Re: spamc timeouts continued with 3.2.2
On Jul 28, 2007, at 9:31 PM, Pedro LaWrench wrote: What happens if mail comes in while the cron-based expire is running? Will spamc timeout then? I stop most of the mail system (except for Postfix) at night while I do the expire and backup the database. It finishes much faster and I don't have expire problems.
Re: spamc timeouts continued with 3.2.2
--- Micha³ Jêczalik <[EMAIL PROTECTED]> wrote: > On Fri, 27 Jul 2007, Pedro LaWrench wrote: > > > I run spamc from SA 3.2.2 on one host, and spamd from SA 3.2.2 on another > host. > > 3.2 was supposed to fix timeouts (* bug 3466: do Bayes expiration, if > > required, after results have been passed back to the client from spamd; > this > > helps avoid client timeouts.), but a packet capture shows the following: > > > > - spamc sends message to spamd > > - spamd sends results back > > [spamd performs Bayes expiration] > > [spamc times out] > > - spamd sends FIN to close TCP connection > > > > So even though the results are sent back prior to Bayes expiration, spamc > will > > still wait until the TCP connection is closed. > > > > How can this get fixed? > > By disabling auto expire and doing it by cron. > -- > Micha³ Jêczalik, +48.603.64.62.97 > INFONAUTIC, +48.33.487.69.04 > > What happens if mail comes in while the cron-based expire is running? Will spamc timeout then? -PL Building a website is a piece of cake. Yahoo! Small Business gives you all the tools to get online. http://smallbusiness.yahoo.com/webhosting
Re: zero score rules still show up in 3.2.2
On Thu, 2007-07-26 at 13:30 -0500, McDonald, Dan wrote:
> I may have dreamed it, but I thought I remembered a discussion about
> removing rules with a zero score from spam reports. I upgraded one of
> my systems to 3.2.2 today (Mandriva Corporate Server 4.0, perl 5.8.7,
> called from amavisd-new 2.5.2) and still see zero scores from plugins
> displayed:
Bug 5519. http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5519
guenther
--
char *t="[EMAIL PROTECTED]";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}
RE: Have Spamassassin forward mail to Spam Folder
Silly enough for which part? Dumping or using the other rules here? I've
had my implementation in place for a year and a half and am confident in
dumping scores of 10+. The highest false positive I have seen in that time
was a 6.2.
When I attempted to implement using /usr/bin/spamassassin I saw nothing
being done in the maillog. When I switched back to using spamc it appears
to be working again.
> -Original Message-
> From: jdow [mailto:[EMAIL PROTECTED]
> Yes if you are silly enough.
> {^_^}
> - Original Message -
> From: "Skip Brott" <[EMAIL PROTECTED]>
>
>
> > These are more appropriately procmail questions, but
> >
> > Do you know if this ruleset will process before or after attempted
> > delivery
> > to the user (and thus triggering the .forward file)? Is there a
> > difference
> > between using /usr/bin/spamassassin versus using
> /usr/bin/spamc ? And can
> > I
> > still use this rule to dump spam with high scores?:
> > :0
> > * ^X-Spam-Level: \*\*\*\*\*\*\*\*\*\*
> > /dev/null
> >
> >> :0fwhb
> >> | /usr/bin/spamassassin
> >> :0H
> >> * ^X-Spam-Status: Yes
> >> /var/spool/mail/spam
> >
> >
Re: Bye for good FuzzyOCR - SVN Performance
Steve West wrote, on 26/07/07 10:59 AM: decoder wrote: Try using the SVN Version (revision 132). This is basically the same as the latest 3.5.x release but some issues with SA 3.2.x were fixed. Best regards, Chris We are running SA 3.2.1 and just wondering if anyone using the SVN version on a production server processing 10,000 to 15,000 emails per day? Specifically, I would be interested in hearing if the SVN is stable enough to be used in a production environment and what type of load gets added. thx, SW Just upgraded to 3.2.2. Running FuzzyOCR SVN. Looking ok thus far. We use to have (many revisions ago :)) spamd climb high in CPU and also children taking too long to complete, but this has not been noticed to date. It is not _our_ busy time of the year but : TOTAL Volume Messages Addresses Hosts Delayed Failed Received1306MB 41693 14264 8 0.0%546 1.3% Delivered 1306MB 41149 4854725 Rejects109324 35410 Temp Rejects 106 5 We reject at connect (500 errors) so the Reject line would be SPAM / Block list totals. JES
