The mailing list for listmasters using Sympa

[Tom Fillmore <address@concealed>]

Hi -

Wow, that is interesting...

Based on what I see it seems things are being processed in reverse order, or perhaps the whole process should be nested.

To me the next step is to maximize library verbocity, then walk through the perl code.  I'm no perl guy, so I cannot help you there.

A couple quick things to try though that I've found helpful in a very few circumstances are:

• Restart your processes (NOT reload) in this order:

• Amavis
• Clamav (or whatever av you use, if any)
• spamassassin
• Opendkim
• Postfix
• sympa

Test run mhonarc manually, confirm no errors (more below)
Copy the list in question to a new list, remove verp and see if you get the same results  (I don't use verp but copying a list has sometimes proved helpful)
After doing this check the logs to see if anything wonky shows up for any of the above. 


Rationale:
I just replaced a 6.2.13 server that crashed with the latest-and-greatest 6.2.42, built from source.  Comparatively speaking it's much easier to install but as always there are a few issues that could be show stoppers if you didn't know how to handle them.  I'll list a couple here because I noticed some _process_ AND _lib_ dependency things that popped up.  They could be part of the problems you are experiencing but are not throwing errors or log messages.

Amavis - monitor cpu usage and decide if that is causing things to hang or terminate.  For a while on the older machine I just replaced I would just stop the service if usage stayed too high.  I never did find what the cause of it was (though I suspected memory leaks) and it seems to have gone away with the new install.  Still, it caused problems in other areas.

Spamassassin - probably a version issue that has been fixed, but on the older machine it would fail as a service but all other processes would not always see that.  I noticed opendkim was sensitive to that

Opendkim - wonderful service, is invaluable especially to the major carriers.  Because it is security-related it's hyper-sensitive to _everything_ and can really muck up things if it does not see what it expects to see in the way and order it wants to see it.  For example - as a matter of course, if you have to restart any services in the sympa universe (eg spamassassin), ALWAYS RESTART OPENDKIM AND RESTART IT JUST BEFORE RESTARTING POSTFIX.  It will not always tell you it is having problems if you don't restart it, but will if you do.  That extra five seconds is worth its weight in gold.

Opendkim - following the exact installation steps of opendkim requires it to have sole access to the private key.  It will not allow additional members to be added to it's group and will not sign messages if there is, so your two options there are to turn off DKIM signing per list and let the mta decide, or copy the private keys file to a sympa-only folder and leave DKIM signing turned on in your lists.  I've tried the first option and noticed that clicking the distribute link vs the web ui distribution link produces different signing results.  Because of that I'm going to make a sympa-only copy of the private ley (...and now, every single security person is feeling a bad disturbance in 'the force' for that comment...  :) ).

Mhonarc - ugh.  Nasty, nasty, nasty.  Does what it says it will but is slow and takes more memory than it seems it should.  The version installed by cpan had a bug in it not detected by the sympa wizard file and sympa would start but fail only the message service.  Fortunately the error thrown pointed out the simple solution.  Making that change fixed everything.

About dependencies - - I tried to let cpanm handle the heavy lifting for that but there are _so many_ new installs and lib version dependencies that I could never get it to complete successfully.   Basically, it failed at it's (one) job.  So, I cheated.  I removed all files created by the source install, then installed sympa using apt-get, then removed sympa but left all the perl libs.  Rerunning cpanm finished successfully with only one add, and sympa started and ran as expected.  FYI I'm using Devuan ASCII.  I'm not a systemd guy.  Love Debian, but cut my teeth on Sllackware.

Anyway, thanks for reading this far.  This turned out to be longer than I'd thought, but I truly hope anything above helps get you further towards a solution.

Thanks

Tom Fillmore
southern California

On 04/27/2019 02:28, dee heffem wrote:

On 4/26/19 5:05 PM, Fillmore, Tom wrote:

Hi -

What shows in the logs?

Tom Fillmore

I see some entries like this related to distribution:

Apr 26 16:36:26 lists sympa[4252]: err List::send_msg() List::send_msg,
could not send message to distribute from
address@concealed (verp desabled)



Other log entries like this but not sure they are errors:

Apr 25 16:16:49 lists sympa[25317]: notice List::send_msg() No VERP
subscribers left to distribute message from
address@concealed to list location3.group3


Below is a part of the logfile where a moderator approved a waiting
message (I've santized org info) and sympa sent "QUIET DISTRIBUTE..."
message which went to the upstream relay without distributing the
approved message. Is the do_modindex "no message and no document" telling?

Thank you



Apr 26 16:24:55 lists wwsympa[4340]: info [robot domain.net] [session
40723086638312] [client 192.168.33.244] [user address@concealed]
[list its] main::do_distribute()
do_distribute(01bb3865684be85bbe81a4810d9fab34)
Apr 26 16:24:55 lists wwsympa[4340]: notice main::do_distribute()
Moderator approves message: its 01bb3865684be85bbe81a4810d9fab34
Apr 26 16:24:55 lists wwsympa[4340]: info [robot domain.net] [session
40723086638312] [client 192.168.33.244] [user address@concealed]
[list its] main::do_modindex() do_modindex
Apr 26 16:24:55 lists wwsympa[4340]: err [robot domain.net] [session
40723086638312] [client 192.168.33.244] [user address@concealed]
[list its] main::do_modindex() do_modindex: no message and no document
Apr 26 16:24:57 lists sympa[4252]: notice main::DoFile() Processing
/address@concealed ; sender:
address@concealed#012 ; message-id:
<address@concealed>
Apr 26 16:24:57 lists sympa[4252]: info main::DoSendMessage() Processing
web message for address@concealed
Apr 26 16:24:58 lists sympa[4252]: info main::DoSendMessage() Message
for address@concealed sent
Apr 26 16:24:58 lists postfix/pickup[5185]: 41599147: uid=115
from=<address@concealed>
Apr 26 16:24:58 lists postfix/cleanup[6138]: 41599147:
message-id=<address@concealed>
Apr 26 16:24:58 lists postfix/qmgr[3681]: 41599147:
from=<address@concealed>, size=465, nrcpt=1 (queue active)
Apr 26 16:24:58 lists postfix/smtp[6141]: 41599147:
to=<address@concealed>, relay=cliff.domain.net[10.22.1.6]:2525,
delay=0.34, delays=0.3/0.01/0.03/0.01, dsn=2.0.0, status=sent (250 2.0.0
Ok: queued as 37111129)
Apr 26 16:24:58 lists postfix/qmgr[3681]: 41599147: removed