dspam-devel Mailing List for DSPAM (Page 88)
Brought to you by:
paulcockings,
sbajic
You can subscribe to this list here.
2009 |
Jan
|
Feb
(8) |
Mar
(13) |
Apr
(21) |
May
(70) |
Jun
(92) |
Jul
(125) |
Aug
(149) |
Sep
(52) |
Oct
(86) |
Nov
(116) |
Dec
(163) |
---|---|---|---|---|---|---|---|---|---|---|---|---|
2010 |
Jan
(113) |
Feb
(28) |
Mar
(24) |
Apr
(107) |
May
(59) |
Jun
(9) |
Jul
(38) |
Aug
(503) |
Sep
(16) |
Oct
(12) |
Nov
(19) |
Dec
(17) |
2011 |
Jan
(2) |
Feb
(20) |
Mar
(2) |
Apr
(3) |
May
(36) |
Jun
(40) |
Jul
(2) |
Aug
(76) |
Sep
(10) |
Oct
(10) |
Nov
(14) |
Dec
(8) |
2012 |
Jan
(1) |
Feb
(9) |
Mar
(4) |
Apr
(9) |
May
(8) |
Jun
(15) |
Jul
|
Aug
(18) |
Sep
(6) |
Oct
(6) |
Nov
(5) |
Dec
(6) |
2013 |
Jan
(16) |
Feb
(13) |
Mar
(5) |
Apr
(1) |
May
|
Jun
(17) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(3) |
Dec
|
2014 |
Jan
(4) |
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(5) |
Aug
(1) |
Sep
|
Oct
|
Nov
(7) |
Dec
|
2016 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
(1) |
Oct
|
Nov
|
Dec
|
2019 |
Jan
(5) |
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
2020 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
|
Oct
|
Nov
|
Dec
|
2021 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(1) |
From: Paul C. <ds...@cy...> - 2009-04-15 21:08:55
|
-------------- New Website -------------- Something seems to be up with the link I gave you in my last email. You should find the new website on the nuclearelephant.com domain http://www.nuclearelephant.com/ - Sorry! |
From: Andreas S. <ma...@cy...> - 2009-04-15 21:01:30
|
On Wednesday 15 April 2009 22:36:22 Paul Cockings wrote: > ---------------- > GIT Relocation > ---------------- > > Sourceforge now supports GIT, so we've moved the GIT repo for dspam to > Sourceforge! Yippee! > You can find the new web-git front end here: > http://dspam.git.sourceforge.net/ Great, are you guys still interested in switching to CMake, if yes I will migrate it. This needs ~2 hours. Cheers, -- andreas |
From: Paul C. <ds...@cy...> - 2009-04-15 20:36:32
|
-------------- New Website -------------- We've just completed the move of the old nuclearelephant.com domain and we have new website! http://dpsam.sf.net --------------- RABL Service --------------- We've had some reports of people still trying to use rabl.nuclearelephant.com service for a DNSBL As far as the new devel team know this service was an experiment some years ago and to our knowledge wasn't running on the nuclearelephant.com server at the time of hand-over from Sensory Networks. We suspect that as the nuclearelephant.com domain is no longer responding to RABL requests then some poorly implemented DNSBL clients are treating the lookup as a negative rather than failure to lookup. The RABL was a cool idea, so we hope to revisit this subject at some point in the future but for now the service doesn't exist. If you think you might know something about the old rabl service that might be helpful, please post your knowledge on the mailing lists - it'll be greatly appreciated. ---------------- GIT Relocation ---------------- Sourceforge now supports GIT, so we've moved the GIT repo for dspam to Sourceforge! Yippee! You can find the new web-git front end here: http://dspam.git.sourceforge.net/ -------------- New Release -------------- In line with listed todo's on the wiki homepage (although a little behind our original time-scale), we are in the final stages of making the new release of dspam 3.9.0 available. More news will be announced on the lists shortly ------------ How to help ------------ Lastly, thankyou for your continued interest in Dspam. If you have any ideas, bugfixes, patches or suggestions please head on over to the new website (http://dspam.sf.net) or the wiki pages (http://apps.sourceforge.net/mediawiki/dspam/index.php?title=Main_Page) to see how you can help the project! The Dspam Admin Team |
From: Julien V. <ju...@ki...> - 2009-04-05 16:01:49
|