R: R: R: R: R: Question about SQL-based AWL
> -Messaggio originale- > Da: Luca Bertoncello [mailto:[EMAIL PROTECTED] > > "Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > > > Luca, I'm meaning this behavior should be enforced only by turning > the > > user_awl_sql_override_username switch on, and then the SQL 'username' > column > > would be filled with the username to be used to connect to the sql > db, as > > opposed to the current SA user. > > > > Actually, when user_awl_sql_override_username is off, you should get > a > > per-user AWL in which the sql 'username' column is filled with the > current > > SA user. Unfortunately, PersistentAddrList caches the SA username at > > startup, which is when the SA username is the one of the user > starting SA > > (nobody in your case). It seems to me that this caching defeats any > attempt > > to change the AWL username later. > > > > You can't obtain a per-user AWL by tweaking > user_awl_sql_override_username, > > then... > > Ach! > Does it mean, that it is not possible to have a per-user-AWL if I use > the SpamAssassin-Daemon? Right, that's what I meant. FWIK, every and each SA-powered daemon do create a fresh SA instance at startup, then use it to inspect mail directed to different users by first "tweaking" the current user in SA. Since PersistentAddrList caches the user at startup, this tweaking can't work to AWL... Giampaolo > > Thanks > Luca Bertoncello > ([EMAIL PROTECTED])
RE: sa-compile failes from cron job
Thanks, that one solved the issue. Catalin -Original Message- From: Daryl C. W. O'Shea [mailto:[EMAIL PROTECTED] Sent: Saturday, July 28, 2007 9:55 AM To: Catalin Miclaus Cc: Spamassassin Users List Subject: Re: sa-compile failes from cron job Catalin Miclaus wrote: > I have SA 3.2.1 and FreeBSD 6.2. > > I'm trying to run sa-compile by using a weekly cron job but I have this > error: > Can't exec "re2c": No such file or directory at > /usr/local/bin/sa-compile line 278, <$fh> line 2944. > > command failed! at /usr/local/bin/sa-compile line 279, <$fh> line 2944. > If I run same script from CLI is working fine. Add the path to re2c to your cron environment's PATH variable. Daryl
Re: sa-compile failes from cron job
On Sat, 28 Jul 2007 at 04:54 -0400, [EMAIL PROTECTED] confabulated: Catalin Miclaus wrote: I have SA 3.2.1 and FreeBSD 6.2. I?m trying to run sa-compile by using a weekly cron job but I have this error: Can't exec "re2c": No such file or directory at /usr/local/bin/sa-compile line 278, <$fh> line 2944. command failed! at /usr/local/bin/sa-compile line 279, <$fh> line 2944. If I run same script from CLI is working fine. Add the path to re2c to your cron environment's PATH variable. I noticed the same error. After adding /usr/local/bin to the path in the crontab for root (i.e.: PATH=/usr/local/bin at the top) all errors were resolved. The other error that showed up here was gpg couldn't be found. Thanks for the tip! --- _|_ (_| |
Re: R: R: R: R: Question about SQL-based AWL
"Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > Luca, I'm meaning this behavior should be enforced only by turning the > user_awl_sql_override_username switch on, and then the SQL 'username' column > would be filled with the username to be used to connect to the sql db, as > opposed to the current SA user. > > Actually, when user_awl_sql_override_username is off, you should get a > per-user AWL in which the sql 'username' column is filled with the current > SA user. Unfortunately, PersistentAddrList caches the SA username at > startup, which is when the SA username is the one of the user starting SA > (nobody in your case). It seems to me that this caching defeats any attempt > to change the AWL username later. > > You can't obtain a per-user AWL by tweaking user_awl_sql_override_username, > then... Ach! Does it mean, that it is not possible to have a per-user-AWL if I use the SpamAssassin-Daemon? Thanks Luca Bertoncello ([EMAIL PROTECTED])
R: R: R: R: Question about SQL-based AWL
