Re: rspamd 3.2 core dumps

[Thomas Klausner <wiz%NetBSD.org@localhost>]

On Sun, Sep 04, 2022 at 09:22:36AM +0200, oskar%fessel.org@localhost wrote:
> since rspamd 3.2 i get the following on every connection:
> 2022-09-04 00:02:33 #6032(rspamd_proxy) <06b5f1>; proxy; rspamd_mime_parse_message: cannot find content-type for a message, assume text/plain
> 2022-09-04 00:02:33 #6032(rspamd_proxy) <06b5f1>; proxy; rspamd_message_parse: loaded message; id: <1662238946.719281.20880.nullmailer%spam.ping.de@localhost>; queue-id: <05DF43DB453>; size: 2573; checksum: <d326992218f97bb686821a98ddf17044>
> 2022-09-04 00:02:33 #6032(rspamd_proxy) <06b5f1>; proxy; rspamd_mime_part_detect_language: detected part language: en
> 2022-09-04 00:02:33 #6032(rspamd_proxy) rspamd_crash_sig_handler: caught fatal signal 11(Segmentation fault), pid: 6032, trace: 
> 2022-09-04 00:02:33 #6032(rspamd_proxy) rspamd_crash_sig_handler: please see Rspamd FAQ to learn how to dump core files and how to fill a bug report
> 2022-09-04 00:02:33 #11312(main) <f11e73>; main; rspamd_check_termination_clause: rspamd_proxy process 6032 terminated abnormally with exit code 0 by signal: Segmentation fault but NOT created core file (throttled=no); core file limits: 9223372036854775807 current, 9223372036854775807 max
> 2022-09-04 00:02:33 #11312(main) <f11e73>; main; rspamd_cld_handler: respawn process rspamd_proxy in lieu of terminated process with pid 6032
> 2022-09-04 00:02:35 #11312(main) <f11e73>; main; rspamd_fork_worker: prepare to fork process rspamd_proxy (0); listen on: localhost:11332
> 2022-09-04 00:02:35 #26348(rspamd_proxy) <f11e73>; main; rspamd_worker_set_limits: use system max file descriptors limit: 1024B cur and 3.32KiB max
> 2022-09-04 00:02:35 #26348(rspamd_proxy) <f11e73>; main; rspamd_worker_set_limits: use system max core size limit: 7.99EiB cur and 7.99EiB max
> 2022-09-04 00:02:35 #26348(rspamd_proxy) <f11e73>; main; rspamd_handle_child_fork: starting rspamd_proxy process 26348 (0); listen on: localhost:11332
> 
> going back to rspamd 3.1nb2, everything works as expected.
> I would really like to get a core file and file a bug report - but the FAQ looks quite evasive on that subject to me (i.e. i don’t have a clue why this does not dump core).  
> Anyone else seeing this?

Yes, I've seen this, on 9.99.99/amd64. That was the reason I've added
wip/rspamd31.

I've heard it works on Solaris, and one suspect is pcre2, but a) that
passes its selfs tests on my system, and b) it is not optional any
longer (as it was in older rspamd versions).

I haven't debugged this further.
 Thomas


> Could it be related to the ancient netbsd version - that has been very stable so far?
> NetBSD zappa.unna.ping.de 9.99.74 NetBSD 9.99.74 (ZAPPA) #0: Fri Oct 30 12:45:40 CET 2020  root%zappa.unna.ping.de@localhost:/moon/obj/sys/arch/amd64/compile/ZAPPA amd64
> 
> ZAPPA is GENERIC-$Revision: 1.589 with added options LOCKDEBUG, GATEWAY, IPSEC_DEBUG, MROUTING, PIM, ALTQ*, USBVERBOSE, RF_INCLUDE* and spdmem* as well as sdtemp* devices.
> 
> Cheers
> 	Oskar
>