Re: Apache logs problem
Thanks for the tip Ceki ! * Yes, I am using the -p flag (incremental) * I did not know about the -f one and just checked the man page. It seems that by using it webalizer will process all logs from previous dates as if had the time/date of the last known process log. Let's imagine a cron job for day D : it will need to process D-1.gz (not too easy because we need to consider when D is at the end of the month - but doable) and D.gz. For D-1.gz, no -f flag should be used. For D.gz, I will need to use the D.gz flag but then won't all logs accumulate for the time of the last entry in D-1.gz, instead of the correct times at which these log happened ? Also, it looks a bit complicated to me. It would seem normal that D.gz would contain all logs for the D day (as was happening before the past month). So, I would prefer to correct that behaviour instead of trying to accomodate this [unless someone tells me this behaviour is actually correct for such and such reason]. * Do you know who is managing daedalus logs so that I can contact him and solve this out ? Thanks -Vincent - Original Message - From: Ceki Gülcü [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, October 01, 2001 1:11 PM Subject: Re: Apache logs problem Hello Vincent, Assuming you are using webalizer, have you tried using the -f (fold sequence errors) and the -p (preserve state) flags? Regards, Ceki At 11:39 29.09.2001 +0100, Vincent Massol wrote: It seems there is a problem of dates with the generated apache logs in /x2/logarchive/www/2001 For example if you look at 2001/09/24.gz, the last entry is dated 25/Sep/2001:00:21:09 and the first entry of 2001/09/25.gz is dated 25/Sep/2001:00:00:04 !! Now, even worse, I have tried to find in 24.gz the first entry that is in 25.gz with no success. It is not there ! It means there is a problem of dates and the entries do not seem to be duplicate ... !?? This is a big problem for handling logs, especially for generating statistics. This problem is quite new (maybe 1 month at most) and it worked fine before. Can anyone help ? Thanks -Vincent Massol - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Ceki Gülcü - http://qos.ch The world owes Israel a great debt for destroying Saddam's French built nuclear reactor in 1981 and thus preventing nuclear blackmail in the region and perhaps beyond. -- Garry Kasparov (yes, the chess player) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Apache logs problem
Hello Vincent, At 13:57 01.10.2001 +0100, Vincent Massol wrote: Thanks for the tip Ceki ! Most welcome. * Yes, I am using the -p flag (incremental) OK. * I did not know about the -f one and just checked the man page. It seems that by using it webalizer will process all logs from previous dates as if had the time/date of the last known process log. Let's imagine a cron job for day D : it will need to process D-1.gz (not too easy because we need to consider when D is at the end of the month - but doable) and D.gz. For D-1.gz, no -f flag should be used. For D.gz, I will need to use the D.gz flag but then won't all logs accumulate for the time of the last entry in D-1.gz, instead of the correct times at which these log happened ? Also, it looks a bit complicated to me. It would seem normal that D.gz would contain all logs for the D day (as was happening before the past month). So, I would prefer to correct that behaviour instead of trying to accomodate this [unless someone tells me this behaviour is actually correct for such and such reason]. Right, it's a pain. On the other hand, doesn't webalizer ignore out of order entries? Due to out of order entries, your stats will not be 100% accurate but 99.95% should be good enough for our purposes... * Do you know who is managing daedalus logs so that I can contact him and solve this out ? It's probably Brian but I do not know for sure. I am very curious about the reason behind the disordered log entires. Thanks -Vincent - Original Message - From: Ceki Gülcü [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, October 01, 2001 1:11 PM Subject: Re: Apache logs problem Hello Vincent, Assuming you are using webalizer, have you tried using the -f (fold sequence errors) and the -p (preserve state) flags? Regards, Ceki At 11:39 29.09.2001 +0100, Vincent Massol wrote: It seems there is a problem of dates with the generated apache logs in /x2/logarchive/www/2001 For example if you look at 2001/09/24.gz, the last entry is dated 25/Sep/2001:00:21:09 and the first entry of 2001/09/25.gz is dated 25/Sep/2001:00:00:04 !! Now, even worse, I have tried to find in 24.gz the first entry that is in 25.gz with no