Piler 1.35 on Ubuntu 18.04 - piler-smtp SIGSEGV

2021-02-09 Thread Wagner, Patrick




Hello everyone,

a few weeks ago 2 of our Piler instances began segfaulting - more 
precisely, several times per month (but not every day night) at about 
02:30 am local time the MS Exchange server that has the Piler instance 
setup as journaling target connects to piler-smtp, upon which piler-smtp 
segfaults after trying to store an email that clearly doesn't get parsed 
correctly by piler-smtp as it is represented as gibberish (and gibberish 
cannot be renamed...)


See below for a few of the latest examples. Interestingly, the SIGSEGV 
did not happen Sunday and Monday night, meaning the day following a 
non-weekday? Not going on long enough that I could call it a pattern, 
but in any case it would be nice if piler-smtp didn't segfault in the 
first place.


What can I do to supply more useful information to get this fixed? Or 
has there been meaningful work on piler-smtp in the meantime that this 
could / would've already been fixed?



Thanks, Patrick


Feb  4 02:30:01 hostname CRON[24428]: (piler) CMD (/usr/bin/find 
/var/www/piler/tmp -type f -name i.\* -exec rm -f {} \;)
Feb  4 02:30:01 hostname CRON[24430]: (piler) CMD (/usr/bin/find 
/var/piler/error -type f|wc -l > /var/piler/stat/error)
Feb  4 02:30:01 hostname CRON[24429]: (piler) CMD 
(/usr/local/libexec/piler/indexer.main.sh)
Feb  4 02:30:01 hostname CRON[24431]: (piler) CMD (/usr/bin/indexer 
--quiet note1 --rotate)
Feb  4 02:30:01 hostname CRON[24436]: (piler) CMD (/usr/bin/indexer 
--quiet tag1 --rotate)

Feb  4 02:30:01 hostname piler: INDEXER INFO: merging to main started
Feb  4 02:30:01 hostname CRON[24426]: (CRON) info (No MTA installed, 
discarding output)
Feb  4 02:30:01 hostname CRON[24425]: (CRON) info (No MTA installed, 
discarding output)
Feb  4 02:30:20 hostname piler-smtp[1144]: connected from 
192.168.165.252:58292 on fd=6 (active connections: 1)
Feb  4 02:30:20 hostname piler-smtp[1144]: connected from 
192.168.165.252:58294 on fd=9 (active connections: 2)
Feb  4 02:30:20 hostname piler-smtp[1144]: disconnected from 
192.168.165.252 on fd=6 (1 active connections)
Feb  4 02:30:20 hostname piler-smtp[1144]: disconnected from 
192.168.165.252 on fd=9 (0 active connections)
Feb  4 02:30:27 hostname piler-smtp[1144]: connected from 
192.168.165.154:41139 on fd=6 (active connections: 1)

Feb  4 02:32:18 hostname piler-smtp[1144]: client 192.168.165.154 timeout
Feb  4 02:32:18 hostname piler-smtp[1144]: disconnected from 
192.168.165.154 on fd=6 (0 active connections)

Feb  4 02:32:18 hostname piler-smtp[1144]: Removing 6KHX6ZMY7OD1R0KG
Feb  4 02:32:18 hostname piler-smtp[1144]: received: #013}^?, 
from=, size=4940, client=192.168.165.154, fd=6
Feb  4 02:32:18 hostname piler-smtp[1144]: ERROR: couldn't rename 
#013}^? to 2/ #013}^? (reason: no such file or directory)
Feb  4 02:32:18 hostname piler-smtp[1144]: got signal: 11, Segmentation 
fault

Feb  4 02:32:18 hostname piler-smtp[1144]: piler-smtp has been terminated

Feb  5 02:32:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:57818 on fd=6 (active connections: 1)
Feb  5 02:32:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:57820 on fd=9 (active connections: 2)
Feb  5 02:32:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:32:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=9 (0 active connections)
Feb  5 02:33:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:60458 on fd=6 (active connections: 1)
Feb  5 02:33:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:60460 on fd=9 (active connections: 2)
Feb  5 02:33:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:33:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=9 (0 active connections)
Feb  5 02:34:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:34840 on fd=6 (active connections: 1)
Feb  5 02:34:15 hostname piler-smtp[28329]: connected from 
192.168.165.252:34842 on fd=9 (active connections: 2)
Feb  5 02:34:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:34:15 hostname piler-smtp[28329]: disconnected from 
192.168.165.252 on fd=9 (0 active connections)
Feb  5 02:34:29 hostname piler-smtp[28329]: connected from 
192.168.165.154:25626 on fd=6 (active connections: 1)
Feb  5 02:35:01 hostname CRON[30058]: (piler) CMD (/usr/bin/find 
/var/piler/error -type f|wc -l > /var/piler/stat/error)
Feb  5 02:35:01 hostname CRON[30060]: (root) CMD (command -v debian-sa1 
> /dev/null && debian-sa1 1 1)
Feb  5 02:35:01 hostname CRON[30059]: (piler) CMD 
(/usr/local/libexec/piler/indexer.delta.sh)
Feb  5 02:35:01 hostname CRON[30063]: (piler) CMD (/usr/bin/find 
/var/www/piler/tmp -type f -name i.\* -exec rm -f {} \;)
Feb  5 02:35:01 hostname piler: INDEXER ERROR: indexer merging to main 
index is already running. It started at Fr 5. Feb 02:30:01 CET 

Re: Piler 1.35 on Ubuntu 18.04 - piler-smtp SIGSEGV

2021-02-09 Thread sj




Hello Patrick,

I've been reported such issues when a backup system paused or froze
the virtual machine running piler. I suspect that it might be the
case for you based on the time and days you mentioned.

