[logs] Re: What's normal?

From: Paul Ebersman (list-loganalysisat_private)
Date: Tue Aug 20 2002 - 07:51:45 PDT

  • Next message: Rene Pfeiffer: "Re: [logs] Logging: World Domination"

    tbird> 2) Given a particular operating system and/or system purpose
    tbird> (such as a UNIX mail server, or a Windows Domain Controller, or
    tbird> whatever), what are the (pick your favorite integer) 15 most
    tbird> frequently logged messages in the elusive "typical"
    tbird> environment?  What do they mean?  Do we have sample data?
    
    ebersman> I think the key problem is that in order to do this, you
    ebersman> have to define "standard" platform/usage profiles. Most of
    ebersman> us can't even get that for the company we're working for at
    ebersman> the time, much less across the Internet. B^)
    
    tbird> Yeah, yeah, yeah, and for decades, system administrators have
    tbird> talked themselves out of doing the experiment with arguments
    tbird> exactly this way.
    
    Part of my point is that different apps will need different
    knobs/alarms at significantly different usage levels. What I need to
    maintain 1000 users and 20,000 emails a day isn't what I'd need if I
    had 2Meg users and 20 machines churning 2Meg emails/day each.
    
    The amount of constraints we'd have to put on variables to get
    statistically accurate samples would make the results of limited
    use. If we could lock down configs and apps with no variation based on
    "we've still got these P1s, use them!" or "our funding depends on
    using the froobaz mailer even if sendmail works better" or whatever
    other stupidity realities throws us, we probably wouldn't have most of
    our security/system problems.
    
    tbird> Look, everyone, presumably at least some of us have access to
    tbird> log data on a "live server."  I posit that we'll learn really
    tbird> really interesting things by taking a day's or a week's worth
    tbird> of data and looking at the messages.
    [...]
    tbird> My hope is that by providing this sort of information we'll
    tbird> make it easier for people to get up to speed on what is and is
    tbird> not typical >>for them<<.
    
    Fair enough. But I'd still guess that for most newbies, the real first
    problem is learning to drink from the firehose. Coming up with some
    semi-standard way of chewing through megabytes of raw logs to figure
    out your baseline would be useful even if we *never* come up with a
    standardized configuration.
    
    Most folks don't learn any forensic techniques until something's
    already exploded and most literature I've seen deals with those
    explosions. Everyone always says "log and you'll know what's normal"
    but I haven't seen too many folks give good examples of how to weed
    through things during normal operations.
    
    ebersman> I'll posit the next straw man to torch: what would be useful
    ebersman> is a standardized methodology for how to turn two weeks of
    ebersman> verbose logging into a template against which to compare
    ebersman> "normal", "abnormal" and "catastrophic" at your particular
    ebersman> site/application. This does assume your first point of
    ebersman> somewhat standardized logging being available on all
    ebersman> critical OSs and Apps.
    
    tbird> Remember that by "standardized logging" I'm not >>even<<
    tbird> worrying about log formats or severities --just message
    tbird> categories.  Taking yet another step back.
    
    Agreed. My use of "standardized logging" is simply message format, not
    message content. I'm also making the assumption that when I turn the
    logging knob to "log everything" that your goal of programming
    standards of logging for OS and App has turned into something useful.
    _______________________________________________
    LogAnalysis mailing list
    LogAnalysisat_private
    https://lists.shmoo.com/mailman/listinfo/loganalysis
    



    This archive was generated by hypermail 2b30 : Tue Aug 20 2002 - 08:03:19 PDT