Hi, I am not able to build dspam against the newer sqlite3 3.6.xx series, while everything is OK with 3.5.95. Here is the error message: checking sqlite3.h usability... yes checking sqlite3.h presence... yes checking for sqlite3.h... yes checking SQLite header version... 3.6.12 checking how to link SQLite libraries... failure configure: error: Required version of libsqlite not found I have no idea what I could do to try and provide more information for this issue. Note that I have reported this bug on Debian BTS first, but thought the new development team would be interested in fixing this in git. Cheers, Julien -- Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org Rejoignez maintenant plus de 4 500 personnes, associations, entreprises et collectivités qui soutiennent notre action |
From: Chris L. <ch...@le...> - 2009-04-04 20:40:01
|
Hi, I've been trying to submit diffs against a 3.8.1 for the patches I recently described on this list, however, I'm unable to finish compiling it on my system (Mac OS X 10.5.6, gcc 4.0.1, etc) Anyone have any idea how to get past the following problem? > libtool: link: /usr/bin/gcc-4.0 -dynamiclib -Wl,-undefined - > Wl,dynamic_lookup -o .libs/libdspam.7.dylib .libs/libdspam.o .libs/ > heap.o .libs/diction.o .libs/base64.o .libs/buffer.o .libs/ > util.o .libs/nodetree.o .libs/error.o .libs/decode.o .libs/ > pref.o .libs/read_config.o .libs/config_shared.o .libs/bnr.o .libs/ > hash.o .libs/list.o .libs/external_lookup.o .libs/tokenizer.o .libs/ > hash_drv.o -L/opt/local/lib -lm -pthread -pthread - > install_name /opt/local/lib/libdspam.7.dylib -compatibility_version > 8 -current_version 8.0 -Wl,-single_module > ld: duplicate symbol _agent_config in .libs/read_config.o and .libs/ > pref.o -Chris |
From: Hugo M. <hug...@fc...> - 2009-03-25 22:25:33
|
Sylvain Garcia wrote: > Paul Cockings wrote: > >> Sylvain Garcia wrote: >> >>> Hi all, i'm working on integration of dspam in OBM http://www.obm.org, >>> a groupware. >>> I have been tested experiemental debian packages of dspam-community on >>> http://kirya.net. >>> >>> Intégration works fine, i must fix some code on OBM in order to have >>> perfect integration. >>> >>> So I would like know, if you have a date or some indication about one >>> first release of dspam-community. What'is the the remaining works on >>> dspam-community? >>> >>> tx >>> >>> >> We're in the process of going back over all the commits from 3.8.0 to >> date, evaluating each one. This process is taking lots of time, so the >> release isn't due in the next few days. I'd recommend the code >> currently in GIT as this code is being run by many people in production >> (its not without problems, but currently the best). I'm really excited >> about your plans to integrate dspam-community to OBM, please let us know >> your feedback. If you'd like to copy your setup instructions to the >> wiki, then login to the wiki and then drop me a mail. I've then upgrade >> your wiki permissions. If you find any problems with dspam, or your'd >> like some features please add them to the trackers. >> >> Best regards, >> Paul >> >> > Ok, thanks for this complete answer. No problem for feedback about this > integration, I could write setup instruction about configuration, but > the goal is to package installation and setup into obm!! ;) (deb and rpm). > I choose this configuration with dspam and OBM: > * postgresql. OBM is deployed on many big infrastructure ( Some french > ministry and private french organization :D ) so we must support between > 1000 at 80000 users. About this, what is average size by user in posgres > database? Of course i know this query is difficult... > * ldap check user in obm-ldap, I think that i need modify OBM to store > "obm id database" in ldap to have uniq identifier because in obm we can > change login, so if user change login he can lost his dspam database... > > Hello Sylvain, If you choose to use the latest cvs/git release, you can use an experimental feature, which i've been using with great success for the last 2 years, which provides external user lookup. Currently it supports built-in LDAP user lookup and external program lookup (for those not using ldap). That will allow you to define which unique ldap attribute you want to use as your dspam username. That way your users can both change logins and email address and you can keep mapping them to the defined unique and unchangeable ldap attribute. Look in the dspam.conf file for ExtLookup. Best regards, Hugo Monteiro. -- ci.fct.unl.pt:~# cat .signature Hugo Monteiro Email : hug...@fc... Telefone : +351 212948300 Ext.15307 Web : http://hmonteiro.net Centro de Informática Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa Quinta da Torre 2829-516 Caparica Portugal Telefone: +351 212948596 Fax: +351 212948548 www.ci.fct.unl.pt ap...@fc... ci.fct.unl.pt:~# _ |