success. It is not there ! It means there is a problem of dates and the entries do not seem to be duplicate ... !?? This is a big problem for handling logs, especially for generating statistics. This problem is quite new (maybe 1 month at most) and it worked fine before. Can anyone help ? Thanks -Vincent Massol - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Ceki Gülcü - http://qos.ch The world owes Israel a great debt for destroying Saddam's French built nuclear reactor in 1981 and thus preventing nuclear blackmail in the region and perhaps beyond. -- Garry Kasparov (yes, the chess player) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Ceki Gülcü - http://qos.ch The world owes Israel a great debt for destroying Saddam's French built nuclear reactor in 1981 and thus preventing nuclear blackmail in the region and perhaps beyond. -- Garry Kasparov (yes, the chess player) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
BCEL @ apache
Hi, Many months ago I asked Markus Dahm whether or not he would be interested in BCEL (Byte Code Engineering tool) being a Jakarta project as I was using and thought it a very cool package, and I believe Xalan uses it for their compiler, but nothing resulted from the initial conversation. Markus has approached me now stating that he would be interested in BCEL becoming a Jakarta project so I thought I would see what others thought. BCEL has many applications including compilers, analysis tools and who knows what else. Thoughts? -- jvz. Jason van Zyl http://tambora.zenplex.org http://jakarta.apache.org/turbine http://jakarta.apache.org/velocity http://jakarta.apache.org/alexandria http://jakarta.apache.org/commons - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: BCEL @ apache
Jason van Zyl wrote: Many months ago I asked Markus Dahm whether or not he would be interested in BCEL (Byte Code Engineering tool) being a Jakarta project as I was using and thought it a very cool package, and I believe Xalan uses it for their compiler, but nothing resulted from the initial conversation. Markus has approached me now stating that he would be interested in BCEL becoming a Jakarta project so I thought I would see what others thought. BCEL has many applications including compilers, analysis tools and who knows what else. Velocity depends on BCEL's predecessor - JavaClass. Avalon's Excalibur includes a GNU licensed competitor named gnu.bytecode from a package named Kawa, but no longer seems to be depending on it. I see this as an indication that there is an interest in such functionallity by Java based Apache projects, but I would prefer to understand the history behind Velocity and Excalibur's choices before proceeding. - Sam Rubyb - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: BCEL @ apache
On 10/1/01 11:31 AM, Jon Stevens [EMAIL PROTECTED] wrote: on 10/1/01 7:56 AM, Jason van Zyl [EMAIL PROTECTED] wrote: Hi, Many months ago I asked Markus Dahm whether or not he would be interested in BCEL (Byte Code Engineering tool) being a Jakarta project as I was using and thought it a very cool package, and I believe Xalan uses it for their compiler, but nothing resulted from the initial conversation. Markus has approached me now stating that he would be interested in BCEL becoming a Jakarta project so I thought I would see what others thought. BCEL has many applications including compilers, analysis tools and who knows what else. Thoughts? What is his developer/user community like? I belong to the mailing list and there is never that much traffic but I think that's due to the completeness of the tool. As it is based on the language spec it is pretty much finished as a tool. Markus can correct me if I'm wrong. Maybe there are changes in Merlin that will need to be accomodated, I don't know. I think it is primarily Markus who works on it which is why he wants to consider giving it to Apache. He won't have time to maintain it. I'm leaning towards being conservative and saying that we don't need another project right now since we just added Lucene and moved Cactus to a TLP (top level project). I don't think there is a rush, but I think BCEL is an amazingly powerful tool with many possibilities. I don't think it would add a lot of overhead as it is a very specialized tool and the mailing traffic is pretty minimal. I would volunteer to help the move if it was decided to move it to jakarta. -jon - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- jvz. Jason van Zyl http://tambora.zenplex.org http://jakarta.apache.org/turbine http://jakarta.apache.org/velocity http://jakarta.apache.org/alexandria http://jakarta.apache.org/commons - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: BCEL @ apache
