dovecot

dovecot@dovecot.org

September 2020

77 participants
102 discussions

Re: replicator: Panic: data stack: Out of memory
by Thomas Tsiakalakis 01 Sep '20

01 Sep '20

So nobody has any idea why this could happen? Let me know if I should provide more Information Thanks Thomas Tsiakalakis Team Applikationsbetrieb GDV Dienstleistungs-GmbH Tel: +49(40)33449-4318 Fax: E-Mail: Thomas.Tsiakalakis(a)gdv-dl.de<mailto:Thomas.Tsiakalakis@gdv-dl.de> GDV Dienstleistungs-GmbH Glockengießerwall 1 D-20095 Hamburg www.gdv-dl.de<http://www.gdv-dl.de> Niederlassungen: Wilhelmstraße 43 / 43 G 10117 Berlin Frankenstraße 18 20097 Hamburg Sitz und Registergericht: Hamburg HRB 145291 USt.-IdNr : DE 205183123 Geschäftsführer: Dr. Jens Bartenwerfer Fred di Giuseppe Chiachiarella Aufsichtsratsvorsitzender: Werner Schmidt ------------------------------------------------------------------ Diese E-Mail und alle Anhänge enthalten vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe der E-Mail ist nicht gestattet. This e-mail and any attached files may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorised copying, disclosure or distribution of the material in this e-mail is strictly forbidden. Von: Thomas Tsiakalakis Gesendet: Dienstag, 23. Juni 2020 11:25 An: dovecot(a)dovecot.org Betreff: AW: replicator: Panic: data stack: Out of memory I managed to convince the System that I really want a core dump Von: Thomas Tsiakalakis Gesendet: Donnerstag, 18. Juni 2020 16:26 An: 'dovecot(a)dovecot.org' <dovecot(a)dovecot.org> Betreff: replicator: Panic: data stack: Out of memory Hi, I have set up 2 SLES 15 Hosts with dovecot and replication. Everything seems to work fine but every time a message is delivered, I get an out of memory error in the logs. The Replication itself seems to work fine though. I increased default_vsz_limit to 512M but the only thing that changed was that dovecot was trying to allocate 1073741864 bytes instead of 268435496 As I said, I’m running SLES 15 SP1 and Dovecot 2.3.10 (0da0eff44) (both Hosts have the same version) Each Host currently has 8GB of Memory. # free -h total used free shared buff/cache available Mem: 7.8Gi 278Mi 7.3Gi 47Mi 183Mi 7.2Gi Swap: 4.0Gi 0B 4.0Gi # journalctl -f Jun 18 15:55:48 mail1 postfix/pickup[3457]: 18533C009C8: uid=0 from=<root> Jun 18 15:55:48 mail1 postfix/cleanup[3669]: 18533C009C8: message-id=<20200618135548.18533C009C8@mail1> Jun 18 15:55:48 mail1 postfix/qmgr[3458]: 18533C009C8: from=<root@mail1>, size=431, nrcpt=1 (queue active) Jun 18 15:55:48 mail1 dovecot[1833]: lmtp(3673): Connect from local Jun 18 15:55:48 mail1 dovecot[1833]: replicator: Panic: data stack: Out of memory when allocating 268435496 bytes Jun 18 15:55:48 mail1 dovecot[1833]: replicator: Error: Raw backtrace: /usr/lib64/dovecot/libdovecot.so.0(backtrace_append+0x42) [0x7f8346d6d262] -> /usr/lib64/dovecot/libdovecot.so.0(backtrace_get+0x1e) [0x7f8346d6d37e] -> /usr/lib64/dovecot/libdovecot.so.0(+0xebcee) [0x7f8346d77cee] -> /usr/lib64/dovecot/libdovecot.so.0(+0xebd91) [0x7f8346d77d91] -> /usr/lib64/dovecot/libdovecot.so.0(i_fatal+0) [0x7f8346cd1f8b] -> /usr/lib64/dovecot/libdovecot.so.0(t_pop_last_unsafe+0) [0x7f8346d73660] -> /usr/lib64/dovecot/libdovecot.so.0(+0xe7a50) [0x7f8346d73a50] -> /usr/lib64/dovecot/libdovecot.so.0(+0x10eb58) [0x7f8346d9ab58] -> /usr/lib64/dovecot/libdovecot.so.0(+0xe3ac5) [0x7f8346d6fac5] -> /usr/lib64/dovecot/libdovecot.so.0(buffer_write+0xf2) [0x7f8346d6fdb2] -> dovecot/replicator(replicator_queue_push+0xde) [0x5648989a5cde] -> dovecot/replicator(+0x5464) [0x5648989a5464] -> dovecot/replicator(+0x4a0b) [0x5648989a4a0b] -> dovecot/replicator(+0x4c02) [0x5648989a4c02] -> /usr/lib64/dovecot/libdovecot.so.0(io_loop_call_io+0x69) [0x7f8346d90c99] -> /usr/lib64/dovecot/libdovecot.so.0(io_loop_handler_run_internal+0x134) [0x7f8346d92574] -> /usr/lib64/dovecot/libdovecot.so.0(io_loop_handler_run+0x4c) [0x7f8346d90d9c] -> /usr/lib64/dovecot/libdovecot.so.0(io_loop_run+0x38) [0x7f8346d90fc8] -> /usr/lib64/dovecot/libdovecot.so.0(master_service_run+0x13) [0x7f8346d01873] -> dovecot/replicator(main+0x1a0) [0x5648989a3cb0] -> /lib64/libc.so.6(__libc_start_main+0xea) [0x7f83468f534a] -> dovecot/replicator(_start+0x2a) [0x5648989a3d5a] Jun 18 15:55:48 mail1 dovecot[1833]: lmtp(test(a)example.com)<3673><q5sLCGRy615ZDgAATn/NZw>: Warning: replication(test(a)example.com): Sync failure: Jun 18 15:55:48 mail1 dovecot[1833]: lmtp(test(a)example.com)<3673><q5sLCGRy615ZDgAATn/NZw>: Warning: replication(test(a)example.com): Remote sent invalid input: - Jun 18 15:55:48 mail1 dovecot[1833]: lmtp(test(a)example.com)<3673><q5sLCGRy615ZDgAATn/NZw>: msgid=<20200618135548.18533C009C8@mail1>: saved mail to INBOX Jun 18 15:55:48 mail1 dovecot[1833]: lmtp(3673): Disconnect from local: Client has quit the connection (state=READY) Jun 18 15:55:48 mail1 postfix/lmtp[3672]: 18533C009C8: to=<test(a)example.com>, orig_to=<test(a)domain.com>, relay=mail1[private/dovecot-lmtp], delay=0.17, delays=0.02/0.01/0.02/0.13, dsn=2.0.0, status=sent (250 2.0.0 <test(a)example.com> q5sLCGRy615ZDgAATn/NZw Saved) Jun 18 15:55:48 mail1 postfix/qmgr[3458]: 18533C009C8: removed Jun 18 15:55:48 mail1 dovecot[1833]: replicator: Fatal: master: service(replicator): child 2948 killed with signal 6 (core not dumped - https://dovecot.org/bugreport.html#coredumps - set /proc/sys/fs/suid_dumpable to 2)

2 3

shutdown_clients ignored on Director upgrade/restart
by Tom Sommer 01 Sep '20

01 Sep '20

When yum updates a Dovecot Director, it restarts all processes - however a few imap-login and pop3-login processes hang because they are still proxying connections. It's like the now orphaned processes are stalled, and needs a kill -9 to die. This is a problem because those stalled connections are still connected to backends, so when new connections are made from the same account, they do not end up on the same backend and thus index corruption happens (same account, two different backends). shutdown_clients is enabled but it appears Dovecot restarts regardless of verifying all processes/clients are shut down. Just had this happen upon yum update from dovecot-2.3.11.3-3 to dovecot-2.3.11.3-4 -- Tom

1 0