Loading ...
Sorry, an error occurred while loading the content.

Re: [webalizer] Old reports are (sometimes) flushed

Expand Messages
  • Bradford L. Barrett
    ... Incremental mode is for use only when you are rotating logs.. if you are using it against live logs, you can, and usually will, lose data. -- Bradford L.
    Message 1 of 19 , Feb 6, 2003
    • 0 Attachment
      On Thu, 6 Feb 2003, Kees de Keizer wrote:

      > Setting "Incremental yes" save the stats info for the current month
      > to a file and that is used again when you generate stats for the same
      > month. I.E. On my personal server I generate the stats every hour (to
      > get nearly live info) and with the use of the incremental file this is
      > done within a minute: http://stats.de-keizer.net/rip/ and also on
      > http://stats.de-keizer.net/rip/apache/

      Incremental mode is for use only when you are rotating logs.. if you are
      using it against live logs, you can, and usually will, lose data.

      --
      Bradford L. Barrett brad@...
      A free electron in a sea of neutrons DoD#1750 KD4NAW

      The only thing Micro$oft has done for society, is make people
      believe that computers are inherently unreliable.
    • Kees de Keizer
      ... I ve been using it the way I described it for more than a year now, and I never experienced any trouble with it. Just once I had a corrupted logfile and
      Message 2 of 19 , Feb 6, 2003
      • 0 Attachment
        Bradford L. Barrett wrote on 06-02-2003:

        > > On my personal server I generate the stats every hour (to get
        > > nearly live info) and with the use of the incremental file this
        > > is done within a minute: http://stats.de-keizer.net/rip/
        > Incremental mode is for use only when you are rotating logs.. if
        > you are using it against live logs, you can, and usually will,
        > lose data.

        I've been using it the way I described it for more than a year
        now, and I never experienced any trouble with it. Just once
        I had a corrupted logfile and that was caused by a corrupted
        referrer.
        --
        Kees de Keizer NT Administrator T: +31-10-2448344
        #easynet Nederland PGP: 0x24E3770B F: +31-10-2448356
        http://www.easynet.nl/ kees.de.keizer@...
        We gaan door, maar waarvoor? [Rent]
      • Bradford L. Barrett
        ... And you won t notice any trouble .. the program was designed to be run once a month against a full months worth of logs.. _only_ if you need to rotate
        Message 3 of 19 , Feb 6, 2003
        • 0 Attachment
          On Thu, 6 Feb 2003, Kees de Keizer wrote:

          > > Incremental mode is for use only when you are rotating logs.. if
          > > you are using it against live logs, you can, and usually will,
          > > lose data.
          >
          > I've been using it the way I described it for more than a year
          > now, and I never experienced any trouble with it. Just once
          > I had a corrupted logfile and that was caused by a corrupted
          > referrer.

          And you won't notice any 'trouble'.. the program was designed to be run
          once a month against a full months worth of logs.. _only_ if you need to
          rotate more than once a month should you use Incremental mode, and then,
          only against the _rotated_ log file. Depending on site traffic, I have
          seen upwards of 5-10% data loss when trying to run against a live log
          using incremental mode. Nothing will appear out of the ordinary except
          your numbers will be inaccurate, and you generally will never know it.
          If you want to run against live logs, disable incremental mode.

          --
          Bradford L. Barrett brad@...
          A free electron in a sea of neutrons DoD#1750 KD4NAW

          The only thing Micro$oft has done for society, is make people
          believe that computers are inherently unreliable.
        • Kees de Keizer
          ... If I had to use it that way, it s complete useless for me. Our customers don t want info once a month. For our production machines we use a daily run
          Message 4 of 19 , Feb 6, 2003
          • 0 Attachment
            Bradford L. Barrett wrote on 06-02-2003:

            > [Using Incremental mode on live logs]
            > And you won't notice any 'trouble'.. the program was designed to be run
            > once a month against a full months worth of logs.. _only_ if you need to
            > rotate more than once a month should you use Incremental mode, and then,
            > only against the _rotated_ log file.

            If I had to use it that way, it's complete useless for me. Our
            customers don't want info once a month. For our production machines
            we use a daily run around midnight.

            > Depending on site traffic, I have seen upwards of 5-10% data loss
            > when trying to run against a live log using incremental mode.
            > Nothing will appear out of the ordinary except your numbers will
            > be inaccurate, and you generally will never know it.

