I just recently converted one of our limit groups to be machine-level, but since then I’ve seen a few instances where it looks like two slaves that are running on the same machine are still holding separate stubs. Is this a known issue in 7.1.2.1?
Are you running Pulse? Would you be able to send the Pulse logs for when this happened? There is a bug we are aware of where Repository Repair can sometimes remove a limit stub from a machine that is still actively using it in the Orphaned Limit Stub check.
Maybe I’m missing something, but that sounds a bit like the opposite problem.
In my case, we have certain machines that run 4 slaves each. Yesterday I noticed that there was a machine-level limit that had one of these machines listed twice in the “Stub Holders” column while two of its slaves were running tasks that required that limit, and the total number of machines listed there matched the reported number of stubs in use. I haven’t seen reliable occurrences of it since then, so maybe it was a fluke.
The bug I was mentioning does not check the stub back in, but deletes the holder from the list of holders. In your case, it could have removed the machine from the list of machines holding the limit at which point another slave from the same machine checked it out.
But if it removed the machine from the list of holders, why would two occurrences of the same machine name show up in the list?