Let me try to answer why I said the server was shutting down automatically." I was definitely influenced by the following comment made by Tom on 21 March over email while responding to some hudson job failure messages:
We've been seeing occasional failures of other test suites (GF admin devtests, nucleus admin devtests) in which there seems to be this behavior too, where it looks like the server just dies.
I will let Tom explain what made him come to this conclusion.
When I had seen QL failing in my local environment once (unfortunately I don't have the logs available to look at now - ask Bill if he has it), it was to do with the cluster test. The nature of failure made me come to similar conclusion as Tom.
You may be spot on with your analysis that server is not really crashing or automatically shutting down. Rather, it's just that asadmin client is failing to correctly detect that it's started. If that's the case, it is definitely as severe a problem as we initially thought and you can update the synopsys and priority. If we don't want to fix it now, I am hoping we can fix our tests to be more resilient.
Finally, do we know why we have started to see this behavior (failing to detect server has started correctly) only recently? I can't recollect to have seen such issue before. What change has trigged this issue?