Re: [Assp-test] Assp dying
Masood, warning: got unexpected signal SEGV. These error are most related to the regex engine of Perl. Maybe the reason is a very high (100%) CPU workload over some time. Worker_2: Can't call method fetchrow_arrayref on an undefined value at The MySQL server was not able to do a SQL-select statement. Is the MySQL server running on the same system ? Maybe Perl (assp) , MySQL, ClamAV and possibly some other processes are fighting for system resources!? It looks like the system is going out of system resources (memory - CPU-time). Keep in mind (as I've told you) - running 10 SMTP-Workers (13 threads) + MySQL + ClamAV on a very high workload system will need a (dual? ) quadcore system (switch off hyperthreading - this will produce an unneeded overhead with Perl)! How much memory is used by assp? What is the average CPU usage of assp in relation to the whole system? Thomas Von:Masood Rahim mas...@intertune.com An: assp-test@lists.sourceforge.net Datum: 17.03.2010 18:31 Betreff:Re: [Assp-test] Assp dying So after almost 20 hours of testing with using the mysql database instead of the flatfiles and the BerkeleyDB options for caching, the assp v2 process on this box that I am testing it on is still dying. It has died twice. There is an exception.log file but it doesn't seem to provide any information around the time the process died. I just see many entries for the following: Mar-16-10 12:37:33 Error: Worker_2: Can't call method fetchrow_arrayref on an undefined value at /usr/local/lib/perl5/site_perl/5.10.1/Tie/RDBM.pm line 166. I did find the following debugSignal.txt file which looks like it does contain some errors from right around the time before the server crashed: Mar-08-10 16:00:38 warning: got unexpected signal ALRM in Worker_10001: package - main, file - assp.pl, line - 38606! Mar-08-10 18:15:00 warning: got unexpected signal ALRM in Main_Thread: package - main, file - assp.pl, line - 37079! Mar-11-10 20:36:57 warning: got unexpected signal SEGV in Worker_3: package - main, file - assp.pl, line - 17523! Mar-13-10 00:36:01 warning: got unexpected signal ALRM in Main_Thread: package - Encode::Guess, file - /usr/local/lib/perl5/5.10.1/i686-linux/Encode/Guess.pm, line - 139! Mar-13-10 00:37:33 warning: got unexpected signal ALRM in Main_Thread: package - IO::Socket::INET, file - /usr/local/lib/perl5/5.10.1/i686-linux/IO/Socket/INET.pm, line - 270! Mar-13-10 00:38:29 warning: got unexpected signal ALRM in Main_Thread: package - Encode::Guess, file - /usr/local/lib/perl5/5.10.1/i686-linux/Encode/Guess.pm, line - 134! Mar-13-10 01:44:25 warning: got unexpected signal SEGV in Worker_10: package - main, file - assp.pl, line - 18160! Mar-13-10 05:15:18 warning: got unexpected signal SEGV in Worker_4: package - main, file - assp.pl, line - 17554! Mar-13-10 08:39:14 warning: got unexpected signal SEGV in Worker_9: package - main, file - assp.pl, line - 17554! Mar-15-10 11:48:50 warning: got unexpected signal SEGV in Worker_9: package - main, file - assp.pl, line - 17554! Mar-16-10 06:34:40 warning: got unexpected signal SEGV in Worker_3: package - Email::Simple, file - /usr/local/lib/perl5/site_perl/5.10.1/Email/Simple.pm, line - 89! Mar-17-10 09:52:32 warning: got unexpected signal SEGV in Worker_1: package - main, file - assp.pl, line - 17907! Any help is greatly appreciated, I am out of ideas and hope the information above helps. Thanks, Masood -Original Message- From: assp-test-requ...@lists.sourceforge.net [mailto:assp-test-requ...@lists.sourceforge.net] Sent: Wednesday, March 17, 2010 4:41 AM To: assp-test@lists.sourceforge.net Subject: Assp-test Digest, Vol 32, Issue 33 Send Assp-test mailing list submissions to assp-test@lists.sourceforge.net To subscribe or unsubscribe via the World Wide Web, visit https://lists.sourceforge.net/lists/listinfo/assp-test or, via email, send a message with subject or body 'help' to assp-test-requ...@lists.sourceforge.net You can reach the person managing the list at assp-test-ow...@lists.sourceforge.net When replying, please edit your Subject line so it is more specific than Re: Contents of Assp-test digest... Today's Topics: 1. Re: Assp dying (Thomas Eckardt) 2. Re: DB_File (Thomas Eckardt) 3. Re: Assp dying (GrayHat) 4. Re: DB_File (GrayHat) 5. Re: DB_File (Scott MacLean) 6. Re: Assp dying (Scott MacLean) -- Message: 1 Date: Wed, 17 Mar 2010 06:45:09 +0100 From: Thomas Eckardt thomas.ecka...@thockar.com Subject: Re: [Assp-test] Assp dying To: ASSP development mailing list assp-test@lists.sourceforge.net Message-ID: titc.3692ca544f.of60da4747.bdb0ae74-onc12576e9.001f8437-c12576e9.001f9...@t hockar.com Content-Type: text/plain; charset=us-ascii Inside this GZip archive is a file named assp.pl, which unpacks to 396,392 bytes. I'm
Re: [Assp-test] Assp dying
I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
[Assp-test] Mail meets message scoring lower limit but is not tagged.
