https://issues.apache.org/bugzilla/show_bug.cgi?id=56703
--- Comment #7 from Rainer Jung <rainer.j...@kippdata.de> --- Hi Martin, yes I think r1334399 was a bit to much. There are cases where ajp_abort_endpoint() is called with shutdown==JK_FALSE because the shutdown already happened in code lower down. That code doesn't adjust the connected counter, so we have to do it. That part was correct before r1334399, wrong between r1334399 and r1618000 and should be OK after my commit r1618000. The other case was when the socket "sd" was already INVALID. In that case my code inspection told me that we should not adjust the connected counter, so here r1334399 was right and r1618000 still is right in not adjusting connected when sd is INVALID. I do hope that your production proves us right. I see no risk of running the patch in production, because apart from the connected counter nothing has changed. So only your monitoring/status page display will change. The full 1.2.41 dev has a few more changes/fixes, but they should be low risk and I'm not aware of new problems. Thanks for the cooperation. Looking forward for your production results. Regards, Rainer -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
