[sa-learn] already scanned and tagged mails
I'm getting a nice bunch of SPAM that is not recognized as such. Now I have manually sorted out those mails and want to feed SA. Nearly all mails, 'ham' or 'spam' contain the SA header tags (X-Spam-Checker-Version, X-Spam-Level and X-Spam-Status). Do I have to filter out these headers before learning or is »sa-learn« (and »spamc« too) ignoring these fields? thanks Lars
Re: [sa-learn] already scanned and tagged mails
From: Lars Uhlmann [EMAIL PROTECTED]
I'm getting a nice bunch of SPAM that is not recognized as such. Now I
have manually sorted out those mails and want to feed SA. Nearly all
mails, 'ham' or 'spam' contain the SA header tags (X-Spam-Checker-Version,
X-Spam-Level and X-Spam-Status). Do I have to filter out these headers
before learning or is »sa-learn« (and »spamc« too) ignoring these
fields?
Nope, just be sure you feed sa-learn properly given the format the
in which the messages are stored and whether they are ham or spam.
{^_^}
Re: Spamassassin Bayes
debug: tests=ALL_TRUSTED,MISSING_DATE,MISSING_SUBJECT,NO_REAL_NAME Here is a BIG problem! You got an ALL_TRUSTED on this spam. Assuming you fed in a spam that came from outside and the headers are still there showing that, it shoudl NEVER get an ALL_TRUSTED indication. This results in a negative score, being added to the total, and in this case is probably responsible for the major problems here. You MUST set trusted_networks correctly. so i have read something about spamassassin and the configuration of trusted networks. I found an example: trusted_networks 123.12.34.56 123.12.35/24 That line will specify that the host at 123.12.34.56, and all hosts in the 123.12.35.0 - 123.12.35.255 address range, are to be trusted. But i dont understand it really. I have a webserver with an Ip Adress. How must i configure it? Thank you very much for your help marcus _ Haben Spinnen Ohren? Finden Sie es heraus mit dem MSN Suche Superquiz via http://www.msn-superquiz.de Jetzt mitmachen und gewinnen!
need help with spamd-mysql config required_hits problem please
hey there
first time mailing to the list, so apologies for any gaffs i make. I've
searched the archive at GMANE for my problem but I can't find much associated
with it.
problem:
I've got Postfix 2.2 (using virtual mailboxes through mysql), spamassassin 3
(getting it's required_hits and whitelists from the mysql db), clamd, and
latest version of mysql all trying to work together.
my particular problem is that, right now, I have these lines in
postfix/master.cf:
smtp inet n - y - - smtpd -o
content_filter=spamassassin
spamassassin unix - n n - - pipe user=filter
argv=/usr/bin/spamc -f -e /usr/sbin/sendmail -oi -f ${sender} ${recipient}
This WOULD be fine for what I need except for the fact that I want each user
to have control over his/her required hits factor (through php-sa/mysql).
As shown, spam runs as filter and thus checks the sql database using the
filter user:
Jan 14 21:11:55 krynn spamd[2497]: spamd: processing message
[EMAIL PROTECTED] for filter:41031
So with each mail going through spamassassin as the user filter then it does
not read each user's required hits in mysql, just the user filter's.
IF I comment out the spamassassin sections above in master.cf and use the
mailbox_command in main.cf to pipe it through a site-wide /etc/procmailrc,
THEN it works (more or less).
I'm looking at this page:
http://www.akadia.com/services/postfix_spamassassin.html
and wondering if i can cobble a bash script together that will filter out the
email address from the To: line and use THAT as the user spamc runs as so
that, when it goes to the spamd daemon, it contacts the mysql database and
checks the required hits for THAT user's email address rather then filter.
Does it have anything to do with the -Q flag on spamd's command line?
Solution: ?? :)
Confusing as heck isnt it?
any ideas please? ANY ideas would be helpful right now but I dont really want
to install mailscanner or amavis if it can be avoided please.
Thanks very much for any help/pointers etc.
