Rspamd unexpectedly stops working

[gibzer]

Classification (Please choose one option):

• Crash/Hang/Data loss
• WebUI/Usability
• Serious bug
• Other bug
• Feature
• Enhancement

Reproducibility (Please choose one option):

• Always
• Sometimes
• Rarely
• Unable
• I didn’t try
• Not applicable

Rspamd version:

1.6.4

Operation system, CPU, memory and environment:

Ubuntu 16.04

Description (Please provide a descriptive summary of the issue):

Rspamd sometimes unexpectedly stops working. Though all Rspamd processes work (main, rspamd_proxy, controller, normal, log_helper, hs_helper), no one is down. Rspamd log is clean like just no mails is coming. In Postfix log I have this:

Oct 18 10:29:33 mail postfix/smtpd[5768]: warning: milter inet:localhost:11332: can't read SMFIC_OPTNEG reply packet header: Connection timed out
Oct 18 10:29:33 mail postfix/smtpd[5768]: warning: milter inet:localhost:11332: read error in initial handshake
Oct 18 10:29:33 mail postfix/smtpd[5768]: NOQUEUE: milter-reject: CONNECT from mmail8.mail365.ru[83.222.116.101]: 451 4.7.1 Service unavailable - try again later; proto=SMTP
Oct 18 10:29:33 mail postfix/smtpd[5768]: NOQUEUE: milter-reject: EHLO from mmail8.mail365.ru[83.222.116.101]: 451 4.7.1 Service unavailable - try again later; proto=SMTP helo=<mmail8.mail365.ru>

This could happen once a month or twice a week, so I can't reveal the trends.

Compile errors (if any):

Steps to reproduce:

Expected results:

Actual results:

Debugging information (see details here):

Configuration:

Additional information:

[fenice2]

This sounds like it might be a problem with the MTA not rspamd. Can you check if the milter protocol is set correctly to "6"? You can check the protocol version with "postconf milter_protocol". Which MTA are you using?

[gibzer]

I'm not sure that problem is because of MTA. I use Postfix. When this happens, if I restart Rspamd - everything starts to work. So no need to restart MTA.

smtpd_milters = inet:localhost:11332, inet:localhost:54000
non_smtpd_milters = inet:localhost:11332, inet:localhost:54000
milter_protocol = 6
milter_mail_macros = i {mail_addr} {client_addr} {client_name} {auth_authen}
milter_default_action = tempfail

[fenice2]

I see you have multiple milters in there, what's the second one for? Is that the one that might be failing when mail flow stops? I believe the "milter_default_action" applies to both milters unless you specific different actions for each milter. Have you tried setting "milter_default_action" to accept to see if that allows mail to flow?

[gibzer]

The second is AVG antivirus milter. Log shows that this is exactly Rspamd. And after Rspamd restart error is gone. I tried milter_default_action = accept, mail flows, but without Rspamd check.

[gibzer]

Problem still persists. Got 2 crashes within 3 days. Rspamd logs are clear, all processes are working. I have to use Monit to check Postfix logs for Rspamd errors and restart it...

[jrampin]

@gibzer are you still getting this issue? Have you found a workaround? I've got this once two weeks ago, and once 2 days ago...

[gibzer]

Yes, I still have it. The only one workaround is to use Monit to restart Rspamd when Postfix says in log that it can't connect to Rspamd.
Monit config:

check file maillog with path /var/log/mail.log
    if match "warning: milter inet:localhost:11332: can't read SMFIC_OPTNEG reply packet header: Connection timed out" then exec "/usr/sbin/service rspamd restart"

[jrampin]

@gibzer I've done 2 things:

1. Increase the memory up to 16GB of RAM
2. Added both parameters on redis.conf file

---

/etc/redis/redis.conf

---

maxmemory 1024mb
maxmemory-policy volatile-lru

It's been a while since I don't get that error anymore. Both servers reach up to 10GB of used memory, and then it releases memory after a while.

Hopefully it can help you! :)