[Rspamd-Users] per-user statistics - best practice? - spam-learning

[Stefan Bauer]

Hi rspamd-users,

we have around 600 mailboxes to protect with rspamd. We're using
per_user option for statistics in combination with autolearn.

Am i right, that spam/ham learned per user, will only be used in
rspamd-decisions for this specific user?

Wouldnt it be better to skip per_user learning?

I might remember that per_user was mentioned in the rspamd docs
previously but was now movd to (optional).

Ideas/thoughts?

Stefan