From: Sylvain G. <syl...@ob...> - 2009-03-25 21:09:09
|
Paul Cockings wrote: > Sylvain Garcia wrote: >> Hi all, i'm working on integration of dspam in OBM http://www.obm.org, >> a groupware. >> I have been tested experiemental debian packages of dspam-community on >> http://kirya.net. >> >> Intégration works fine, i must fix some code on OBM in order to have >> perfect integration. >> >> So I would like know, if you have a date or some indication about one >> first release of dspam-community. What'is the the remaining works on >> dspam-community? >> >> tx >> > We're in the process of going back over all the commits from 3.8.0 to > date, evaluating each one. This process is taking lots of time, so the > release isn't due in the next few days. I'd recommend the code > currently in GIT as this code is being run by many people in production > (its not without problems, but currently the best). I'm really excited > about your plans to integrate dspam-community to OBM, please let us know > your feedback. If you'd like to copy your setup instructions to the > wiki, then login to the wiki and then drop me a mail. I've then upgrade > your wiki permissions. If you find any problems with dspam, or your'd > like some features please add them to the trackers. > > Best regards, > Paul > Ok, thanks for this complete answer. No problem for feedback about this integration, I could write setup instruction about configuration, but the goal is to package installation and setup into obm!! ;) (deb and rpm). I choose this configuration with dspam and OBM: * postgresql. OBM is deployed on many big infrastructure ( Some french ministry and private french organization :D ) so we must support between 1000 at 80000 users. About this, what is average size by user in posgres database? Of course i know this query is difficult... * ldap check user in obm-ldap, I think that i need modify OBM to store "obm id database" in ldap to have uniq identifier because in obm we can change login, so if user change login he can lost his dspam database... In the first time i configure dspam without quarantaine, and use sieve script to sort mail in mailbox of user. Then I wouldlike use an other quarantaine than dspam-cgi. The actual quarantain system store mail on dspam server, but I can install this because the big consummer will refuse install apache and other package on their DMZ server... So for quarantaine system I think ( but I think yet..) that we will deliver mail in an other cyrus mailbox (perhaps an other cyrus server) and we will devellope own quarantaine ui, perhaps in minig... (http://www.minig.org). -- Sylvain Garcia syl...@ob... OBM - The opensource groupware - http://www.obm.org |
From: Paul C. <ds...@cy...> - 2009-03-24 09:41:35
|
Sylvain Garcia wrote: > Hi all, i'm working on integration of dspam in OBM http://www.obm.org, a > groupware. > I have been tested experiemental debian packages of dspam-community on > http://kirya.net. > > Intégration works fine, i must fix some code on OBM in order to have > perfect integration. > > So I would like know, if you have a date or some indication about one > first release of dspam-community. What'is the the remaining works on > dspam-community? > > tx > We're in the process of going back over all the commits from 3.8.0 to date, evaluating each one. This process is taking lots of time, so the release isn't due in the next few days. I'd recommend the code currently in GIT as this code is being run by many people in production (its not without problems, but currently the best). I'm really excited about your plans to integrate dspam-community to OBM, please let us know your feedback. If you'd like to copy your setup instructions to the wiki, then login to the wiki and then drop me a mail. I've then upgrade your wiki permissions. If you find any problems with dspam, or your'd like some features please add them to the trackers. Best regards, Paul |
