DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT <http://nagoya.apache.org/bugzilla/show_bug.cgi?id=27079>. ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND INSERTED IN THE BUG DATABASE.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=27079 On client restart server no longer responds [EMAIL PROTECTED] changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From [EMAIL PROTECTED] 2004-02-20 12:48 ------- Unforunately all the mutex implementations have their own nuances. This particular nuance of pthread cross-process mutexes is why it is not the default on AIX. FWIW, Solaris offers APIs for recovery of a cross-process pthread mutex after the holder crashes and Apache 2 uses those APIs, but some Apache users and customers of Apache-based servers have unfortunately found out that the recovery after the mutex holder crashes is not reliable. The accept mutex has to be changed in that situation to something besides pthread. What Apache MPMs could do is implement an exception hook and release the accept mutex in the exception hook if it was being held. This wouldn't catch all accept mutex scenarios (e.g., some thread caused a crash right as the OS gave the listener thread ownership of the mutex and the MPM didn't get a chance to take note that it held it). It wouldn't help for any mutexes used by modules either. And regardless of the mutex issues, the original cause of the crash still hs to be fixed. I'm reopening the bug and will close it as WONTFIX, as there is work that could potentially be done to help this situation but at the present time we would not plan to do it ;) --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