            There is no way to give accurate data. Whether you use log files
            or a javascript utily on the site. And I don't believe I'm having
            5-10% data loss during the 12-15 minutes webalizer runs each day.

            > If you want to run against live logs, disable incremental mode.

            Thanks for the tip, but that will make it completely useless
            for me and our customers.
            --
            Kees de Keizer NT Administrator T: +31-10-2448344
            #easynet Nederland PGP: 0x24E3770B F: +31-10-2448356
            http://www.easynet.nl/ kees.de.keizer@...
            We take the loads from off the roads [Starlight Express]
          • Stanley Stramel
            I agree! Our customers require that we have near live stats. Bradford, is the loss of data occrring during the Webalizer is running on the live log file and
            Message 5 of 19 , Feb 6, 2003
            • 0 Attachment
              I agree!  Our customers require that we have near live stats.
               
              Bradford, is the loss of data occrring during the Webalizer is running on the live log file and it may be dropping records that are being added to the log file during that execution?
               
              Stan
              -----Original Message-----
              From: Kees de Keizer [mailto:kees.de.keizer@...]
              Sent: Thursday, February 06, 2003 9:45 AM
              To: Bradford L. Barrett
              Cc: webalizer@yahoogroups.com
              Subject: Re: [webalizer] Old reports are (sometimes) flushed

              Bradford L. Barrett wrote on 06-02-2003:

              > [Using Incremental mode on live logs]
              > And you won't notice any 'trouble'.. the program was designed to be run
              > once a month against a full months worth of logs.. _only_ if you need to
              > rotate more than once a month should you use Incremental mode, and then,
              > only against the _rotated_ log file.

              If I had to use it that way, it's complete useless for me. Our
              customers don't want info once a month. For our production machines
              we use a daily run around midnight.

              > Depending on site traffic, I have seen upwards of 5-10% data loss
              > when trying to run against a live log using incremental mode.
              > Nothing will appear out of the ordinary except your numbers will
              > be inaccurate, and you generally will never know it.

              There is no way to give accurate data. Whether you use log files
              or a javascript utily on the site. And I don't believe I'm having
              5-10% data loss during the 12-15 minutes webalizer runs each day.

              > If you want to run against live logs, disable incremental mode.

              Thanks for the tip, but that will make it completely useless
              for me and our customers.
              --
              Kees de Keizer          NT Administrator    T: +31-10-2448344
              #easynet Nederland      PGP:  0x24E3770B    F: +31-10-2448356
              http://www.easynet.nl/          kees.de.keizer@...
              We take the loads from off the roads [Starlight Express]

              To unsubscribe from this group, send an email to:
              webalizer-unsubscribe@egroups.com
              Webalizer homepage: http://www.webalizer.org
              Webalizer for NT: http://www.medasys-lille.com/webalizer/




              Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
            • Bradford L. Barrett
              ... As do I.. actually, I run it twice a day, but _not_ in incremental mode. Data loss will occur otherwise, and the amount will be determined by the amount of
              Message 6 of 19 , Feb 6, 2003
              • 0 Attachment
                > > [Using Incremental mode on live logs]
                > > And you won't notice any 'trouble'.. the program was designed to be run
                > > once a month against a full months worth of logs.. _only_ if you need to
                > > rotate more than once a month should you use Incremental mode, and then,
                > > only against the _rotated_ log file.
                >
                > If I had to use it that way, it's complete useless for me. Our
                > customers don't want info once a month. For our production machines
                > we use a daily run around midnight.

                As do I.. actually, I run it twice a day, but _not_ in incremental mode.
                Data loss will occur otherwise, and the amount will be determined by the
                amount of traffic the individual site generates.. small sites may not
                experience any.. large sites will experience a lot.

                > There is no way to give accurate data. Whether you use log files
                > or a javascript utily on the site.

                Huh?

                > And I don't believe I'm having 5-10% data loss during the 12-15 minutes
                > webalizer runs each day.