From: Sylvain G. <syl...@ob...> - 2009-03-23 15:58:53
|
Hi all, i'm working on integration of dspam in OBM http://www.obm.org, a groupware. I have been tested experiemental debian packages of dspam-community on http://kirya.net. Intégration works fine, i must fix some code on OBM in order to have perfect integration. So I would like know, if you have a date or some indication about one first release of dspam-community. What'is the the remaining works on dspam-community? tx -- syl...@ob... |
From: Chris L. <ch...@le...> - 2009-03-18 19:30:46
|
Hi, I've been using dspam for some time (amavisd+postfix+clamav+dspam) and have been very satisfied with Dspam's performance. However, the whole setup that I have is intricately dependent on the performance of spamassassin, perl, etc, so I decided to venture out into dspam+clamav territory and drop amavisd and spamassassin. In the process of doing so I ran across what appear to be several unaddressed problems, and I believe I have fixes for them. I'm happy to see that a new dspam community is assembling, and I look forward to future revs of the software. I've submitted three bugs to the tracker: ID: 2692393 User detection improper for non-domainscale setup ID: 2692425 DNSBL Detection and Inoculation are broken ID: 2692428 WEBUI doesn't handle \r characters at the end of signatures I'm just using DIGEST/SSL authentication through apache for access to the WEBUI, so I didn't want to use any domain scale capability, which would complicate authentication. My coding skills are limited, but I've managed to implement fixes for all three problems. When I can separate the fixes of 2692393 from 2692425 I'll submit diffs. I've been an open source user for nearly two decades now, but I have yet to participate in a project. With the fixes above, my Mac OS X Server 10.5.6 installation (postfix, mysql 5.067, clamav 0.94.2) works smoothly inoculating off the DNSBLs and allowing maintenance through the WEBUI. I'm looking forward to helping out where I can. I'll use the tools to submit feature requests and other bugs that I'm aware of... -Chris |
From: Jyothsni R. <jyo...@gm...> - 2009-03-17 14:50:48
|
Any particular reason for this? Because it performs worse sometimes when I add stemming, especially with respect to false positives. 2009/3/17 MKondrin <mko...@hp...> > Jyothsni Reddy wrote: > >> Hi, >> >> Does dspam do stemming? >> >> No. > >> Thanks, >> Jyothsni >> ------------------------------------------------------------------------ >> >> >> ------------------------------------------------------------------------------ >> Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are >> powering Web 2.0 with engaging, cross-platform capabilities. Quickly and >> easily build your RIAs with Flex Builder, the Eclipse(TM)based development >> software that enables intelligent coding and step-through debugging. >> Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Dspam-devel mailing list >> Dsp...@li... >> https://lists.sourceforge.net/lists/listinfo/dspam-devel >> >> > > |
From: Mate G. <mai...@mo...> - 2009-03-17 11:09:24
|
> > -------- Original-Nachricht -------- >> Datum: Mon, 16 Mar 2009 20:55:26 +0100 >> Von: Gabri Mate <mai...@mo...> >> An: dsp...@li... >> Betreff: [Dspam-devel] forged from header - not scanning > >> Hey there! >> > Hello. > > >> I've recently noticed that if i recieve a spam with a forged header, so >> it appears that i have sent it that mail to myself from the address in >> question, then dspam won't process it. >> >> I have an account which i check with fetchmail and i pass the message to >> dspam through procmail: >> >> #dspam >> :0fw >> | /usr/local/bin/dspam --stdout --deliver=innocent,spam --mode=teft >> --feature=noise --user dotty >> >> but if the sender is the same as the recipient, then dspam won't analyze >> it. I haven't found a configuration option for this issue yet. >> >> Do You have any advice? >> > 1) Can you post the full header of the message dspam does deliver after > not scanning the message? I'm not in front of that machine, but i can tell that there were not a single header from dspam. So the message wasn't through dspam. > > 2) Could you post some log