Re: webalizer help
- Errr, you're using IIS, so you can't of course run the script I told
you. Ejem. Try this:
webalizer [your_options] c:\rotated_logs\access_log*
If rotated logs have proper names that get the files ordered correctly
when listing, then webalizer will work well. For names like
"access_log20040102" (a log rotated in 02/Jan/2004) it will work fine.
If logs are entered not in order, then all entries should be lumped
together in a single day (200000 visits in day x and 0 visits from day
x to today, sometimes with days before day x appearing correctly,
meaning that some entry from day x has appeared incorrectly after day
x+1 has been processed)
NOTE: Remember to delete everything in the output directory before
trying what I told you, so the old history information is not used
Have a nice day.
--- In email@example.com, Enric Naval <enventa2000@y...>
> --- Michael Stewart <mstewartga@y...> wrote:
> > I cant get the moderator of the yahoo group to
> > accept my membership.
> > I have been pending membership for a week.
> > I have webalizer, recently installed, running on IIS
> > 6 processing 3
> > websites.
> > For some reason, after each process I get only 1
> > days worth of info
> > and it deletes the previous days.
> > The history file always contains only one line and
> > the file is time
> > stamped at the time of processing.
> > Any ideas?
> > I appreciate your help.
> > Regads,
> > Michael Stewart
> (Please next time post also your webalizer.conf,
> cutting thigs for privacy if you need to, so we can
> help better.)
> Let's me guess, you are rotating logs daily, and in
> your webalizer.conf you have a line saying:
> #Incremental no
> or perhaps:
> Incremental no
> (the default value is 'no').
> You are also of course processing only the logfile you
> have just rotated that day, not all past rotated logs.
> Because webalizer has not been told with "Incremental
> yes" to remember past log files then it will only
> display statistics about the single daily file it has
> just processed.
> OK, you have two solutions:
> solution 1:
> Put a line saying "Incremental yes" in your
> webalizer.conf. This will create a file called
> webalizer.current in the output directory.
> Then process all old rotated log files in order.
> You'll see how webalizer will add them to the
> statistics. If the incremental info gets corrupted,
> you'll get to delete webalizer.current and reprocess
> everything again (you can make it fastly with a
> for logfile in $(ls -r /var/log/httpd/access_log.*);
> webalizer [some options] $logfile
> You may also process all old files and then save the
> output directory, writing down which was the last file
> processed (or simply looking at what is the last day
> appearing in the stats). This way you can recover the
> directory and reprocess only from that day.
> NOTE: if you delete old logfiles, and incremental
> becomes corrupted, you'll lose the data!
> solution 2:
> cat all logfiles together in a single file and then
> add every night the rotated file to the end. Make
> webalizer re-process that file every night. Prepare to
> buy more RAM for webalizer so it doesn't resort to
> To prevent problems:
> With incremental on, just let webalizer go on
> analysing. Before it process 1 Jan 2004, save a copy
> of the stats in a different folder, so you'll have a
> nice summary page for all 12 months of 2003. Do the
> same before it process 1 Jan 2004.
> Last: Your server is not doing reverse-resolving for
> visitors'IPs. The webalizer is not resolving either,
> that's why you have no countries stats. To solve this,
> uncoment the DNSChildren line, so it will attempt to
> resolve all IPs it finds (expect your DNS server to be
> hammered for good if you process lots of entries! You
> can limit the "children" number, "0" disables
> resolving, all sites are IPs and no countries are
> found. Of course "0" is the default value... Normally,
> people relies in Apache to resolve addresses).
> Buf, sorry for being so verbose. Hope I'm useful.
> Enric Naval
> Estudiante de Informática de Gestión en la Udl (Lleida)
> Do you Yahoo!?
> Yahoo! Tax Center - File online by April 15th
- Then webalizer only uses more RAM when you add more entries to the
webalizer.conf file? Entries like SearchEngine xxx x,
GroupSite/HideSite, GroupAgent/HideAgent, etc?
How much RAM would every extra entry add on average? 100 bytes? 64
bytes/each plus overhead for hash tables?
--- In firstname.lastname@example.org, "Bradford L. Barrett" <brad@m...>
> > solution 2:
> > cat all logfiles together in a single file and then
> > add every night the rotated file to the end. Make
> > webalizer re-process that file every night. Prepare to
> > buy more RAM for webalizer so it doesn't resort to
> > swap.
> FYI: There is no difference in RAM usage between using one big log
> or processing lots of smaller files using incremental mode. Thesame
> memory is consumed regardless of the method used.log
> Also, Option #3: Just rotate your logs once a month. Process the
> whenever you want to get an updated report, and once at the end ofthe
> month after rotation to get the final monthly report. This is the
> preferred and _recommended_ method to use.
> Bradford L. Barrett brad@m...
> 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.
> Then webalizer only uses more RAM when you add more entries to theThe use of entries such as the above have a minimal effect on the amount
> webalizer.conf file? Entries like SearchEngine xxx x,
> GroupSite/HideSite, GroupAgent/HideAgent, etc?
of RAM used, but a huge impact on processing time since each entry must
be compared against each log record. If you have lots and lots of
PageType/IndexAlias/Hide/Group/Ignore/Include/SearchEngine entries, they
will have a large impact on processing time since each one must be
processed for each log record, and the impact is linear depending on the
number of such entries.
> How much RAM would every extra entry add on average? 100 bytes? 64Depends. Some just keep the match string in a linked list, so it only
> bytes/each plus overhead for hash tables?
takes however many bytes required for the string + linkage. Others
will also create a hash table entry of whatever size required for the
object being referenced. The size depends on the object, since each
object maintains different metrics particular to the object.
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.