So, it's happened again:

Feb 27 05:11:22 quattro courierd: SHUTDOWN: respawnhi limit reached.

I suspect it's this process:

mail       951  0.0  0.0  1920 1008 ?        S    Feb20   0:00 courieresmtp 1 
mindspring.com

I grabbed too much information just because I needed to get the sytem
running again and only had one shot at getting some information:

quattro:~# ls -al /proc/951/fd
total 0
dr-x------    2 mail     mail            0 Feb 27 13:32 .
dr-xr-xr-x    3 mail     mail            0 Feb 27 13:32 ..
lr-x------    1 mail     mail           64 Feb 27 13:32 0 -> pipe:[30783557]
l-wx------    1 mail     mail           64 Feb 27 13:32 1 -> pipe:[30778911]
lrwx------    1 mail     mail           64 Feb 27 13:32 2 -> /dev/null
lrwx------    1 mail     mail           64 Feb 27 13:32 3 -> socket:[30783558]
lrwx------    1 mail     mail           64 Feb 27 13:32 4 -> 
/var/spool/courier/msgs/2/C719502
lr-x------    1 mail     mail           64 Feb 27 13:32 5 -> 
/etc/courier/hosteddomains.dat
lrwx------    1 mail     mail           64 Feb 27 13:32 6 -> socket:[30783564]

quattro:~# lsof -n | grep " 951 "
courieres   951     mail  cwd    DIR       48,3      4096    443624 /usr/lib/courier
courieres   951     mail  rtd    DIR       48,2      4096         2 /
courieres   951     mail  txt    REG       48,3    115996    246399 
/usr/lib/courier/libexec/courier/modules/esmtp/courieresmtp
courieres   951     mail  mem    REG       48,2    471781    246243 /lib/ld-2.2.2.so
courieres   951     mail  mem    REG       48,2     79024    246256 /lib/libdl-2.2.2.so
courieres   951     mail  mem    REG       48,3     29050    180634 
/usr/lib/libgdbm.so.2.0.0
courieres   951     mail  mem    REG       48,2   1236396    246252 /lib/libc-2.2.2.so
courieres   951     mail    0r  FIFO        0,0            30783557 pipe
courieres   951     mail    1w  FIFO        0,0            30778911 pipe
courieres   951     mail    2u   CHR        1,3              213413 /dev/null
courieres   951     mail    3u  unix 0xe051bd00            30783558 socket
courieres   951     mail    4u   REG       48,5       198    719502 
/var/spool/courier/msgs/2/C719502
courieres   951     mail    5r   REG       48,2     12621     82698 
/etc/courier/hosteddomains.dat
courieres   951     mail    6u  IPv4   30783564                 TCP 
205.252.89.20:3094->207.69.200.138:smtp (ESTABLISHED)

quattro:~# strace -p 951
read(6,  <unfinished ...>

quattro:~# netstat -anpo | grep 951
tcp        0      0 205.252.89.20:3094      207.69.200.138:25       ESTABLISHED 
951/courieresmtp    off (0.00/0/0)
unix  0      [ ]         DGRAM                    30783558 951/courieresmtp    

quattro:~# rpm -q courier
courier-0.37.2.20020131-je1

The only changes I made were use the user "mail" instead of "daemon".

I haven't looked at the code for courier yet, but it looks like it uses
blocking sockets and no alarm's. Although, I find it highly suspicious
that this happens so rarely if the code does use a select before read.

Sam, any ideas?

JE


_______________________________________________
courier-users mailing list
[EMAIL PROTECTED]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users

Reply via email to