I've got an end to end machine build job that goes something like this:
- Boot to WinPE
- Deploy Image
- Boot to Production
- Apply System Configuration (Join Domain)
- Run Script A
- Deploy Software A
- Deploy Software B
- Deploy Software C
- Deploy Software D
- Run Script B
- Reboot
The issue that I'm havin is that when running the job, it will fail at some task along the line...or sometimes even succeed all the way through. I can run the full job on the same machine four times and get four different results; the failures will be anywhere from the Boot to Production just after imaging to the Reboot at the end. And when it fails, drilling into the task, the task shows that it failed with "The task timed out on the Task Server while waiting for the agent to pick it up" message, usually with ~90 minute execution time (when these tasks succeed, it's usually ~1-5 min execution time).
Looking at the Agent's Task history, the point of failure task doesn't show up. Looking at the agent logs, I see the agent complete the previous task instance, then not get the next task and just stop there. I can send another task from the console (either after waiting a bit or immediately after the failure pops) and the client gets and executes it just fine.
Has anybody experienced this and how did you workaround/resolve it?