                Probably not.. depends on size of the site. The problem is not in the
                length of time to process, but the handling of timestamps in incremental
                mode. And you will not notice the loss unless you compare the results
                to those run the correct way. I've been doing this since 1997, and have
                done test results with hundreds of sites ranging in size from small
                mom and pop shops to huge porn sites with many millions of hits a day.
                Offically, you should never, ever, run against live logs when you are
                using incremental mode.. to do so means you run the risk of producing
                inaccurate results, with a greater risk the larger the site is. If you
                run in incremental mode, you should _only_ process rotated logs.. If
                you want to run against live logs, then disable incremental mode and
                run against them as much as you like.

                > > If you want to run against live logs, disable incremental mode.
                >
                > Thanks for the tip, but that will make it completely useless
                > for me and our customers.

                I fail to see why.. thosands of ISPs do it every day without a problem.

                --
                Bradford L. Barrett brad@...
                A free electron in a sea of neutrons DoD#1750 KD4NAW

                The only thing Micro$oft has done for society, is make people
                believe that computers are inherently unreliable.
              • Kees de Keizer
                ... Sorry, but if I had to do it that way, my servers would only be busy with generating stats. Maybe unix machine can handle multiple instances of webalizer
                Message 7 of 19 , Feb 6, 2003
                • 0 Attachment
                  Bradford L. Barrett wrote on 06-02-2003:

                  > [Using Incremental mode on live logs]
                  > > > .. the program was designed to be run once a month against
                  > > > a full months worth of logs.. _only_ if you need to rotate
                  > > > more than once a month should you use Incremental mode, and
                  > > > then, only against the _rotated_ log file.
                  > > If I had to use it that way, it's complete useless for me. Our
                  > > customers don't want info once a month. For our production machines
                  > > we use a daily run around midnight.
                  > As do I.. actually, I run it twice a day, but _not_ in incremental mode.
                  > Data loss will occur otherwise, and the amount will be determined by the
                  > amount of traffic the individual site generates.. small sites may not
                  > experience any.. large sites will experience a lot.

                  Sorry, but if I had to do it that way, my servers would only be
                  busy with generating stats. Maybe unix machine can handle multiple
                  instances of webalizer better but on a windows machine the impact
                  of running webalizer is quite high. That's why I run it around
                  mignight on our production machines.

                  > > There is no way to give accurate data. Whether you use log files
                  > > or a javascript utily on the site.
                  > Huh?

                  In this case I was talking about actual site visitors. I.E. proxies
                  do not generate a a log entry when they present the site to a visitor
                  using that proxy.

                  > If you run in incremental mode, you should _only_ process rotated
                  > logs.. If you want to run against live logs, then disable incremental
                  > mode and run against them as much as you like.

                  Ask Bill if he can generate rotated logfiles with a different
                  extension or any other way so they can be easily selected with
                  a wildcard. Since this is not the case you have to run it against
                  a live log file.

                  > > > If you want to run against live logs, disable incremental mode.
                  > > Thanks for the tip, but that will make it completely useless
                  > > for me and our customers.
                  > I fail to see why.. thosands of ISPs do it every day without a
                  > problem.

                  I never said I have a problem. You try to talk me into one.
                  --
                  Kees de Keizer NT Administrator T: +31-10-2448344
                  #easynet Nederland PGP: 0x24E3770B F: +31-10-2448356
                  http://www.easynet.nl/ kees.de.keizer@...
                  Lovers are very special people [Funny Girl]
                • Bradford L. Barrett
                  ... The limitations of your platform of choice is unfortunate.. however, the implication that it s ok to run against live logs when using incremental mode is
                  Message 8 of 19 , Feb 6, 2003
                  • 0 Attachment
                    > Sorry, but if I had to do it that way, my servers would only be
                    > busy with generating stats. Maybe unix machine can handle multiple
                    > instances of webalizer better but on a windows machine the impact
                    > of running webalizer is quite high. That's why I run it around
                    > mignight on our production machines.

                    The limitations of your platform of choice is unfortunate.. however,
                    the implication that it's ok to run against live logs when using
                    incremental mode is inaccurate, regardless of any justifications
                    you can make.

                    > > > There is no way to give accurate data. Whether you use log files
                    > > > or a javascript utily on the site.
                    > > Huh?
                    >
                    > In this case I was talking about actual site visitors. I.E. proxies
                    > do not generate a a log entry when they present the site to a visitor
                    > using that proxy.