> -Messaggio originale- > Da: Luca Bertoncello [mailto:[EMAIL PROTECTED] > > "Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > > > Ok. I can't understand if you are using spamd or amavisd, but you are > > probably running SA in a daemon which instantiate SA once and then > switches > > Yes, of course! I forgot to say it... > > > You are however right: this behavior should be enforced only by using > the > > user_awl_sql_override_username setting in SA, so it should not show > without > > it. One could easily regard it as a bug, then. > > OK! I seen this statement, but I didn't undestood how it works... > It expects a parameter, but I don't know what I have to write... Luca, I'm meaning this behavior should be enforced only by turning the user_awl_sql_override_username switch on, and then the SQL 'username' column would be filled with the username to be used to connect to the sql db, as opposed to the current SA user. Actually, when user_awl_sql_override_username is off, you should get a per-user AWL in which the sql 'username' column is filled with the current SA user. Unfortunately, PersistentAddrList caches the SA username at startup, which is when the SA username is the one of the user starting SA (nobody in your case). It seems to me that this caching defeats any attempt to change the AWL username later. You can't obtain a per-user AWL by tweaking user_awl_sql_override_username, then... Giampaolo > > Could you give me an example? > > Thanks a lot! > Luca Bertoncello > ([EMAIL PROTECTED])
Re: R: R: R: Question about SQL-based AWL
"Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > Ok. I can't understand if you are using spamd or amavisd, but you are > probably running SA in a daemon which instantiate SA once and then switches Yes, of course! I forgot to say it... > You are however right: this behavior should be enforced only by using the > user_awl_sql_override_username setting in SA, so it should not show without > it. One could easily regard it as a bug, then. OK! I seen this statement, but I didn't undestood how it works... It expects a parameter, but I don't know what I have to write... Could you give me an example? Thanks a lot! Luca Bertoncello ([EMAIL PROTECTED])
R: R: R: Question about SQL-based AWL
> -Messaggio originale-
> Da: Luca Bertoncello [mailto:[EMAIL PROTECTED]
>
> "Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb:
>
> > What is your milter? What is your setup? This may influence stuff
> like
> > Back'n'Whitelists as well as autowhitelist.
>
> I use Exim and I scan the E-Mail with this rule:
>
> warn message = X-Spam-Score: $spam_score ($spam_bar)
> spam= ${acl_m9}:true
>
> acl_m9 contains the username on the system.
>
> After the documentation of Exim, it will give the username to
> SpamAssassin
> (and this is confirmed by the using of personalized Black-/Whitelists).
Ok. I can't understand if you are using spamd or amavisd, but you are
probably running SA in a daemon which instantiate SA once and then switches
user at will. Now, if this is the case the
Mail::SpamAssassin::PersistentAddrList class is instantiated only once too.
At creation time, Mail::SpamAssassin::PersistentAddrList copies the current
SA user in its own object, and then keeps using the copy.
Thereby, when spamd/amavisd switches the SA user to the one which will get
the e-mail, the username copy in the Mail::SpamAssassin::PersistentAddrList
object wouldn't be affected, thereby the problem.
This is something I already saw in my postgresql awl tables, but I'm quite
happy with it because, after all, I prefer AWL to work globally, since it
would be much less useful in a per-user way.
You are however right: this behavior should be enforced only by using the
user_awl_sql_override_username setting in SA, so it should not show without
it. One could easily regard it as a bug, then.
Giampaolo
> Any idea?
>
> Thanks
> Luca Bertoncello
> ([EMAIL PROTECTED])
Re: R: R: Question about SQL-based AWL
"Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb:
> What is your milter? What is your setup? This may influence stuff like
> Back'n'Whitelists as well as autowhitelist.
I use Exim and I scan the E-Mail with this rule:
warn message = X-Spam-Score: $spam_score ($spam_bar)
spam= ${acl_m9}:true
acl_m9 contains the username on the system.
After the documentation of Exim, it will give the username to SpamAssassin
(and this is confirmed by the using of personalized Black-/Whitelists).
Any idea?
