On Sat, 07 Aug 2010 15:32:47 +0200 Tom Hendrikx <t...@whyscream.net> wrote:
> On 07/08/10 14:58, Stevan Bajić wrote: > > On Sat, 07 Aug 2010 14:44:00 +0200 > > Tom Hendrikx <t...@whyscream.net> wrote: > > > >> Hi, > >> > > Hallo Tom, > > > >> After upgrading to 3.9.1-rc (and later git HEAD), I noticed lots of > >> errors in my logging: > >> > >> 2010-08-07T14:13:08+02:00 vmail dspam[13054]: bailing on error -2 > >> 2010-08-07T14:13:08+02:00 vmail dspam[13054]: process_message returned > >> error -5. delivering. > >> > >> When this started, headers were also no longer added to the delivered > >> messages. After some debugging I found this in the debug log: > >> > >> 13054: [08/07/2010 14:13:08] _pgsql_drv_getpwnam: returning NULL for > >> query on name: classicifation_user that returned a null result > >> 13054: [08/07/2010 14:13:08] _ds_getall_spamrecords: unable to > >> _pgsql_drv_getpwnam(classicifation_user) > >> 13054: [08/07/2010 14:13:08] _ds_getall_spamrecords() failed > >> 13054: [08/07/2010 14:13:08] user_classify() returned error -5 > >> 13054: [08/07/2010 14:13:08] destroying/freeing configuration > >> 13054: [08/07/2010 14:13:08] _pgsql_drv_getpwnam returning cached name > >> tomhe...@example.net. > >> 13054: [08/07/2010 14:13:08] destroying/freeing configuration > >> > >> classification_user was in my intial setup intended to contain 'global > >> classification' data, but I never got started to merging stuff into it. > >> The user exists in the pgsql backend (dspam_virtual_uids), but has no > >> training data there. > >> Are you sure that the user "classification_user" existed in dspam_virtual_uids? Could you post the content of your old group file? Could you post the output of "select * from dspam_virtual_uids where username = 'classification_user'"? Do you have more lines from the log? I mean: Could you post more lines from the log from above? How can "_pgsql_drv_getpwnam(classicifation_user)" return NULL if the user exists/existed in dspam_virtual_uids? > >> DSPAM 3.9.0 did not have any problems with this empty user. > >> > > DSPAM 3.9.0 had totally messed up and wrong working group support. We got > > that fixed: > > > > http://dspam.git.sourceforge.net/git/gitweb.cgi?p=dspam/dspam;a=commit;h=22714e4f390d48187777b14e7585ede4d180c376 > > > > http://dspam.git.sourceforge.net/git/gitweb.cgi?p=dspam/dspam;a=commit;h=80b89d634481fa2eca17a6666f67e6d0117b76f8 > > > > Didn't know that it was broken, as I got never around to testing it. > Great :) > > > > > > >> I fixed the > >> issue for now, by removing the group file. > >> > >> When I have some new training data, I'll also try merging this back to > >> the classification user and see if it solves the issue. > >> > > You need to have a valid user. Else it will not work. > > That is no reason for spitting out an uninformal error message, and > abort processing. DSPAM should still process the message, but without > using the data from the classification user. > > -- > Regards, > Tom > ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Dspam-devel mailing list Dspam-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspam-devel
