Re: [logs] Problems/Questions on PEO

From: Devin Kowatch (devinkat_private)
Date: Wed Apr 03 2002 - 12:05:10 PST

  • Next message: Tina Bird: "[logs] LogWatch 2.5 still vulnerable (fwd)"

    On Tue, Apr 02, 2002 at 02:06:42PM -0500, Sweth Chandramouli wrote:
    > > How difficult is it to go back and
    > > reconsile the two logs?
    > 	It depends on how rigorous you want to be.  If the loghosts
    > and the hosts logging to them are all on a quiet switched network with
    > no intervening WAN links, you can probably get away with having each
    > loghost pull over the most recently rolled logfiles from the other hosts
    > and each doing a diff against the most recently rolled local logfiles.
    > Once the networks start getting saturated or you start getting WAN-level
    > latency, however, the machines will start receiving messages in different
    > orders and with different timestamps.  The timestamp issue can be solved
    > by just comparing the files after cutting the timestamps out, on the
    > assumption that if all of the lines are present and in the same order
    > then the timestamps are irrelevant.  Once the orders start differing,
    > however, things get more interesting; you need to do things like
    > comparing "rolling windows" of each file and trying to match a given
    > line in the local file to any of the lines in the current window of the
    > alternate file that has the same message text and a timestamp within
    > some reasonable interval (a la diff, but with extra logic for the
    > timestamps).
    
    Next question.  How well does this scale?  How big are the log files
    that you do this on?  We get about 2GB/week, and I was wondering if
    something like this would be feasable.
    
    thanks,
    -- 
    Devin Kowatch
    devinkat_private
    
    ---------------------------------------------------------------------
    To unsubscribe, e-mail: loganalysis-unsubscribeat_private
    For additional commands, e-mail: loganalysis-helpat_private
    



    This archive was generated by hypermail 2b30 : Wed Apr 03 2002 - 14:20:47 PST