Steve
Gain an extra 25%! (was Purging the Spamassassin Database)
Jason Frisvold xenophage0 at gmail writes: I've been investigating some recent slowness issues with our mail servers and I noticed that the spamassassin database is getting rather large. We process approximately 300,000 mails a day (or more). The bayes_token database is over 1.8 Gig at the moment. (Actually, 1.8 Gig for the data, and 1.3 Gig for the index) If you have fewer than 65,000 accounts you could halve the size of the id columns. Those folks *not* storing per-user Bayes statistics -- i.e. with only one line in bayes_vars -- could take the id column down to one byte, a TINYINT. ALTER TABLE bayes_token MODIFY id SMALLINT UNSIGNED NOT NULL; ALTER TABLE bayes_expire MODIFY id SMALLINT UNSIGNED NOT NULL; ALTER TABLE bayes_seen MODIFY id SMALLINT UNSIGNED NOT NULL; ALTER TABLE bayes_vars MODIFY id SMALLINT UNSIGNED NOT NULL; The last three won't recover much space, but table indexes should be kept the same type and size. Also, try the following. Since you are keeping Bayes per user, your spam_count and ham_count probably don't get very high so you can take those down to SMALLINTs. SELECT MAX(ham_count), MAX(spam_count) FROM bayes_token; ALTER TABLE bayes_token MODIFY spam_count SMALLINT UNSIGNED NOT NULL, MODIFY ham_count SMALLINT UNSIGNED NOT NULL; That took 25% off my bayes_token data file and index. Since MySQL likes to keep indexes in VM, I think it was worthwhile. -- Ard
Error building 64-bit on FC2
All - I am running Fedora Core 2 on an AMD Athlon(tm) 64 Processor 3000+. It is up to date with the latest packages available. I grabbed the latest SA tarball from a mirror, and ran rpmbuild -ta against it. It fails with this: Manifying blib/man3/Mail::SpamAssassin::DnsResolver.3pm Manifying blib/man3/Mail::SpamAssassin::SubProcBackChannel.3pm Manifying blib/man3/Mail::SpamAssassin::Plugin::RelayCountry.3pm + /usr/bin/make spamc/libspamc.so /usr/bin/make -f spamc/Makefile spamc/libspamc.so make[1]: Entering directory `/home/thomas.cameron/redhat/BUILD/Mail- SpamAssassin-3.1.0' gcc -rdynamic -Wl,-rpath,/usr/lib64/perl5/5.8.3/x86_64-linux-thread- multi/CORE spamc/libspamc.c spamc/utils.c \ -o spamc/libspamc.so -shared -ldl /usr/bin/ld: /tmp/cc2TpgXy.o: relocation R_X86_64_32S can not be used when making a shared object; recompile with -fPIC /tmp/cc2TpgXy.o: could not read symbols: Bad value collect2: ld returned 1 exit status make[1]: *** [spamc/libspamc.so] Error 1 make[1]: Leaving directory `/home/thomas.cameron/redhat/BUILD/Mail- SpamAssassin-3.1.0' make: *** [spamc/libspamc.so] Error 2 error: Bad exit status from /var/tmp/rpm-tmp.97589 (%build) RPM build errors: Bad exit status from /var/tmp/rpm-tmp.97589 (%build) Odd thing is, if I just grab the tarball and run perl Makefile.PL /dev/null make make install DESTDIR=/var/tmp/sa it installs just fine to /var/tmp/sa Thoughts? Thomas
Re: Error building 64-bit on FC2
On Mon, Jan 16, 2006 at 12:07:25AM -0600, Thomas Cameron wrote: Odd thing is, if I just grab the tarball and run perl Makefile.PL /dev/null make make install DESTDIR=/var/tmp/sa it installs just fine to /var/tmp/sa Thoughts? Yeah, the spec file does 1 thing that the normal install doesn't, which has the error, namely build libspamc.so. You can grab http://www.kluge.net/~felicity/random/spec.patch and apply it to the tarball version, then build the RPM. -- Randomly Generated Tagline: Fatbot: I heard that in one single night you drank a whole keg, streaked across campus, and crammed fifty-eight humans into a phone booth. Bender: Yeah, well, a lot of 'em were children pgpR7z41src8t.pgp Description: PGP signature