Updates: 1) xsltc and velocity appear to be compatible with the latest BCEL. 2) BCEL's build.xml has minor errors - I'll post details to BCEL's mailing list. 3) BCEL includes and depends on gnu regexp. The dependency does not appear to be deep. At the moment, I'm -1 on including a third regexp package in Jakarta. - Sam Ruby - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: FW: Apache logs problem
There's a very simple explanation for this - sometimes transactions (like a big download) take a long time to complete, and the file descriptors to logs persist even after the midnight roll is done, so when that transaction completes the record is logged to the old log. This is OK, we're talking about a couple hits out of millions. Just don't rely on the file 2001/09/24.gz only containing hits from that day, or hits from 9/24 only appearing in 2001/09/24.gz. For a question of relevancy: there are 4200070 hits recorded in 09/24.gz. Of those, 12307 hits are actually from the 25th. That's higher than I thought, but tracked it down to the fact that a graceful restart (-USR1) doesn't appear to shut down persistant connections. Also, don't rely on the hits in the logs being in strict time order, as logfile writing is buffered, and different buffers from different child processes may flush at different times. Brian -- Forwarded Message From: Vincent Massol [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Date: Sat, 29 Sep 2001 11:39:50 +0100 To: [EMAIL PROTECTED] Subject: Apache logs problem It seems there is a problem of dates with the generated apache logs in /x2/logarchive/www/2001 For example if you look at 2001/09/24.gz, the last entry is dated 25/Sep/2001:00:21:09 and the first entry of 2001/09/25.gz is dated 25/Sep/2001:00:00:04 !! Now, even worse, I have tried to find in 24.gz the first entry that is in 25.gz with no success. It is not there ! It means there is a problem of dates and the entries do not seem to be duplicate ... !?? This is a big problem for handling logs, especially for generating statistics. This problem is quite new (maybe 1 month at most) and it worked fine before. Can anyone help ? Thanks -Vincent Massol - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- End of Forwarded Message -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= CollabNet | open source | do what's right - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: BCEL @ apache
On Tue, 2 Oct 2001 01:50, Sam Ruby wrote: Jason van Zyl wrote: Many months ago I asked Markus Dahm whether or not he would be interested in BCEL (Byte Code Engineering tool) being a Jakarta project as I was using and thought it a very cool package, and I believe Xalan uses it for their compiler, but nothing resulted from the initial conversation. Markus has approached me now stating that he would be interested in BCEL becoming a Jakarta project so I thought I would see what others thought. BCEL has many applications including compilers, analysis tools and who knows what else. Velocity depends on BCEL's predecessor - JavaClass. Avalon's Excalibur includes a GNU licensed competitor named gnu.bytecode from a package named Kawa, but no longer seems to be depending on it. Naah -we should delete that ;) Though Avalon will have dependencies on BCEL in the future (though not for about 4-5 months). I see this as an indication that there is an interest in such functionallity by Java based Apache projects, but I would prefer to understand the history behind Velocity and Excalibur's choices before proceeding. Excaliburs choice was via ignorance of other toolkits. I managed to find w other opensource toolkits and a few commercial toolkits. In pure technical terms, BCEL is one of the best (if not the best) that I know of (though some commercial toolkits have better docs). I would have chosen BCEL over gnu.bytecode if I had known about it. The one issue is lack of community. I have been on the BCEL/JavaClass mailing list for a while and there is little traffic except from users but I guess I see this because BCEL is a niche product and relatively high quality one (hence why no one has jumped on board to help). Another issue is that it uses gnu.regexp but this could easily be replaced by one of the jakarta flavours I think. Anyways I think it is a great package code-wise and would be great in that sense. If he wants to add it to jakarta and we have people here that are willing to champion it then - sounds good to me ;) -- Cheers, Pete *-* | Does the name `Pavlov' ring a bell? | *-* - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