The headers don¹t say that the score was less than the lower limit and show no scoring other than than the bayesian, which meets the lower limit. The sender wasn¹t whitelisted. Any ideas why it wasn¹t tagged and was marked as ok mail? I¹m running 1.7.1.2 10 Mar-18-10 07:58:06 13485-23265 208.46.178.158 jaksk...@reference.com to: sh...@fredco-md.net [scoring] Bayesian Check - Prob: 1.0 = spam; Mar-18-10 07:58:07 13485-23265 [MessageOK] 208.46.178.158 jaksk...@reference.com to: sh...@fredco-md.net -- Message OK -- [You have 1 new message] - okmail/You_have_1_new_message__61.eml; X-Assp-Delay: sh...@fredco-md.net was delayed for 12m 1s; 18 Mar 2010 07:58:05 -0400 X-Assp-Received-SPF: none ip=208.46.178.158 mailfrom=jaksk...@reference.com helo=arlington-greene.com X-Assp-Score: 45 (Bayesian Probability: 1.) Received: from arlington-greene.com ([208.46.178.158] helo=arlington-greene.com) with IPv4:25 by mailgw2.frederickcountymd.gov; 18 Mar 2010 07:58:04 -0400 Received: from reference.com ([72.17.202.50]) by arlington-greene.com with Microsoft SMTPSVC(6.0.3790.3959); Thu, 18 Mar 2010 07:46:03 -0400 From: HSBC Bank jaksk...@reference.com Paul K. Dickson Systems Administrator Interagency Information Technologies Frederick County Government Frederick, MD 21701 pdick...@frederickcountymd.gov 301-600-2399/x12399 -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
Re: [Assp-test] Mail meets message scoring lower limit but is not tagged.
ASSP development mailing list assp-test@lists.sourceforge.net schreibt: The headers dont say that the score was less than the lower limit and show no scoring other than than the bayesian, which meets the lower limit. The sender wasnt whitelisted. The GUI says: MessageScoring will tag messages with totalscore over this limit -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
Re: [Assp-test] Assp dying
Fritz I will try to get the information for you on other v1 versions that I have installed. Thomas I have offloaded the DNS and MYSQL to separate servers now. This server is only running ASSP. I have reduced it down to the default 5 threads. The external dns server I am using on my local network has ample RAM and CPU speed. And the Mysql server I am using has xeon processors with 4 gb of ram. There should be no reasons why assp is dying now on this particular server. If it does die now it's either related to one of the following: 1) a problem with the assp code 2) a problem with one of the perl modules being used 3) assp v2 cannot support 275,000 connections on a dual penitum 600Mhz system with 512 MB RAM By the way I don't even do ClamAV or any other virus scanning on this assp server. Masood -Original Message- From: assp-test-requ...@lists.sourceforge.net [mailto:assp-test-requ...@lists.sourceforge.net] Sent: Wednesday, March 17, 2010 11:27 PM To: assp-test@lists.sourceforge.net Subject: Assp-test Digest, Vol 32, Issue 35 Send Assp-test mailing list submissions to assp-test@lists.sourceforge.net To subscribe or unsubscribe via the World Wide Web, visit https://lists.sourceforge.net/lists/listinfo/assp-test or, via email, send a message with subject or body 'help' to assp-test-requ...@lists.sourceforge.net You can reach the person managing the list at assp-test-ow...@lists.sourceforge.net When replying, please edit your Subject line so it is more specific than Re: Contents of Assp-test digest... Today's Topics: 1. Re: Assp dying (Fritz Borgstedt) 2. Re: DB_File (Scott MacLean) 3. Re: Assp dying (Thomas Eckardt) -- Message: 1 Date: Wed, 17 Mar 2010 19:16:55 +0100 From: Fritz Borgstedt f...@iworld.de Subject: Re: [Assp-test] Assp dying To: ASSP development mailing list assp-test@lists.sourceforge.net Message-ID: sig.1692b9268e.fc.000f455506dfc44a3b9aca00628e129b.6dfc...@iworld.de Content-Type: text/plain; charset=ISO-8859-1 ASSP development mailing list assp-test@lists.sourceforge.net schreibt: So after almost 20 hours of testing with using the mysql database instead of the flatfiles and the BerkeleyDB options for caching, the assp v2 process on this box that I am testing it on is still dying. It has died twice. Would you please inform me what information you have about your tries to run newer versions of ASSP V1. (Newer as 1.4.3.1). It seems obvious (at least to me) that your problems started already in version ASSP V1 when you tried to run newer versions. 