sieve issue after upgrad

[Tim Dickson]

On 20/01/2019 11:29, Stephan Bosch wrote:
>
>
> Op 20/01/2019 om 00:21 schreef Tim Dickson via dovecot:
>> On 19/01/2019 17:54, Stephan Bosch wrote:
>>> relevant info from /var/log/dovecot.log (domain changed for privacy)
>>>>
>>>> Jan 19 09:27:09 lda(tim): Debug: Loading modules from directory: 
>>>> /usr/lib64/dovecot
>>>> Jan 19 09:27:09 lda(tim): Debug: Module loaded: 
>>>> /usr/lib64/dovecot/lib90_sieve_plugin.so
>>>> Jan 19 09:27:09 lda(tim): Debug: auth USER input: tim 
>>>> system_groups_user=tim uid=1003 gid=100 home=/home/tim
>>>> Jan 19 09:27:09 lda(tim): Debug: Effective uid=1003, gid=100, 
>>>> home=/home/tim
>>>> Jan 19 09:27:09 lda(tim): Debug: maildir++: 
>>>> root=/var/spool/dovecot/tim, index=, indexpvt=, control=, 
>>>> inbox=/var/spool/dovecot/tim, alt=
>>>> Jan 19 09:27:09 lda(tim): Debug: none: root=, index=, indexpvt=, 
>>>> control=, inbox=, alt=
>>>> Jan 19 09:27:09 lda(tim): Debug: Destination address: 
>>>> <tim at mail.myserver.net> (source: user at hostname)
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: Pigeonhole version 0.5.1 
>>>> (d9bc6dfe) initializing
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: include: sieve_global is 
>>>> not set; it is currently not possible to include `:global' scripts.
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: file storage: Using active 
>>>> Sieve script path: /home/tim/.dovecot.sieve
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: file storage: Using script 
>>>> storage path: /home/tim/sieve
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: file storage: Relative path 
>>>> to sieve storage in active link: sieve/
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: file storage: Using Sieve 
>>>> script path: /home/tim/.dovecot.sieve
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: file script: Opened script 
>>>> `test' from `/home/tim/.dovecot.sieve'
>>>> Jan 19 09:27:09 lda(tim): Debug: sieve: Using the following 
>>>> location for user's Sieve script: /home/tim/.dovecot.sieve
>>>
>>> Here, all Sieve involvement abruptly seems to stop, which is very 
>>> very weird.
>>>
>>> I'd expect this to end in some sort of error message. Is there a 
>>> /var/log/dovecot.error or something like that? You can verify where 
>>> things are logged using `sudo doveadm log find`.
>>>
>>>
>> output of doveadm log find
>> Looking for log files from /var/log
>> Debug: /var/log/dovecot.log
>> Info: /var/log/dovecot.info
>> Warning: /var/log/maillog
>> Error: /var/log/maillog
>> Fatal: /var/log/maillog
>>
>> I have listed below the section of /var/log/maillog which appears to 
>> be relevant . (domain changed for privacy)
>>
>> Jan 18 23:13:46 mail dovecot: lda(tim): Error: sieve: Failed to 
>> initialize script execution: Invalid postmaster_address: invalid 
>> address `postmaster@' specified for the postmaster_address setting
>> Jan 18 23:13:46 mail sm-mta[866]: x0INDhct000865: 
>> to=<tim at myserver.net>, ctladdr=<heather at myserver.net> (1002/100), 
>> delay=00:00:00, xdelay=00:00:00, mailer=local, pri=33291, dsn=2.0.0, 
>> stat=Sent
>> Jan 18 23:21:14 mail dovecot: master: Warning: Killed with signal 15 
>> (by pid=905 uid=0 code=kill)
>>
>> There are other mails with the same "invalid address `postmaster@' to 
>> other users.
>> Is a postmaster address is required for sieve to work.? if so where 
>> should it go. is it the sieve_user_email declaration ?
>> thanks.
>
> Aha! So, you tripped over a 2.3.1 bug. The best way to work around 
> this is by setting a proper explicit value for the global 
> postmaster_address setting (https://wiki.dovecot.org/LDA). So, 
> Pigeonhole does not define this setting; it is defined by Dovecot 
> LDA/LMTP.
>
> Regards,
>
> Stephan.
>
Brilliant. that fixed it. the scripts are running again, and my test, 
and out of office is working. So it is just that there is no default for 
that field, and the example comments need to be updated.  (i just 
checked the 2.3.4 source and the example 15-lda.conf still implies a 
default, or has the behaviour been fixed in newer versions so it works?) 
I'll let the slackware build maintainer know so it can be added to the 
readme for 2.3.1 version.

many thanks everyone for your assistance.
regards, Tim