deployed onto a "bad node" which I didn't think was possible given how the Fabric Controller works.

Nonetheless, if you run into this issue - Azure support is your best bet.

Aaronontheweb
  • 8,224
  • 6
  • 32
  • 61