entries showing that dspam did not or did > processed the message? I can't see anything in the logs, even in the procmail log files. Maybe somehow the procmail filter won't process a message from 'myself'? > > > btw: There is no such switch in the dsapm configuration to disable > complete scanning of messages. Every message get's tagged in the headers > by dspam. So if it is processed by dspam, then there must be a entry in > the headers. If there is no entry in the headers, then you have not > processed the message with dspam. Yes, that's what i thought. Maybe i have to look around procmail to figure out the problem. > > >> Thank You! >> -- >> Gabri Mate >> > Steve > >> ------------------------------------------------------------------------------ >> Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are >> powering Web 2.0 with engaging, cross-platform capabilities. Quickly and >> easily build your RIAs with Flex Builder, the Eclipse(TM)based >> development >> software that enables intelligent coding and step-through debugging. >> Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com >> _______________________________________________ >> Dspam-devel mailing list >> Dsp...@li... >> https://lists.sourceforge.net/lists/listinfo/dspam-devel > > -- > Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: > http://www.gmx.net/de/go/multimessenger01 > > ------------------------------------------------------------------------------ > Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are > powering Web 2.0 with engaging, cross-platform capabilities. Quickly and > easily build your RIAs with Flex Builder, the Eclipse(TM)based development > software that enables intelligent coding and step-through debugging. > Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com > _______________________________________________ > Dspam-devel mailing list > Dsp...@li... > https://lists.sourceforge.net/lists/listinfo/dspam-devel > > !DSPAM:117,49becdb0263228477911171! > > > |
From: Mate G. <mai...@mo...> - 2009-03-17 11:08:37
|
> > -------- Original-Nachricht -------- >> Datum: Mon, 16 Mar 2009 20:55:26 +0100 >> Von: Gabri Mate <mai...@mo...> >> An: dsp...@li... >> Betreff: [Dspam-devel] forged from header - not scanning > >> Hey there! >> > Hello. > > >> I've recently noticed that if i recieve a spam with a forged header, so >> it appears that i have sent it that mail to myself from the address in >> question, then dspam won't process it. >> >> I have an account which i check with fetchmail and i pass the message to >> dspam through procmail: >> >> #dspam >> :0fw >> | /usr/local/bin/dspam --stdout --deliver=innocent,spam --mode=teft >> --feature=noise --user dotty >> >> but if the sender is the same as the recipient, then dspam won't analyze >> it. I haven't found a configuration option for this issue yet. >> >> Do You have any advice? >> > 1) Can you post the full header of the message dspam does deliver after > not scanning the message? I'm not in front of that machine, but i can tell that there were not a single header from dspam. So the message wasn't through dspam. > > 2) Could you post some log entries showing that dspam did not or did > processed the message? I can't see anything in the logs, even in the procmail log files. Maybe somehow the procmail filter won't process a message from 'myself'? > > > btw: There is no such switch in the dsapm configuration to disable > complete scanning of messages. Every message get's tagged in the headers > by dspam. So if it is processed by dspam, then there must be a entry in > the headers. If there is no entry in the headers, then you have not > processed the message with dspam. Yes, that's what i thought. Maybe i have to look around procmail to figure out the problem. > > >> Thank You! >> -- >> Gabri Mate >> > Steve > >> ------------------------------------------------------------------------------ >> Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are >> powering Web 2.0 with engaging, cross-platform capabilities. Quickly and >> easily build your RIAs with Flex Builder, the Eclipse(TM)based >> development >> software that enables intelligent coding and step-through debugging. >> Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com >> _______________________________________________ >> Dspam-devel mailing list >> Dsp...