0

Edit

Ok, so if you care to read the original question, you'll see I've had a fun time with this. I'm running Ubuntu 13.10 - Dovecot v2.1.7 -

And it seems my original problem was due to some misconfiguration - so, I re-configured this and that in /etc/dovecot/ - to no avail ~

So, today, I noticed that the output of doveconf does not reflect my changes no matter what I do - 

service dovecot reload
dovecot reload
doveconf reload
  • How do I get it to read the load the .conf's in /etc/dovecot/ -and probably more importantly -- the .conf's in /etc/dovecot/conf.d/ -?

Original Question:

Our small business moved our email service to an un-managed server, recently - and it works perfectly - !

Except about three days this week, when the imaps port closes, and no one can receive email.

The first time, simply reloading dovecot did the trick - the second time, it worked for a split second and stopped... and ever since then, the only way I can figure out how to get it working again is to restart the whole server.

After Googleing around a bit, I added login_process_size = 64 to dovecot.conf - restarted dovecot, but this morning, I found that it had crashed again. Also, I noticed an error that said that setting was being overridden, but I can't find where.

mail.err gives this: 

Apr 11 09:27:50 Mary dovecot: imap-login: Error: read(anvil) failed: EOF
Apr 11 09:27:50  dovecot: last message repeated 11 times
Apr 11 09:27:50 Mary dovecot: auth: Error: read(anvil-auth-penalty) failed: EOF
Apr 11 09:27:50 Mary dovecot: imap-login: Error: read(anvil) failed: EOF
Apr 11 09:27:55 Mary dovecot: lda: Error: userdb lookup(user@website.com): Disconnected unexpectedly
Apr 11 09:27:55 Mary dovecot: lda: Fatal: Internal error occurred. Refer to server log for more information.

I'm not sure what would be relevant in doveconf - but, one thread mentioned setting a value that is not present in mine -

Also - I've set auth_max_worker_count to 60, just now - and I'm hoping that'll work ---

But if anyone can clue me in on what I need to tweak, what I should look at, or how I should adjust 10-logging.conf to give me more clues on this matter, I would greatly appreciate it..

As per NickW's comment, the output of cat /proc/{dovecot/anvil pit}/limits is: 

Limit                     Soft Limit           Hard Limit           Units     
Max cpu time              unlimited            unlimited            seconds   
Max file size             unlimited            unlimited            bytes     
Max data size             268435456            268435456            bytes     
Max stack size            8388608              unlimited            bytes     
Max core file size        0                    unlimited            bytes     
Max resident set          unlimited            unlimited            bytes     
Max processes             63651                63651                processes 
Max open files            1024                 4096                 files     
Max locked memory         65536                65536                bytes     
Max address space         268435456            268435456            bytes     
Max file locks            unlimited            unlimited            locks     
Max pending signals       63651                63651                signals   
Max msgqueue size         819200               819200               bytes     
Max nice priority         0                    0                    
Max realtime priority     0                    0                    
Max realtime timeout      unlimited            unlimited            us

Now, There are about 15 people connecting - most of which also connect via their cell phone - Those limits look fine to me, but I'm relatively inexperienced in this..

rm-vanda
  • 257
  • 8
  • 18
  • 1
    What do you get back from this command? `cat /proc//limits` ? – NickW Apr 11 '14 at 14:27
  • Well, `ps -ef` lists two processes for anvil - assuming dovecot/anvil - I will edit the question accordingly - – rm-vanda Apr 11 '14 at 14:56
  • 1
    What is your `default_client_limit` set to, and `default_process_limit`? – NickW Apr 11 '14 at 15:17
  • 1000 and 100, respectively --- I ran across something that said "dovecot spawning multiple processes per user" -- when I was researching this - perhaps that is the case? - thanks for your assistance, btw - – rm-vanda Apr 11 '14 at 15:41
  • 1
    That could be the case, do you have any idea about how many connections you have going before the problem shows up? – NickW Apr 11 '14 at 16:08
  • 1
    Also, what version of dovecot, and what OS are you on? – NickW Apr 11 '14 at 16:10
  • No, I don't, but it shouldn't be anymore than 50, at the very most (~40, realistically). - Ubuntu 13.10 - Dovecot 2.1.7 - – rm-vanda Apr 11 '14 at 17:02
  • 1
    You may want to ramp up the log level, add this into your dovecot.conf: `debug_log_path = path/to/debug/log` – NickW Apr 14 '14 at 08:32
  • I ramped up the logging, but its not actually outputting anything at all - any other ideas on what I could do? – rm-vanda Apr 15 '14 at 18:14
  • `syslog` doesn't report any errors, either -- – rm-vanda Apr 15 '14 at 18:14
  • Oh, hey, I noticed that `doveconf` doesn't report the same thing that I've put in my config file - How can I force `doveconf` to read from the `.conf` files in `/etc/dovecot/*` -? – rm-vanda Apr 15 '14 at 18:25
  • The answer was that dovecot.conf was missing the `!include ~` statement. So, if you make an answer, that mentions something about checking the output of doveconf - I'll happily award you the points. – rm-vanda Apr 15 '14 at 21:07
  • Sounds like a plan to me :) – NickW Apr 16 '14 at 08:15

1 Answers1

1

I'm not exactly sure what the problem is, but you can obtain more information about what is going wrong by enabling the debug log debug_log_path = path/to/debug/log and by ensuring that the changes you've made to the configuration are actually seen by dovecot. The easiest way to verify what dovecot is actually using at runtime is the doveconf command, dovecot -a will show you these values.

NickW
  • 10,263
  • 1
  • 20
  • 27