dovecot/pigeonhole and sogo

[Thomas Robers]

Hello,

I have two systems one with dovecot 2.2.31/Pigeonhole 0.4.19 selfcompiled
through the src.rpms from ghettoforge on a CentOS 6.9 server and a second
system of SOGo (3.2.9, nightly) on CentOS 7.
Everything worked together until I made the update to Dovecot 2.2.31
Pigeonhole 0.4.19. Since then I get an error message when I
try to upload a sieve-script. SOGo shows a message popup that "the service
is temporarily not available" and in the log file of Dovecot i get:

> Jul 12 15:26:41 mail2 dovecot: managesieve-login: Disconnected (no auth attempts in 0 secs): user=<>, method=PLAIN, rip=192.168.1.138, lip=192.168.1.135, TLS: Disconnected, session=<25U/xB5U/KLCw8GK>
> Jul 12 15:26:41 mail2 dovecot: auth: Debug: auth client connected (pid=12182)
> Jul 12 15:26:41 mail2 dovecot: managesieve-login: Disconnected (no auth attempts in 0 secs): user=<>, method=PLAIN, rip=192.168.1.138, lip=192.168.1.135, TLS: Disconnected, session=<4zxBxB5U/qLCw8GK>
> Jul 12 15:34:35 mail2 dovecot: auth: Debug: auth client connected (pid=12263)
> Jul 12 15:34:35 mail2 dovecot: managesieve-login: Disconnected (no auth attempts in 0 secs): user=<>, method=PLAIN, rip=192.168.1.138, lip=192.168.1.135, TLS: Disconnected, session=<uiR64B5UDKPCw8GK>
> Jul 12 15:34:35 mail2 dovecot: auth: Debug: auth client connected (pid=12264)
> Jul 12 15:34:35 mail2 dovecot: managesieve-login: Disconnected (no auth attempts in 0 secs): user=<>, method=PLAIN, rip=192.168.1.138, lip=1192.168.1.135, TLS: Disconnected, session=<LeJ74B5UDqPCw8GK>

Does this mean that managesieve-login doesn't get authentication data from SOGo?

SOGo logs for this error:

> S[0x7f368f2bab60]: "IMPLEMENTATION" "Dovecot Pigeonhole"
> S[0x7f368f2bab60]: "SIEVE" "fileinto reject envelope encoded-character vacation subaddress comparator-i;ascii-numeric relational regex imap4flags copy include variables body enotify environment mailbox date index ihave duplicate mime foreverypart extracttext"
> S[0x7f368f2bab60]: "NOTIFY" "mailto"
> S[0x7f368f2bab60]: "SASL" "PLAIN"
> S[0x7f368f2bab60]: "VERSION" "1.0"
> S[0x7f368f2bab60]: OK "Dovecot ready."
> C: AUTHENTICATE "PLAIN" {%d+}
> LOGIN:PASSWORD
> 
> S[0x7f368f2bab60]: NO "Error in MANAGESIEVE command received by server."
> C: AUTHENTICATE "PLAIN" {%d+}
> LOGIN:PASSWORD




Than I compiled Dovecot 2.2.31 together with Pigeonhole 0.4.18 and than everything
works again as expected. I can upload and activate sieve-scripts from the SOGo webinterface.

After that I tried to disable TLS for sieve with ssl = no in the 20-managesieve.conf:

[...]
protocol sieve {
    ssl = no
[...]

to capture the traffic between SOGo and Dovecot/Pigeonhole but then the managesieve-login process
crashes:

> Jul 12 16:28:32 mail2 dovecot: managesieve-login: Fatal: master: service(managesieve-login): child 12941 killed with signal 11 (core dumps disabled) [last ip=192.168.1.138]

Why isn't it possible to disable TLS for sieve and what changes between Pigeonhole 0.4.19 and 0.4.18
could be responsible for the disharmony between SOGo and Pigeonhole. I looked for 'changes' but
could not find any hints regarding this problem.

Is there a chance to get both work together again? And is it possible to disable TLS for sieve or
why crashes the managesieve-login daemon?

Any help would be welcome.

TIA
Thomas Robers