Thanks
Luca Bertoncello
([EMAIL PROTECTED])
R: R: Question about SQL-based AWL
> -Messaggio originale- > Da: Luca Bertoncello [mailto:[EMAIL PROTECTED] > > "Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > > > It is the user in behalf of whom SA is parsing the message. > > > > This may mean the user actually receiving the message, but in many > setups > > (like, in example, amavisd-new) this is the user running the milter > > software. You are probably running SA as the "nobody" user. > > Now the other question: why does SpamAssassin use the user > configuration for > the Black-/Whitelists and not for the AutoWhitelist? What is your milter? What is your setup? This may influence stuff like Back'n'Whitelists as well as autowhitelist. Giampaolo > > Thanks > Luca Bertoncello > ([EMAIL PROTECTED])
Re: R: Question about SQL-based AWL
"Giampaolo Tomassoni" <[EMAIL PROTECTED]> schrieb: > It is the user in behalf of whom SA is parsing the message. > > This may mean the user actually receiving the message, but in many setups > (like, in example, amavisd-new) this is the user running the milter > software. You are probably running SA as the "nobody" user. Now the other question: why does SpamAssassin use the user configuration for the Black-/Whitelists and not for the AutoWhitelist? Thanks Luca Bertoncello ([EMAIL PROTECTED])
R: Question about SQL-based AWL
> -Messaggio originale- > Da: Luca Bertoncello [mailto:[EMAIL PROTECTED] > > Hi, list! > > I use SpamAssassin 3.2.2 and I configured it to use a MySQL-DB to > manage the > Bayes and the AWL. > > So, I created the table awl after the documentation, and I write all > the data > in local.cf. > All runs perfectly! > > Now the question: in the table awl I have a field username, but this is > always nobody. > I tought, it is the user that receive the E-Mails, but it doesn't seems > so. > > Can someone explain me, what does this field do? It is the user in behalf of whom SA is parsing the message. This may mean the user actually receiving the message, but in many setups (like, in example, amavisd-new) this is the user running the milter software. You are probably running SA as the "nobody" user. Giampaolo > > Thanks a lot > Luca Bertoncello > ([EMAIL PROTECTED])
Question about SQL-based AWL
Hi, list! I use SpamAssassin 3.2.2 and I configured it to use a MySQL-DB to manage the Bayes and the AWL. So, I created the table awl after the documentation, and I write all the data in local.cf. All runs perfectly! Now the question: in the table awl I have a field username, but this is always nobody. I tought, it is the user that receive the E-Mails, but it doesn't seems so. Can someone explain me, what does this field do? Thanks a lot Luca Bertoncello ([EMAIL PROTECTED])
Re: sa-compile failes from cron job
Catalin Miclaus wrote: I have SA 3.2.1 and FreeBSD 6.2. I’m trying to run sa-compile by using a weekly cron job but I have this error: Can't exec "re2c": No such file or directory at /usr/local/bin/sa-compile line 278, <$fh> line 2944. command failed! at /usr/local/bin/sa-compile line 279, <$fh> line 2944. If I run same script from CLI is working fine. Add the path to re2c to your cron environment's PATH variable. Daryl
sa-compile failes from cron job
Hello guys, I have SA 3.2.1 and FreeBSD 6.2. I'm trying to run sa-compile by using a weekly cron job but I have this error: cd /tmp/.spamassassin90746Zl0Orztmp cd Mail-SpamAssassin-CompiledRegexps-body_0 re2c -i -b -o scanner1.c scanner1.re Can't exec "re2c": No such file or directory at /usr/local/bin/sa-compile line 278, <$fh> line 2944. command failed! at /usr/local/bin/sa-compile line 279, <$fh> line 2944. If I run same script from CLI is working fine. These are permissions on re2c file: [EMAIL PROTECTED] xxx]# ls /usr/local/bin/re2c -r-xr-xr-x 1 root wheel 131348 Jul 3 06:42 /usr/local/bin/re2c Catalin
Re: spamc timeouts continued with 3.2.2
On Fri, 27 Jul 2007, Pedro LaWrench wrote: I run spamc from SA 3.2.2 on one host, and spamd from SA 3.2.2 on another host. 3.2 was supposed to fix timeouts (* bug 3466: do Bayes expiration, if required, after results have been passed back to the client from spamd; this helps avoid client timeouts.), but a packet capture shows the following: - spamc sends message to spamd - spamd sends results back [spamd performs Bayes expiration] [spamc times out] - spamd sends FIN to close TCP connection So even though the results are sent back prior to Bayes expiration, spamc will still wait until the TCP connection is closed. How can this get fixed? By disabling auto expire and doing it by cron. -- Michał Jęczalik, +48.603.64.62.97 INFONAUTIC, +48.33.487.69.04
