[Mailman-Users] mailman 2.0 beta4 -- connection to sendmail hangs

[Todd C. Miller]

I'm seeing connections to port 25 on the local host hanging when
qrunner tries to do its thing.  Doing a syscall trace of qrunner
reveals that after it has written the message to sendmail, it gets:

  9182 python   RET   sendto 16384/0x4000
  9182 python   CALL  read(0x9,0xde000,0x4000)
  9182 python   RET   read -1 errno 35 Resource temporarily unavailable
  9182 python   CALL  gettimeofday(0xdfbfc694,0)
  9182 python   RET   gettimeofday 0
  9182 python   CALL  poll(0x61000,0x1,0)
  9182 python   RET   poll 0
  9182 python   CALL  gettimeofday(0xdfbfc624,0)
  9182 python   RET   gettimeofday 0
  9182 python   CALL  poll(0x61000,0x2,0xffffffff)

At this opint things hang indefinately.  Now, the EAGAIN it gets
from read(2) is a bit fishy.  The sendmail process does still
exist at this point:

root     24536  0.0  1.9   776  1204 ??  I      2:50PM    0:00.04 sendmail: server localhost [127.0.0.1] child wait (sendmail)

For the time being I've switched to using Sendmail instead of SMTPDirect
as the delivery module.  This is sendmail 8.10.1 BTW.

 - todd