Rspamd soft-reject emails with getMail (rowihei)

it is not under our control, is not on our resources, either you had a network issue, or rspamd from upstream got an undetermined issue

for what I read, your email has been well soft-rejected

it is officially a rspamd issue from a lets encrypt certificate

1 Like

I would like you try to increase the task_timeout = 8s; in /etc/rspamd/rspamd.conf

It is a template so any settings change in the filter panel will erase your change.

Lets say 15s

think to restart rspamd

cc @davidep

we have a timeout of 120s for getMail, in theory we should have the same

Which is why softreject should NEVER be applied against POP or IMAP access of mailboxes, only unsolicited mail received directlyat the server.


This is not related to greylist, but due to a task timeout

Exactly, how do I disable soft rejects, greylisting is disabled.

read my message please

Sorry, just confused. I will try the timeout suggested and respond in the other thread.
But, if this is not a ‘soft reject’, but why does it display as such in the rspamd web interface?

Well please stay in your thread…this is possibly another issue…your is at Rspamd soft reject email with getMail (wbilger)

A post was merged into an existing topic: Rspamd soft reject email with getMail (wbilger)

we thought of this effect at beginning - we let stay the mails 14 days on the smarthost

Today we set task_timeout = 15s and now let’s wait

1 Like

:thinking: 8 seconds…

This is the reject reason. However the rspamc-getmail wrapper is invoked with 120s timeout: where do the 8s come from? Our new clamav error handling?

This is our before.sieve filter talking. IIUC it wants to discard the message.

This is likely to be the timeout origin. During clamav db reloads rspamd waits a while.

Hypotesis: the timeout handling with getmail has got a regression bug and discards instead of retrying? Let’s try to reproduce @stephdl:

kill -STOP <pidofclamd>

Possible workaround: disable clamav official signatures to speed up db reloads.

not sure @davidep
in /etc/rspamd/rspamd.conf

# Timeout for messages processing (must be larger than any internal timeout used)
task_timeout = 8s;

When rspamd2.0 has been tested we have considered to increase it to 120s since it is our higher timeout we have, but it is LONG

1 Like

If the rspamc-getmail timeout argument is not honored, we can remove it.

Increasing the timeout is not a complete solution.

The temporary scan failure condition must be handled correctly, then we can discuss if 8s has to be increased.

I often see imap bandwithd around 200kB, I do not know why but imap is not so often fast.

For a text email I think it is enough, but for an email with an attachment it is of course longer

If I shoot in the night

10MB / 0,2MB/s = 50 seconds

Clearly the task_timeout is down

@rowihei does the email gets an attachment !

Yes, PDF attachment ~ 600 kByte

A post was merged into an existing topic: Rspamd soft reject email with getMail (wbilger)