Error on IO with server fuzzy1.rspamd.com

Since about a week I have additional fuzzy errors on rspamd besides the already existing map errors. This error might be triggered through marking some emails in Thunderbird as spam, but not sure.

These errors are occurring:

<5d76e0>; csession; fuzzy_check_io_callback: got error on IO with server fuzzy1.rspamd.com:11335(127.0.0.1:11335), on read, 111, Connection refused

<ff3c54>; csession; fuzzy_check_io_callback: got error on IO with server fuzzy2.rspamd.com:11335(127.0.0.1:11335), on read, 111, Connection refused

<dcc3b7>; task; fuzzy_check_io_callback: got error on IO with server fuzzy1.rspamd.com:11335(127.0.0.1:11335), on read, 111, Connection refused

<5f770a>; task; fuzzy_check_io_callback: got error on IO with server fuzzy2.rspamd.com:11335(127.0.0.1:11335), on read, 111, Connection refused

I did spend many hours in documentations and the internet to gain some understanding. But honestly I’m unable to cope with it. I did look into most configurations found under /etc/rspamd/ and /var/lib/rspamd/. But I just don’t understand all the parameters…

In /etc/rspamd/rspamd.conf I did see, that local fuzzy storage is disabled. And since the error contains the local address 127.0.0.1:11335, I made the thought, that this might be linked to the error. Could this be a cause for the errors? So far I was not able to understand how this is linked with localhost etc.

nano /etc/rspamd/rspamd.conf

#Local fuzzy storage is disabled by default

worker "fuzzy" {
    bind_socket = "localhost:11335";
    count = -1; # Disable by default, see #4677 for details
    .include "$CONFDIR/worker-fuzzy.inc"
    .include(try=true; priority=1,duplicate=merge) "$LOCAL_CONFDIR/local.d/worker-fuzzy.inc"
    .include(try=true; priority=10) "$LOCAL_CONFDIR/override.d/worker-fuzzy.inc"
}

A nslookup for fuzzy 1 & 2 brought the following summary:

# nslookup fuzzy1.rspamd.com
Server:         127.0.0.53
Address:        127.0.0.53#53

Non-authoritative answer:
Name:   fuzzy1.rspamd.com
Address: 127.0.0.1

I’m stuck and don’t know if I just should ignore all the errors on my email-server or if I should find solutions for them. But since I don’t like errors, I prefer to solve them.

Pioneer FreedomBox with 25.13
.

Edit:
This morning I have made an adjustment in /etc/rspamd/options.inc:

# How often maps are checked
map_watch_interval = 120min;

As I faced the many map errors after update to trixie, I did increase the value from 5min (default) to 60min and the the map errors became very few (and this is the only change so far on my system).
The adjustment from this morning from 60 to 120 min did let the map errors disappear and since about seven hours I noticed no fuzzy errors either.
Could it be, that these errors relate to a filed cache or something like this because of often updated databases?

I will observe and let you know, how it will look like.

Ok, did some more investigation because the map error (https://discuss.freedombox.org/t/issues-with-rspamd/3920) hasn’t disappeared.
On a mailcow post (https://community.mailcow.email/d/5394-rspamd-ssl-connect-error-syscall-fail-connection-reset-by-peer/13) I found the maps problem is related to an outdated rspamd version. With rspamd version 3.13.0 the issue should be fixed. FreedomBox is currently running rspamd version 3.12.1.

Maybe the fuzzy error is related to to the same issue with maps and gets fixed with the update. So I need to wait for rspamd 3.13.0 on FreedomBox.
If anybody has more information, let me know.

I just noticed on the RSPAMD web interface under Status, that counting the Hashes started by zero again. This is the reason the fuzzy errors no longer occur, I think. Before starting over again, the value of the Hashes went up into the billions and may filled a limited store?
To avoid fuzzy errors, does a earlier data dump (system) of the Hashes solve the problem?

It looks like the fuzzy error is connected with a very high value of the Hashes.

And the map error will be solved with the rspamd update on 3.13.0.