Hello,

While still young on the internet, our site had to move hosts twice in
the last year. I hope we ended up in a nice and cosy place, but the
'hits' data accumulated in the first two hosts is no longer visible
(Doh).

Assuming the following:

* All hosts run Linux and the same version of webalizer, Analog and
Apache. (I am quite positive about it - please let me know how
critical the version is and I can check it again)
* I am a very competent Unix user (a veteran ;-), but I don't have
ROOT access to the host - it's a shared hosting environment
* I have the full data backup of the previosu sites

The question is:
What are the CONCRETE steps that I need to take to 'graft' the
previous hosts data into the current host data, so that all the
historical information will be visible on the new one, AS IF there was
no interruption in service?

If possible, I would like to be able to do it all offline, and upload
the information to the server when it's ready.

Thanks

MC

---

Clarification of Question by mcgy-ga on 09 Jul 2002 04:21 PDT

Hello,

The clarification is in regard to the expression 'hits data'. What I
meant by the 'hits data' not being avaibale is that the Apache log
files are not available. My hosting companies were in the habit of
deleting the raw data once the Webalizer and Analog programs were run
to generate the daily (or periodic) statistics file.

They claimed that they need the space ...

Anyway, what I have is the full webalizer and analog records. Is it
possible to combine THEM to form one coherent and continous picture?
If so, HOW???

Thanks

MC

Hello MC,

well having the complete apache logs - of current and previous host - is
all you need to regenerate statistics; there's no needing to being root.

So maybe I dont' understand your question: whats' your difficult in adding
previous apache logs to actuals and running webalizer and analog again?

rggb

Hello,

now it's clear :) I do use both Webalizer and Analog programs, but actually
doesn't come in mind to me a method you should use to regenerate Analog stats,
but creating a program that do it...

I do have a suggestion for the Webalizer, using the Incremental data process.
All you have to do is to generate (or correct previously generated) the
'webalizer.current' and 'webalizer.hist' files - or whatever you've named the
files contining the historical data, including data from the previous statistics
you can obtain from the output generated by the program itself, prior the
interruption.

Analyzing the Webalizer sources (preserve.c) it's almost clear whats the data
that should be into these files.

Function 'get_history' read webalizer.hist, that store all monthly totals for
the last 12 months, example

  1 2002 12345 1234 123 123456 1 31 345 234
  2 2002 12345 1234 123 123456 1 28 345 234
  3 2002 12345 1234 123 123456 1 31 345 234
  ...
  11 2001 12345 1234 123 123456 1 30 345 234
  12 2001 12345 1234 123 123456 1 31 345 234

this data, along with actual logs, reproduce the monthly totals.

Function 'restore_state' read webalizer.current file that stores all previous
runs data. Thats quite complicated to explain here but having a look at the
source file should be self-explanatory - if you've a little confidence with
C, anyways it's commented out. Look at 'save_state' function also.

Of course you have to enable incremental data processing too. Simply add a
'Incremental Yes' line to your conffiles (read webalizer manpage for more
details).

Hope this helps.

rggb


PS. I will document myself for Analog, since I wish to reproduce your
'problem' to reduce the execution times :)