@li... >> https://lists.sourceforge.net/lists/listinfo/dspam-devel > > -- > Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: > http://www.gmx.net/de/go/multimessenger01 > > ------------------------------------------------------------------------------ > Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are > powering Web 2.0 with engaging, cross-platform capabilities. Quickly and > easily build your RIAs with Flex Builder, the Eclipse(TM)based development > software that enables intelligent coding and step-through debugging. > Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com > _______________________________________________ > Dspam-devel mailing list > Dsp...@li... > https://lists.sourceforge.net/lists/listinfo/dspam-devel > > !DSPAM:117,49becdb0263228477911171! > > > |
From: MKondrin <mko...@hp...> - 2009-03-17 10:24:57
|
Jyothsni Reddy wrote: > Hi, > > Does dspam do stemming? > No. > Thanks, > Jyothsni > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------------ > Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are > powering Web 2.0 with engaging, cross-platform capabilities. Quickly and > easily build your RIAs with Flex Builder, the Eclipse(TM)based development > software that enables intelligent coding and step-through debugging. > Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com > ------------------------------------------------------------------------ > > _______________________________________________ > Dspam-devel mailing list > Dsp...@li... > https://lists.sourceforge.net/lists/listinfo/dspam-devel > |
From: Jyothsni R. <jyo...@gm...> - 2009-03-17 09:14:07
|
Hi, Does dspam do stemming? Thanks, Jyothsni |
From: Steve <ste...@gm...> - 2009-03-16 22:01:05
|
-------- Original-Nachricht -------- > Datum: Mon, 16 Mar 2009 20:55:26 +0100 > Von: Gabri Mate <mai...@mo...> > An: dsp...@li... > Betreff: [Dspam-devel] forged from header - not scanning > Hey there! > Hello. > I've recently noticed that if i recieve a spam with a forged header, so > it appears that i have sent it that mail to myself from the address in > question, then dspam won't process it. > > I have an account which i check with fetchmail and i pass the message to > dspam through procmail: > > #dspam > :0fw > | /usr/local/bin/dspam --stdout --deliver=innocent,spam --mode=teft > --feature=noise --user dotty > > but if the sender is the same as the recipient, then dspam won't analyze > it. I haven't found a configuration option for this issue yet. > > Do You have any advice? > 1) Can you post the full header of the message dspam does deliver after not scanning the message? 2) Could you post some log entries showing that dspam did not or did processed the message? btw: There is no such switch in the dsapm configuration to disable complete scanning of messages. Every message get's tagged in the headers by dspam. So if it is processed by dspam, then there must be a entry in the headers. If there is no entry in the headers, then you have not processed the message with dspam. > Thank You! > -- > Gabri Mate > Steve > ------------------------------------------------------------------------------ > Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are > powering Web 2.0 with engaging, cross-platform capabilities. Quickly and > easily build your RIAs with Flex Builder, the Eclipse(TM)based development > software that enables intelligent coding and step-through debugging. > Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com > _______________________________________________ > Dspam-devel mailing list > Dsp...@li... > https://lists.sourceforge.net/lists/listinfo/dspam-devel -- Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01 |
From: Gabri M. <mai...@mo...> - 2009-03-16 19:55:47
|
Hey there! I've recently noticed that if i recieve a spam with a forged header, so it appears that i have sent it that mail to myself from the address in question, then dspam won't process it. I have an account which i check with fetchmail and i pass the message to dspam through procmail: #dspam :0fw | /usr/local/bin/dspam --stdout --deliver=innocent,spam --mode=teft --feature=noise --user dotty but if the sender is the same as the recipient, then dspam won't analyze it. I haven't found a configuration option for this issue yet. Do You have any advice? Thank You! -- Gabri Mate |