1.4.3.1 was a development version and is now nearly 2 years old. ASSP V1 with versions above 1.4.3.1 is running in several thousands of installations. My recommendation would be to start with the newest version of ASSP V1 to see what happens. It is even easier to go from there to ASSP V2, because the feature set is much more similar. -- Message: 2 Date: Wed, 17 Mar 2010 15:53:00 -0400 From: Scott MacLean a...@hollsco.com Subject: Re: [Assp-test] DB_File To: Matti Haack m.ha...@haack-it.de, ASSP development mailing list assp-test@lists.sourceforge.net Message-ID: bcf4801c-cd66-47d1-bb6c-4f1a5811f...@mail.frogstar.com Content-Type: text/plain; charset=iso-8859-1; format=flowed Wow. Answers in a wiki? Next you'll come up with a list of answers to questions that are asked frequently. Hmm, what could you call such a thing...? :) Thanks. :) At 08:29 AM 3/17/2010, Matti Haack wrote: yes, ASSP will import files to the MySQL DB! Just have a look at the Wiki (sometimes, you will find help there :) http://sourceforge.net/apps/mediawiki/assp/index.php?title=Supported_da tabases Matti Thomas, Is there an elegant way to move a server from flat files to MSSQL (the server has a fast MSSQL server available, so I'd like to use it) without losing data? At 01:55 AM 3/17/2010, Thomas Eckardt wrote: This depend on the size of the flat files. Use the RDBM (SQL) DB of your choice (MySQL,MSSQL,Oracle,ProstgreSQL,DB2) for the main lists and hashes and BerkeleyDB for the others. In the past we had some problems using BerkeleyDB for the main hashes - but this is fixed. Using BerkeleyDB for the main hashes needs more DB knowledge, if more than 7 SMTP-Workers are running. Thomas -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Matti Haack - Hit Haack IT Service Gmbh
Re: [Assp-test] Assp dying
ASSP development mailing list assp-test@lists.sourceforge.net schreibt: Fritz I will try to get the information for you on other v1 versions that I have installed. No hurry about that. ASSP V1 is running robust and stable in several thousand installations. As I said before, for me it is not rational to jump onto V2 from 1.4.3.1 without being able to run never versions of V1 during a timeframe of 2 years. -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
Re: [Assp-test] Assp dying
I'm going to try this again, but I'm going to get rid of the start /wait, so that the perl process outputs its entire console output (as well as redirecting the STDERR output) into the debug file instead of opening in a second window, and see if there is something to see that way. At 07:50 AM 3/18/2010, Scott MacLean wrote: I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
Re: [Assp-test] Assp dying
AHA. This one did the trick. I logged several hours of resets, and the same thing shows each time: Mar-18-10 16:01:15 [Main_Thread] Info: Main_Thread got connection request Out of memory! Callback called exit at assp.pl line 17570. Now this is interesting...because the server has lots ( 2GB physical) free memory available while ASSP is running, and did at the time of the error, as well. Is there a separate constraint within ASSP that is limiting its memory use, in a way that will cause Perl itself to fault this way? At 01:21 PM 3/18/2010, Scott MacLean wrote: I'm going to try this again, but I'm going to get rid of the start /wait, so that the perl process outputs its entire console output (as well as redirecting the STDERR output) into the debug file instead of opening in a second window, and see if there is something to see that way. At 07:50 AM 3/18/2010, Scott MacLean wrote: I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
[Assp-test] URIBL Question
I noticed that the dbl.spamhaus.org had been added to the default for the URIBL. So I decided to update it back to default and give that new one a try. When I did that I started sending some test messages through from my gmail account with a list of domains in the message and I turned the logging on to verbose for URIBL to see what it gets. When I did that, I saw it kick me out a false positive on newegg.com of all places. Finding this odd I went and checked on spamhaus.org in their domain lookup tool (http://www.spamhaus.org/lookup.lasso) and it showed that newegg.com is not listed. I am using DNS servers internally in my network, DNS isn't installed locally on the same machine in case you were wondering. So I reverted back to my old config without the dbl.spamhaus.org in it and everything seems fine, I still get hits properly on the other 2 lists, but I am not sure why this particular list throws up a false positive. Here is a snip of my log from the message in question: Mar-18-10 17:40:47 59246-01086 [URIBL] 74.125.83.46 mygmailtesta...