https://issues.apache.org/bugzilla/show_bug.cgi?id=47645
--- Comment #7 from HWS <[email protected]> 2009-08-06 06:43:46 PDT --- Before recompiling and doing new tests (which I have to delay till after the next hang), here some remarks: > I had a similar situation >with the event MPM once and this same implementation of apr_pollset_poll(), >but I didn't have time to debug or to narrow it down to one of {Event MPM, >APR, Solaris kernel}. >You will likely be able to work around the problem by selecting a different >poll implementation. As far as I can see, 2.2.11 used exactly the same implementation, without any difficulties. The only change now is setting FD_CLOEXEC. >The TCP layer should answer a SYN packet. Did that happen? >What does netstat show at the time of the hang? If connections are able to >reach the ESTABLISHED state but not responded to, then TCP is doing the right >thing and httpd isn't finding the connection. If none reach the ESTABLISHED >state, then it is a TCP layer issue. I have to wait for the next hang, but as far as I remember, snoop did not show any 'A' packets as reply to 'S' packets. If I remember the netstat output correctly, old connections were visible, but no new ones. >BTW, what exactly did you do to clear up the hang? Often nothing, but then it took hours to recover. Graceful restart always cleared the hang. -- Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