From: Vadim Z. <vz-...@ze...> - 2009-03-16 10:25:15
|
[Johannes, I hope it's ok to cc you this as your contact instructions say to do it for dovecot-antispam emails sent to dovecot mailing list and while I send this one to dspam list I hope you still would find it relevant] Hello, I've been using dspam with Dovecot and its antispam plugin for quite some time without any problems but since relatively recently reclassification of false negatives as spams stopped working and, of course, the quality of spam detection started to suffer. The symptom of the problem is the following log message in syslog: dspam[pid]: Unable to attach DSPAM context dspam[pid]: process_message returned error -2. dropping message. As there is no information about what exactly went wrong there I rebuilt dspam with debug info and found that the real cause of the problem was that mmap() in _hash_drv_open() was failing with ENOMEM ("Cannot allocate memory") which explains why does it fail. Unfortunately why does mmap() itself fail is a complete mystery to me. Here are some of my findings: 1. There is plenty (in excess of a couple gigabytes, compared to ~250MB size of the .css file being mapped) of available physical and virtual memory on this Linux amd64 system (see below for full details). 2. A simple test program calling mmap() on the same .css file works (its full source is included at the end of the message). 3. Even more surprisingly, dspam itself works fine when called from command line as "dspam --source=error --class=spam --signature=... --user zeitlin" 4. Unfortunately the test program still works even when called from antispam plugin so it's not Dovecot/antispam "fault" neither. 5. As a final data point, cssstats dspam tool which also uses the same _hash_drv_open() succeeds -- when called from command line -- too. I have absolutely no ideas about what could explain this behaviour. FWIW I'm not the only one seeing it, see e.g. http://www.shapeshifter.se/2009/03/03/dspam-converting-from-the-hash-driver-to-mysql-driver/ which speaks about exactly the same error in a very similar configuration and shows that it's not specific to Linux as this post is about FreeBSD. And there are plenty of other reports of "Unable to attach DSPAM context" which might be due to this problem (but it could also be due to permission problems of course...). The trouble is that nobody, including the author of the blog post above, found the real reason for mmap() failure. Personally I thought that, as mmap() mappings are inherited by the child process, maybe dovecot or antispam plugin might be mmap()-ing something huge into the process address space and not munmap()-ing it before spawning dspam. But there is nothing like this in /proc/<imap-pid>/maps and I even tried setting "mmap_disable = yes" in /etc/dovecot/dovecot.conf without any effect (well, I did see that maps for the .index file disappeared from /proc/pid/maps, so it was taken into account, it just didn't change anything for the mmap() call failure in dspam). And I can't think of any other reason for mmap() to fail with ENOMEM. For the reference, here are the full details of my system: - OS: Debian GNU/Linux 4.0 on x86_64 - Kernel: 2.6.20 - DSPAM: latest from git - Dovecot: 1.0.13-1~bpo40+1 (Debian backports repository) - Dovecot-antispam from git (behaviour was the same with 1.0 I used before) Dovecot configuration is pretty standard, here is the output of "dovecot -n": # 1.0.13: /etc/dovecot/dovecot.conf log_timestamp: %Y-%m-%d %H:%M:%S protocols: imap login_dir: /var/run/dovecot/login login_executable: /usr/lib/dovecot/imap-login login_greeting_capability: yes mail_privileged_group: mail mail_location: mbox:%h/mail:INBOX=/var/mail/%u mbox_lazy_writes: no mail_plugins: antispam auth default: passdb: driver: pam userdb: driver: passwd plugin: antispam_signature: X-DSPAM-Signature antispam_spam: Junk antispam_dspam_binary: /usr/local/bin/dspam antispam_dspam_args: --user;%n Does anybody have any ideas about what is going on here? Thanks in advance for