[Dovecot] RC6 startup failure: "Login prcess died too early"
Greetings.
I have been running Dovecot-1.0rc2 on a test server for the last couple of weeks. No problems. However, when I installed RC6 this morning, Dovecot refused to launch completely. Here's the relevant snippet from /var/log/maillog:
Aug 7 10:08:36 mercury dovecot: Dovecot v1.0.rc6 starting up Aug 7 10:08:37 mercury dovecot: Login process died too early - shutting down Aug 7 10:08:37 mercury dovecot: child 26127 (login) killed with signal 11 Aug 7 10:08:37 mercury dovecot: child 26128 (login) killed with signal 11
Both RC2 and RC6 compiled cleanly with "--with-pop3d". For reasons beyond my control, I have to set
protocols = imap pop3 disable_plaintext_auth = no ssl_disable = yes
I do have a valid /etc/pam.d/dovecot file (we're using PAM). I made certain that the localized settings for /usr/local/etc/dovecot.conf matched what they were in RC2 for RC6. (I did use the RC6 dovecot-example.conf as source for RC6, then made the local changes. I also tried testing RC6 with my old RC2 config file--same failure).
So, RC2 starts just fine. RC6 fails to start with the messages logged
above. Can anyone give me any pointers? RHEL (Linux Kernel
2.6.9-34.0.2.ELsmp), built from source, nothing out-of-the-ordinary.
Right now, I've reverted back to RC2.
TIA. S JList sjlist@heidelberg.edu
P.S. More info on my earlier post...I had gone directly from RC2 -> RC6. So I wondered if the problem had been introduced sometime between the two. I just now compiled RC5, and it does -not- exhibit the problem below for me. So it seems to be something new in RC6 given my setup.
Thanks,
S JList
S JList wrote:
Greetings.
I have been running Dovecot-1.0rc2 on a test server for the last couple of weeks. No problems. However, when I installed RC6 this morning, Dovecot refused to launch completely. Here's the relevant snippet from /var/log/maillog:
Aug 7 10:08:36 mercury dovecot: Dovecot v1.0.rc6 starting up Aug 7 10:08:37 mercury dovecot: Login process died too early - shutting down Aug 7 10:08:37 mercury dovecot: child 26127 (login) killed with signal 11 Aug 7 10:08:37 mercury dovecot: child 26128 (login) killed with signal 11
Both RC2 and RC6 compiled cleanly with "--with-pop3d". For reasons beyond my control, I have to set
protocols = imap pop3 disable_plaintext_auth = no ssl_disable = yes
I do have a valid /etc/pam.d/dovecot file (we're using PAM). I made certain that the localized settings for /usr/local/etc/dovecot.conf matched what they were in RC2 for RC6. (I did use the RC6 dovecot-example.conf as source for RC6, then made the local changes.
I also tried testing RC6 with my old RC2 config file--same failure).So, RC2 starts just fine. RC6 fails to start with the messages logged above. Can anyone give me any pointers? RHEL (Linux Kernel 2.6.9-34.0.2.ELsmp), built from source, nothing out-of-the-ordinary.
Right now, I've reverted back to RC2.TIA. S JList sjlist@heidelberg.edu
On 7.8.2006, at 17.25, S JList wrote:
I have been running Dovecot-1.0rc2 on a test server for the last
couple of weeks. No problems. However, when I installed RC6 this
morning, Dovecot refused to launch completely. Here's the relevant
snippet from /var/log/maillog:
I guess this fixes it? http://dovecot.org/list/dovecot-cvs/2006- August/006209.html
Timo Sirainen wrote:
On 7.8.2006, at 17.25, S JList wrote:
I have been running Dovecot-1.0rc2 on a test server for the last couple of weeks. No problems. However, when I installed RC6 this morning, Dovecot refused to launch completely. Here's the relevant snippet from /var/log/maillog:
I guess this fixes it? http://dovecot.org/list/dovecot-cvs/2006-August/006209.html
Timo,
Yes--this does seem to fix it. I modified my copy of src/login-common/ssl-proxy-openssl from the RC6 source files, made the change you noted, and it seems to have fixed my problem (crashing on startup if SSL was disabled).
Thanks so much for all of your hard work and this quick reply. I'm running with this change in my RC6 now, and (I'm not a CVS user, nor am I very familiar with it) that this change will be rolled into RC7, then, right?
Thanks again for your response--much appreciated.
S JList sjlist@heidelberg.edu
participants (2)
-
S JList
-
Timo Sirainen