Red Hat Bugzilla – Bug 487290
client connection management needed
Last modified: 2015-05-29 16:12:37 EDT
When zillions of systems are given an action to perform at roughly the same time, the systems check in with a crude “attempt, fail, retry” loop. This is a performance and execution issue.
All customers initiating actions with more systems than Apache/DB connections available.
Those connections should be “managed”, i.e., proactively queued, and the estimated checkins for each noted as to when the expected actions will be initiated.