Since somewhere around 3 weeks ago I’ve been receiving much larger amount of SPAM in my inbox.
It seems that rspamd neural module gives -4 with NEURAL_HAM_LONG and -2 with NEURAL_HAM_SHORT rating to most of spam emails, making them into inbox folder.
The confidence rate of NEURAL module is always -1.000
X-Spamd-Result: default: False [4.04 / 15.00];
RSPAMD_URIBL(4.50)[rosstrah.ru:url];
NEURAL_HAM_LONG(-4.00)[-1.000];
FREEMAIL_REPLYTO_NEQ_FROM_DOM(3.00)[];
NEURAL_HAM_SHORT(-2.00)[-1.000];
BAD_REP_POLICIES(2.00)[];
SUBJECT_ENDS_SPACES(0.50)[];
IP_REPUTATION_HAM(-0.41)[asn: 47764(-0.40), country: RU(-0.01), ip: 94.100.177.128(0.00)];
R_MIXED_CHARSET(0.25)[subject];
MIME_HTML_ONLY(0.20)[];
MIME_BASE64_TEXT(0.10)[];
MIME_GOOD(-0.10)[multipart/mixed];
MX_GOOD(-0.01)[];
XM_UA_NO_VERSION(0.01)[];
FROM_HAS_DN(0.00)[];
DWL_DNSWL_NONE(0.00)[mail.ru:dkim];
R_DKIM_ALLOW(0.00)[mail.ru:s=mail3,mail.ru:s=mail4];
ARC_NA(0.00)[];
TO_MATCH_ENVRCPT_ALL(0.00)[];
HAS_REPLYTO(0.00)[to98spb@bk.ru];
REPLYTO_DOM_NEQ_FROM_DOM(0.00)[];
RCVD_IN_DNSWL_NONE(0.00)[94.100.177.128:from];
FREEMAIL_REPLYTO(0.00)[bk.ru];
BCC(0.00)[];
CLAM_VIRUS_FAIL(0.00)[failed to scan and retransmits exceed];
RCPT_COUNT_ONE(0.00)[1];
MIME_TRACE(0.00)[0:+,1:~,2:~];
FREEMAIL_ENVFROM(0.00)[mail.ru];
ASN(0.00)[asn:47764, ipnet:94.100.176.0/20, country:RU];
HAS_ATTACHMENT(0.00)[];
TO_DN_NONE(0.00)[];
RCVD_COUNT_THREE(0.00)[3];
REPLYTO_DN_EQ_FROM_DN(0.00)[];
RCVD_VIA_SMTP_AUTH(0.00)[];
FREEMAIL_FROM(0.00)[mail.ru];
RCPT_MAILCOW_DOMAIN(0.00)[valdikss.org.ru];
MID_RHS_MATCH_FROMTLD(0.00)[];
DMARC_POLICY_ALLOW(0.00)[mail.ru,reject];
RCVD_TLS_LAST(0.00)[];
HAS_X_PRIO_ONE(0.00)[1];
R_SPF_ALLOW(0.00)[+ip4:94.100.176.0/20];
FROM_EQ_ENVFROM(0.00)[];
DKIM_TRACE(0.00)[mail.ru:+];
ARC_SIGNED(0.00)[valdikss.org.ru:s=msrv:i=1];
GREYLIST(0.00)[pass,body]
Since the confidence does not change over time, did the module broke somehow? Should I reset its data, or the whole rspamd data overall? If so, how should I do that properly?