                    The webalizer was designed to give administrators insight into the usage
                    of their servers, in order to provide insight into capacity planning and
                    future direction. If a request is satisified from a proxy instead of
                    hitting the server, that does not impact the analysis of server
                    performance or usage from that standpoint. The program will provide
                    accurate analysis of actual usage in every way with the exception of
                    visit/entry/exit page analysis (as described in the documentation).

                    > > If you run in incremental mode, you should _only_ process rotated
                    > > logs.. If you want to run against live logs, then disable incremental
                    > > mode and run against them as much as you like.
                    >
                    > Ask Bill if he can generate rotated logfiles with a different
                    > extension or any other way so they can be easily selected with
                    > a wildcard. Since this is not the case you have to run it against
                    > a live log file.

                    Again, the limitations of your platform of choice is not justification
                    for giving people the incorrect assumption that it's ok to run against
                    live logs in incremental mode. To do so is a disservice to those users.

                    > > > > If you want to run against live logs, disable incremental mode.
                    > > > Thanks for the tip, but that will make it completely useless
                    > > > for me and our customers.
                    > > I fail to see why.. thosands of ISPs do it every day without a
                    > > problem.
                    >
                    > I never said I have a problem. You try to talk me into one.

                    I took your statement to mean that it is impossible for you to use
                    the program correctly (ie: to do so would "make it completely useless"),
                    which I would view as a problem. Apparently, you don't think so.

                    --
                    Bradford L. Barrett brad@...
                    A free electron in a sea of neutrons DoD#1750 KD4NAW

                    How do you give Microsoft the benefit of the doubt when you
                    know that if you were to throw it in a room with truth, you'd
                    risk a matter/anti-matter explosion? -- Nicholas Petreley IDG
                  • Kees de Keizer
                    ... But not my choice. We have customers who want to write their websites in BASIC. ... If I want to know the traffic to a server, I ll ask our networking
                    Message 9 of 19 , Feb 6, 2003
                    • 0 Attachment
                      On Thu, 6 Feb 2003, Bradford L. Barrett wrote:

                      > > Sorry, but if I had to do it that way, my servers would only be
                      > > busy with generating stats. Maybe unix machine can handle multiple
                      > > instances of webalizer better but on a windows machine the impact
                      > > of running webalizer is quite high. That's why I run it around
                      > > mignight on our production machines.
                      > The limitations of your platform of choice is unfortunate..

                      But not my choice. We have customers who want to write their
                      websites in BASIC.

                      > > > > There is no way to give accurate data. Whether you use log files
                      > > > > or a javascript utily on the site.
                      > > > Huh?
                      > > In this case I was talking about actual site visitors. I.E. proxies
                      > > do not generate a a log entry when they present the site to a visitor
                      > > using that proxy.
                      > The webalizer was designed to give administrators insight into the usage
                      > of their servers, in order to provide insight into capacity planning and
                      > future direction. If a request is satisified from a proxy instead of
                      > hitting the server, that does not impact the analysis of server
                      > performance or usage from that standpoint. The program will provide
                      > accurate analysis of actual usage in every way with the exception of
                      > visit/entry/exit page analysis (as described in the documentation).

                      If I want to know the traffic to a server, I'll ask our networking
                      department for analysis. They can provide better traffic information
                      that webalizer can, because they measure all the traffic. And nowadays
                      webalizer is used very often to provide statistics to customers, so
                      they can see when, what, how and with what their site has been visited.
                      And they won't miss the 0,5% (if it is that much anyhow) in the stats.

                      > Again, the limitations of your platform of choice is not justification
                      > for giving people the incorrect assumption that it's ok to run against
                      > live logs in incremental mode. To do so is a disservice to those users.

                      I give our customers what they want: Site statistics. Our customers know
                      that they are not 100% accurate if it concern page views and visits, due
                      to the use of i.e. proxies. Webalizer is no longer a tool just for the
                      administrators, but is more and more used for site statistics. Why would
                      you else present referrers, user agents and countries? That information
                      has nothing to do with analysis of the performance.

                      > > I never said I have a problem. You try to talk me into one.
                      > I took your statement to mean that it is impossible for you to use
                      > the program correctly (ie: to do so would "make it completely useless"),
                      > which I would view as a problem. Apparently, you don't think so.