any ideas! VZ Appendix: Source of the test program used to test mmap() which can be executed from dovecot-antispam plugin by setting antispam_dspam_binary to its path and antispam_dspam_args to the name of the .css file to map: #include <stdio.h> #include <errno.h> #include <string.h> #include <unistd.h> #include <fcntl.h> #include <syslog.h> #include <sys/mman.h> #include <sys/types.h> static const int logprio = LOG_LOCAL0 | LOG_DEBUG; int main(int argc, char **argv) { int fd; off_t len; const char *fname; if ( argc == 1 ) { /* ignore any preceding arguments to allow calling it as dspam */ syslog(logprio, "Usage: %s [...whatever...] <file>", argv[0]); return 2; } fname = argv[argc - 1]; fd = open(fname, O_RDWR); if ( fd == -1 ) { syslog(logprio, "open(\"%s\") failed: %s", fname, strerror(errno)); return 1; } len = lseek(fd, 0, SEEK_END); if ( len == -1 ) { syslog(logprio, "lseek(SEEK_END) failed: %s", strerror(errno)); return 1; } if ( mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0) == MAP_FAILED) { syslog(logprio, "mmap() failed: %s", strerror(errno)); return 1; } syslog(logprio, "mmap(\"%s\", 0..%ld) succeeded.\n", fname, len); return 0; } |
From: Paul C. <ds...@cy...> - 2009-02-27 10:28:00
|
Ismail YENIGUL wrote: > Hi, > > I could not see any source code in sf.net svn. > Where can I get the latest code to test? > > Thanks. > > > Take a look at the wiki page: http://apps.sourceforge.net/mediawiki/dspam/index.php?title=Main_Page You can find the instructions for downloading source here: http://apps.sourceforge.net/mediawiki/dspam/index.php?title=Developing Hope this helps, |
From: Ismail Y. <ism...@en...> - 2009-02-26 23:05:10
|
Hi, I could not see any source code in sf.net svn. Where can I get the latest code to test? Thanks. -- Ismail YENIGUL Endersys Ltd. Phone :+90 216-4709423 | Mobile:+90 533 747 36 65 Fax :+90 216-4709508 | web: http://www.endersys.com.tr |
From: Andreas S. <ma...@cy...> - 2009-02-11 17:38:10
|
On Wednesday 11 February 2009 18:32:21 Reto Glauser wrote: > Is this the active, most recent dev-mailing list of dspam? If so, can > someone post some information about the project? Where is its home, what > happened during the last couple of month, what are the plans etc. > > I found http://sourceforge.net/projects/dspam/ > > which shows 6 project members and 11 feature requests - but that's about > it. Who does what and what is still missing? > > I read something about a public git access, but where? > I read something about introducing CMake - is this still wanted? If you want CMake support, I could do in in half a day. -- andreas > > R, > r > > --------------------------------------------------------------------------- >--- Create and Deploy Rich Internet Apps outside the browser with > Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing > skills and code to build responsive, highly engaging applications that > combine the power of local resources and data with the reach of the web. > Download the Adobe AIR SDK and Ajax docs to start building applications > today-http://p.sf.net/sfu/adobe-com > _______________________________________________ > Dspam-devel mailing list > Dsp...@li... > https://lists.sourceforge.net/lists/listinfo/dspam-devel |
From: Reto G. <li...@bl...> - 2009-02-11 17:32:28
|
Is this the active, most recent dev-mailing list of dspam? If so, can someone post some information about the project? Where is its home, what happened during the last couple of month, what are the plans etc. I found http://sourceforge.net/projects/dspam/ which shows 6 project members and 11 feature requests - but that's about it. Who does what and what is still missing? I read something about a public git access, but where? I read something about introducing CMake - is this still wanted? R, r |
From: Eugene <ge...@ge...> - 2009-02-08 15:00:52
|
Hi Hugo, Thanks for the pointer, I subscribed to that list too. I guess our messages would serve as a pointer to the new list for other people =) Best wishes Eugene |
From: Eric P. <epa...@in...> - 2009-02-04 14:21:58
|
-- Eric Paumerat Iniflux 5, rue du Chevalier de St George Paris 75008 Std: 0142603665 Fax: 0142963901 Mob: 0650705349 |
From: Jyothsni R. <jyo...@gm...> - 2009-02-04 05:44:30
|