@gmail.com to: mylocaltesta...@myemaildomain.com info: URI blackberry.com found in text; Mar-18-10 17:40:47 59246-01086 [URIBL] 74.125.83.46 mygmailtesta...@gmail.com to: mylocaltesta...@myemaildomain.com info: URI newegg.com found in text; Mar-18-10 17:40:47 Sending DNS-query to 10.10.10.110 on multi.surbl.org for URIBL checks on newegg.com; Mar-18-10 17:40:47 Sending DNS-query to 10.10.10.110 on black.uribl.com for URIBL checks on newegg.com; Mar-18-10 17:40:47 Sending DNS-query to 10.10.10.110 on dbl.spamhaus.org for URIBL checks on newegg.com; Mar-18-10 17:40:47 Commencing URIBL checks on newegg.com; Mar-18-10 17:40:48 59246-01086 [URIBL] 74.125.83.46 mygmailtesta...@gmail.com to: mylocaltesta...@myemaildomain.com [spam found] -- URIBL: fail, newegg.com listed in 1dbl.spamhaus.org-208.67.216.132; -- [test domains] - spam/1086.eml; Here is the config portion with the URIBL information in it: ValidateURIBL:=1 URIBLWL:= URIBLNP:= URIBLLocal:= URIBLISP:=1 URIBLServiceProvider:=multi.surbl.org|black.uribl.com URIBLmaxreplies:=3 URIBLmaxhits:=1 URIBLmaxtime:=10 URIBLsocktime:=1 URIBLCCTLDS:=file:files/URIBLCCTLDS.txt URIBLmaxdomains:=10 URIBLwhitelist:=doubleclick.net noURIBL:= AddURIBLHeader:=1 URIBLCacheExp:=24 URIBLLog:=2 URIBLError:=554 5.7.1 Blacklisted by URIBLNAME Contact the postmaster of this domain for resolution. Thanks, Marcus -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
Re: [Assp-test] Assp dying
Doing some research into this Out of memory! error from Perl, it apparently only happens when a malloc() call within Perl fails, because memory it wants is not available. That said, my ASSP normally trundles along with about 2 GB of physical memory available, and about the same amount of virtual memory available. My MRTG logs don't show a sudden ramp up at the time of the crashes, in fact, they don't ever show anything other than a straight line in terms of memory usage. Unless a function within ASSP is suddenly causing Perl to say, hey, I need 4 GB of memory, right now! I can't see why Perl would be failing with this error. I'm at a loss. At 06:20 PM 3/18/2010, Scott MacLean wrote: AHA. This one did the trick. I logged several hours of resets, and the same thing shows each time: Mar-18-10 16:01:15 [Main_Thread] Info: Main_Thread got connection request Out of memory! Callback called exit at assp.pl line 17570. Now this is interesting...because the server has lots ( 2GB physical) free memory available while ASSP is running, and did at the time of the error, as well. Is there a separate constraint within ASSP that is limiting its memory use, in a way that will cause Perl itself to fault this way? At 01:21 PM 3/18/2010, Scott MacLean wrote: I'm going to try this again, but I'm going to get rid of the start /wait, so that the perl process outputs its entire console output (as well as redirecting the STDERR output) into the debug file instead of opening in a second window, and see if there is something to see that way. At 07:50 AM 3/18/2010, Scott MacLean wrote: I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev
Re: [Assp-test] Assp dying
I've now got 7 examples logged of this problem occurring, and every time, it's line 17570 that is causing the out of memory error. This line is: while (time - $itime = $maxBombSearchTime $text =~ s/($regex)//) { At 10:09 PM 3/18/2010, Scott MacLean wrote: Doing some research into this Out of memory! error from Perl, it apparently only happens when a malloc() call within Perl fails, because memory it wants is not available. That said, my ASSP normally trundles along with about 2 GB of physical memory available, and about the same amount of virtual memory available. My MRTG logs don't show a sudden ramp up at the time of the crashes, in fact, they don't ever show anything other than a straight line in terms of memory usage. Unless a function within ASSP is suddenly causing Perl to say, hey, I need 4 GB of memory, right now! I can't see why Perl would be failing with this error. I'm at a loss. At 06:20 PM 3/18/2010, Scott MacLean wrote: AHA. This one did the trick. I logged several hours of resets, and the same thing shows each time: Mar-18-10 16:01:15 [Main_Thread] Info: Main_Thread got connection request Out of memory! Callback called exit at assp.pl line 17570. Now this is interesting...because the server has lots ( 2GB physical) free memory available while ASSP is running, and did at the time of the error, as well. Is there a separate constraint within ASSP that is limiting its memory use, in a way that will cause Perl itself to fault this way? At 01:21 PM 3/18/2010, Scott MacLean wrote: I'm going to try this again, but I'm going to get rid of the start /wait, so that the perl process outputs its entire console output (as well as redirecting the STDERR output) into the debug file instead of opening in a second window, and see if there is something to see that way. At 07:50 AM 3/18/2010, Scott MacLean wrote: I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net
Re: [Assp-test] Assp dying
I added this line just before line 17570 that was causing the exception: mlog (0,%%%DEBUG: itime= $itime maxBombSearchTime = $maxBombSearchTime text = $text regex = $regex); I then let it run for a bit, until it crashed again with an out of memory error. The result, extracted from my log file (spat out by the line above), can be seen here: http://ilovewbpf.com/memory.txt The $text contains what is clearly a spam email. The $regex, is, I think what is causing the problem. It is 7,681 characters long - the contents of my bombRe- and when this is being combined with a relatively large email, it's just more than the Perl regex engine can handle. I'm going to reduce the contents of my bombRe and see if this solves the problem. At 12:22 AM 3/19/2010, Scott MacLean wrote: I've now got 7 examples logged of this problem occurring, and every time, it's line 17570 that is causing the out of memory error. This line is: while (time - $itime = $maxBombSearchTime $text =~ s/($regex)//) { At 10:09 PM 3/18/2010, Scott MacLean wrote: Doing some research into this Out of memory! error from Perl, it apparently only happens when a malloc() call within Perl fails, because memory it wants is not available. That said, my ASSP normally trundles along with about 2 GB of physical memory available, and about the same amount of virtual memory available. My MRTG logs don't show a sudden ramp up at the time of the crashes, in fact, they don't ever show anything other than a straight line in terms of memory usage. Unless a function within ASSP is suddenly causing Perl to say, hey, I need 4 GB of memory, right now! I can't see why Perl would be failing with this error. I'm at a loss. At 06:20 PM 3/18/2010, Scott MacLean wrote: AHA. This one did the trick. I logged several hours of resets, and the same thing shows each time: Mar-18-10 16:01:15 [Main_Thread] Info: Main_Thread got connection request Out of memory! Callback called exit at assp.pl line 17570. Now this is interesting...because the server has lots ( 2GB physical) free memory available while ASSP is running, and did at the time of the error, as well. Is there a separate constraint within ASSP that is limiting its memory use, in a way that will cause Perl itself to fault this way? At 01:21 PM 3/18/2010, Scott MacLean wrote: I'm going to try this again, but I'm going to get rid of the start /wait, so that the perl process outputs its entire console output (as well as redirecting the STDERR output) into the debug file instead of opening in a second window, and see if there is something to see that way. At 07:50 AM 3/18/2010, Scott MacLean wrote: I tried this (although I changed delete to del to make sure it functioned correctly). It died three times overnight and duly restarted itself - but did not log anything to asspdbg.log. It is a zero-byte file. assprst.log held exactly what you expect: Thu 03/18/2010 0:54:04.13 Thu 03/18/2010 3:00:51.20 Thu 03/18/2010 7:37:27.56 At 10:12 AM 3/17/2010, GrayHat wrote: @echo off pushd c:\assp :LOOP start /wait perl assp.pl asspdbg.log 21 echo %DATE% %TIME% assprst.log delete pid goto LOOP that should pretty much solve the issue, just run ASSP through the above script, then check the resulting asspdbg.log file to find out any error message(s) (probably coming from a perl module) that would help pinpointing the cause of the issue; the assprst.log file will also contains all dates/times of restarts so it may ease searching through the log file --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test --- --- Download Intel#174; Parallel Studio Eval