We are getting frequent segmentation faults from exim 4.34 parent daemon
on a high volume server. It was happening on 4.20 as well and we
upgraded to latest code to attempt a resolution but it did not help.
Last few lines of strace from seg follows, pid 6081 is the parent daemon
Running linux 2.4.23 SMP kernel.
Any ideas or things I should do to get more debugging information w/o
sacrificing scalability of exim on this high traffic box?
[pid 6081] munmap(0x40017000, 4096 <unfinished ...>
[pid 11158] <... setrlimit resumed> ) = 0
[pid 6081] <... munmap resumed> ) = 0
[pid 11158] close(0 <unfinished ...>
[pid 6081] wait4(-1, <unfinished ...>
[pid 11158] <... close resumed> ) = 0
[pid 6081] <... wait4 resumed> [WIFEXITED(s) && WEXITSTATUS(s) == 0],
WNOHANG, NULL) = 11157
[pid 6081] wait4(-1, 0xbffff494, WNOHANG, NULL) = 0
[pid 6081] rt_sigaction(SIGCHLD, {0x40296c70, [CHLD],
SA_RESTORER|SA_RESTART, 0x401794f8}, <unfinished ...>
[pid 11158] rt_sigaction(SIGCHLD, {SIG_IGN}, <unfinished ...>
[pid 11152] <... read resumed> 0x81e41d0, 8192) = -1 ECONNRESET
(Connection reset by peer)
[pid 11152] alarm(0) = 900
[pid 11152] rt_sigaction(SIGALRM, {0x40296c70, [], SA_RESTORER,
0x401794f8}, NULL, 8) = 0
[pid 11152] time(NULL) = 1086238656
[pid 11152] write(1, "421 fuse1.fusemail.net lost inpu"..., 46) = -1 EPIPE
(Broken pipe)
[pid 11152] --- SIGPIPE (Broken pipe) @ 0 (0) ---
[pid 11152] exit_group(1) = ?
[pid 11158] <... rt_sigaction resumed> NULL, 8) = 0
[pid 6081] <... rt_sigaction resumed> {SIG_DFL}, 8) = 0
[pid 11158] brk(0 <unfinished ...>
[pid 6081] --- SIGCHLD (Child exited) @ 0 (0) ---
[pid 11158] <... brk resumed> ) = 0x81e5000
[pid 6081] --- SIGSEGV (Segmentation fault) @ 0 (0) ---
brk(0x81e7000) = 0x81e7000
getsockopt(1, SOL_IP, IP_OPTIONS, "", [0]) = 0
setsockopt(1, SOL_SOCKET, SO_KEEPALIVE, [1], 4) = 0
getpid() = 11158
time(NULL) = 1086238656
select(3, [2], NULL, NULL, {0, 0}) = 1 (in [2], left {0, 0})
write(1, "554 SMTP synchronization error\r\n", 32) = 32
exit_group(0) = ?
Best,
Bryan Heitman
FuseMail
