> Hi,
>
> While trying to troubleshoot some uwsgi problem I noticed that uwsgi
> 1.4.6 does not reap/respawn after exiting from strace session (strace
> with master PID). Later it's not possible to stop it in the usual way:
> I have to use KILL signal. Sample session:
>
> write(2, "Fri Mar 1 13:08:55 2013 - Respa"..., 69) = 69
> wait4(-1, 0x7fffdaeab4b8, WNOHANG, NULL) = 0
> epoll_wait(225, {}, 1, 1000) = 0
> lseek(2, 0, SEEK_CUR) = 3978138
> wait4(-1, 0x7fffdaeab4b8, WNOHANG, NULL) = 0
> epoll_wait(225, {}, 1, 1000) = 0
> lseek(2, 0, SEEK_CUR) = 3978138
> wait4(-1, 0x7fffdaeab4b8, WNOHANG, NULL) = 0
> epoll_wait(225, 7fffdaeab410, 1, 1000) = -1 EINTR (Interrupted system
> call)
> --- SIGCHLD (Child exited) @ 0 (0) ---
> wait4(-1, [{WIFEXITED(s) && WEXITSTATUS(s) == 0}], WNOHANG, NULL) = 11519
> rt_sigaction(SIGPIPE, {0x1, [], SA_RESTORER, 0x34e900e7c0}, {0x1, [],
> SA_RESTORER, 0x34e900e7c0}, 8) = 0
> clone(child_stack=0,
> flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD,
> child_tidptr=0x2b8ea6a62650) = 11841
> rt_sigaction(SIGPIPE, {0x1, [], SA_RESTORER, 0x34e900e7c0}, NULL, 8) = 0
> write(2, "Fri Mar 1 13:08:57 2013 - Respa"..., 70) = 70
> wait4(-1, 0x7fffdaeab4b8, WNOHANG, NULL) = 0
> epoll_wait(225, <unfinished ...>
> strace: ptrace(PTRACE_CONT,1,133): Input/output error
> Process 11506 detached
>
> Then we see quite some of these:
>
> nobody 11510 11506 0 13:08 ? 00:00:00 [uwsgi] <defunct>
> nobody 11512 11506 0 13:08 ? 00:00:00 [uwsgi] <defunct>
> ...
>
> and logs show only:
>
> Fri Mar 1 13:09:24 2013 - *** psgix.harakiri.commit requested ***
> Fri Mar 1 13:09:24 2013 - ...The work of process 11516 is done. Seeya!
> Fri Mar 1 13:09:27 2013 - *** psgix.harakiri.commit requested ***
> Fri Mar 1 13:09:27 2013 - ...The work of process 11515 is done. Seeya!
> Fri Mar 1 13:09:29 2013 - *** psgix.harakiri.commit requested ***
> Fri Mar 1 13:09:29 2013 - ...The work of process 11514 is done. Seeya!
> Fri Mar 1 13:09:32 2013 - *** psgix.harakiri.commit requested ***
> Fri Mar 1 13:09:32 2013 - ...The work of process 11512 is done. Seeya!
> Fri Mar 1 13:09:34 2013 - *** psgix.harakiri.commit requested ***
> Fri Mar 1 13:09:34 2013 - ...The work of process 11510 is done. Seeya!
>
> In our case we trigger harakiri from app itself.
>
> Marcin
> _______________________________________________
> uWSGI mailing list
> [email protected]
> http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
>
Can you strace the master while this is happening ?
thanks
--
Roberto De Ioris
http://unbit.it
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