The problem in such cases is that piler-smtp is in the middle of
processing an email, then the system is frozen for a few minutes
enough to pass the smtp timeout limit (60 sec), and then when it's
unfrozen then piler-smtp notices the timeout situation, then tears
down the connection. However, at the end of the session it tries
to tear down the already terminated and freed connection data.

So, let's try the following: get the master branch, recompile piler,
however replace the piler-smtp binary only, nothing else. Then let's
see if the problem is solved.

Janos

On 2021-02-09 07:56, Wagner, Patrick wrote:

Hello everyone,

a few weeks ago 2 of our Piler instances began segfaulting - more
precisely, several times per month (but not every day night) at about
02:30 am local time the MS Exchange server that has the Piler instance
setup as journaling target connects to piler-smtp, upon which
piler-smtp segfaults after trying to store an email that clearly
doesn't get parsed correctly by piler-smtp as it is represented as
gibberish (and gibberish cannot be renamed...)

See below for a few of the latest examples. Interestingly, the SIGSEGV
did not happen Sunday and Monday night, meaning the day following a
non-weekday? Not going on long enough that I could call it a pattern,
but in any case it would be nice if piler-smtp didn't segfault in the
first place.

What can I do to supply more useful information to get this fixed? Or
has there been meaningful work on piler-smtp in the meantime that this
could / would've already been fixed?


Thanks, Patrick


Feb  4 02:30:01 hostname CRON[24428]: (piler) CMD (/usr/bin/find
/var/www/piler/tmp -type f -name i.\* -exec rm -f {} \;)
Feb  4 02:30:01 hostname CRON[24430]: (piler) CMD (/usr/bin/find
/var/piler/error -type f|wc -l > /var/piler/stat/error)
Feb  4 02:30:01 hostname CRON[24429]: (piler) CMD
(/usr/local/libexec/piler/indexer.main.sh)
Feb  4 02:30:01 hostname CRON[24431]: (piler) CMD (/usr/bin/indexer
--quiet note1 --rotate)
Feb  4 02:30:01 hostname CRON[24436]: (piler) CMD (/usr/bin/indexer
--quiet tag1 --rotate)
Feb  4 02:30:01 hostname piler: INDEXER INFO: merging to main started
Feb  4 02:30:01 hostname CRON[24426]: (CRON) info (No MTA installed,
discarding output)
Feb  4 02:30:01 hostname CRON[24425]: (CRON) info (No MTA installed,
discarding output)
Feb  4 02:30:20 hostname piler-smtp[1144]: connected from
192.168.165.252:58292 on fd=6 (active connections: 1)
Feb  4 02:30:20 hostname piler-smtp[1144]: connected from
192.168.165.252:58294 on fd=9 (active connections: 2)
Feb  4 02:30:20 hostname piler-smtp[1144]: disconnected from
192.168.165.252 on fd=6 (1 active connections)
Feb  4 02:30:20 hostname piler-smtp[1144]: disconnected from
192.168.165.252 on fd=9 (0 active connections)
Feb  4 02:30:27 hostname piler-smtp[1144]: connected from
192.168.165.154:41139 on fd=6 (active connections: 1)
Feb  4 02:32:18 hostname piler-smtp[1144]: client 192.168.165.154 
timeout

Feb  4 02:32:18 hostname piler-smtp[1144]: disconnected from
192.168.165.154 on fd=6 (0 active connections)
Feb  4 02:32:18 hostname piler-smtp[1144]: Removing 6KHX6ZMY7OD1R0KG
Feb  4 02:32:18 hostname piler-smtp[1144]: received: #013}^?,
from=, size=4940, client=192.168.165.154, fd=6
Feb  4 02:32:18 hostname piler-smtp[1144]: ERROR: couldn't rename
#013}^? to 2/ #013}^? (reason: no such file or
directory)
Feb  4 02:32:18 hostname piler-smtp[1144]: got signal: 11, Segmentation 
fault
Feb  4 02:32:18 hostname piler-smtp[1144]: piler-smtp has been 
terminated


Feb  5 02:32:15 hostname piler-smtp[28329]: connected from
192.168.165.252:57818 on fd=6 (active connections: 1)
Feb  5 02:32:15 hostname piler-smtp[28329]: connected from
192.168.165.252:57820 on fd=9 (active connections: 2)
Feb  5 02:32:15 hostname piler-smtp[28329]: disconnected from
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:32:15 hostname piler-smtp[28329]: disconnected from
192.168.165.252 on fd=9 (0 active connections)
Feb  5 02:33:15 hostname piler-smtp[28329]: connected from
192.168.165.252:60458 on fd=6 (active connections: 1)
Feb  5 02:33:15 hostname piler-smtp[28329]: connected from
192.168.165.252:60460 on fd=9 (active connections: 2)
Feb  5 02:33:15 hostname piler-smtp[28329]: disconnected from
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:33:15 hostname piler-smtp[28329]: disconnected from
192.168.165.252 on fd=9 (0 active connections)
Feb  5 02:34:15 hostname piler-smtp[28329]: connected from
192.168.165.252:34840 on fd=6 (active connections: 1)
Feb  5 02:34:15 hostname piler-smtp[28329]: connected from
192.168.165.252:34842 on fd=9 (active connections: 2)
Feb  5 02:34:15 hostname piler-smtp[28329]: disconnected from
192.168.165.252 on fd=6 (1 active connections)
Feb  5 02:34:15 hostname