                      That's right. I give our customers what they want, and that
                      is site statistics. They only complaint I'll get, is the lack
                      of information about the used operating system. But that is
                      something that won't be a part of webalizer.
                      --
                      Kees de Keizer NT Administrator T: +31-10-2448344
                      #easynet Nederland PGP: 0x24E3770B F: +31-10-2448356
                      http://www.easynet.nl/ kees.de.keizer@...
                    • Bradford L. Barrett
                      ... Because it s already there and it can be reported easily with very little additional overhead. But that is beside the point.. the point is that it is NOT
                      Message 10 of 19 , Feb 6, 2003
                      • 0 Attachment
                        > I give our customers what they want: Site statistics. Our customers know
                        > that they are not 100% accurate if it concern page views and visits, due
                        > to the use of i.e. proxies. Webalizer is no longer a tool just for the
                        > administrators, but is more and more used for site statistics. Why would
                        > you else present referrers, user agents and countries? That information
                        > has nothing to do with analysis of the performance.

                        Because it's already there and it can be reported easily with very little
                        additional overhead.

                        But that is beside the point.. the point is that it is NOT ok to run
                        against live logs when using incremental mode, regardless of your
                        decision, for whatever reasons, to do so anyway.

                        --
                        Bradford L. Barrett brad@...
                        A free electron in a sea of neutrons DoD#1750 KD4NAW

                        The only thing Micro$oft has done for society, is make people
                        believe that computers are inherently unreliable.
                      • Kees de Keizer
                        ... Maybe you should add that more clearly to the README file. The only about it now is something about preventing same timestamps . -- Kees de Keizer
                        Message 11 of 19 , Feb 6, 2003
                        • 0 Attachment
                          On Thu, 6 Feb 2003, Bradford L. Barrett wrote:

                          > the point is that it is NOT ok to run against live logs when
                          > using incremental mode, regardless of your decision, for
                          > whatever reasons, to do so anyway.

                          Maybe you should add that more clearly to the README file. The
                          only about it now is something about "preventing same timestamps".
                          --
                          Kees de Keizer NT Administrator T: +31-10-2448344
                          #easynet Nederland PGP: 0x24E3770B F: +31-10-2448356
                          http://www.easynet.nl/ kees.de.keizer@...
                        • Josh Kuperman
                          I m very confused about this. From my reading of the documentation, it looks like running against live logs should not be a problem as long as you have a
                          Message 12 of 19 , Feb 7, 2003
                          • 0 Attachment
                            I'm very confused about this.

                            From my reading of the documentation, it looks like running against
                            live logs should not be a problem as long as you have a history file
                            keeping copies of the log records so they don't get counted twice. I
                            am started to use webalizer quite a bit for keeping track of both
                            web and proxy (squid) use. In all cases I'm running against live
                            logs. I really need to know what happened yesterday - ideally I'd be
                            processing hourly.

                            So far it is still very unclear to me why the live logs are a
                            problem? I suppose I could force the logs to rotate every day, but I
                            still don't get it. A file is a file, and at least on Unix systems I
                            don't see what the difference is between a live log and an archived
                            log.


                            On Fri, Feb 07, 2003 at 07:17:08AM +0100, Kees de Keizer wrote:
                            > On Thu, 6 Feb 2003, Bradford L. Barrett wrote:
                            >
                            > > the point is that it is NOT ok to run against live logs when
                            > > using incremental mode, regardless of your decision, for
                            > > whatever reasons, to do so anyway.
                            >
                            > Maybe you should add that more clearly to the README file. The
                            > only about it now is something about "preventing same timestamps".

                            --
                            Josh Kuperman
                            josh@...
                          • Bradford L. Barrett
                            ... You don t need incremental mode to run against logs hourly. ... One is being written to while processing, the other isn t. -- Bradford L. Barrett
                            Message 13 of 19 , Feb 7, 2003
                            • 0 Attachment
                              > I really need to know what happened yesterday - ideally I'd be
                              > processing hourly.

                              You don't need incremental mode to run against logs hourly.

                              > I don't see what the difference is between a live log and an archived
                              > log.

                              One is being written to while processing, the other isn't.

                              --
                              Bradford L. Barrett brad@...
                              A free electron in a sea of neutrons DoD#1750 KD4NAW

                              The only thing Micro$oft has done for society, is make people
                              believe that computers are inherently unreliable.
                            Your message has been successfully submitted and would be delivered to recipients shortly.