> > It seems that since upgrading to 3.11 last night, the > > autolearn=yes|no|failed always shows up as failed (on all messages). > > > > Did I miss something? > > Generally "Failed" means that SA somehow can't access the > bayes database to write to it. > > Have you tried sa-learn --sync and then sa-learn --dump magic > to see if those tools find the bayes database correctly? > I had run the --sync last night...
I just ran the --dump (no issues) C:\Perl\bin>sa-learn --dump 0.000 0 3 0 non-token data: bayes db version 0.000 0 74899 0 non-token data: nspam 0.000 0 36273 0 non-token data: nham 0.000 0 122147 0 non-token data: ntokens 0.000 0 1140696522 0 non-token data: oldest atime 0.000 0 1142280376 0 non-token data: newest atime 0.000 0 0 0 non-token data: last journal sync atime 0.000 0 1142162419 0 non-token data: last expiry atime 0.000 0 1466085 0 non-token data: last expire atime delta 0.000 0 31761 0 non-token data: last expire reduction co unt > What about spamassassin --lint to check for config-file errors? > Good suggestion.. I forgot to do that. It found 2 lint issues: use_razor 0 use_dcc 0 Which I removed after remembering about those getting moved to plugins. And I ran --lint again just for sanity sake -- without error. It looks like the autolearn is working now. Does anyone have any ideas on the "misplaced" headers? Or is that a new de facto standard? Steven
