> The Machines have identical configuration since test environment is cloned > from production, I can't figure out what's happening. > Consider I've done the test with only one user working (me). > And the error occurs only in logout.
I'm certain the issue is around the use of executors somewhere. My initial thought was that you'd exhausted the container connector pool managed by some executor. Your mentioning logout gave me another idea. The org.jasig.cas.util.HttpClient component that issues logout messages uses an executor internally. I wonder if it is initially uninitialized, or if it's using connector threads under the hood. In the latter case it could be that perhaps the service you are hitting to logout in production has different characteristics than in test, and it's tying up threads abnormally. We added a flag recently to avoid following redirects to avoid a redirect loop that is fairly common in our environment; that would definitely tie up threads at logout. M -- You are currently subscribed to [email protected] as: [email protected] To unsubscribe, change settings or access archives, see http://www.ja-sig.org/wiki/display/JSG/cas-user
