🚀 go-pugleaf

On Tue, 19 Mar 2002, Robert B. wrote:

> > 400 loadav [innwatch:hiload] 17384 gt 2000
> > to co to ?:)
> Nie wiem :-).

INNWATCH(8)                                           INNWATCH(8)

NAME
       innwatch - monitor innd.

SYNOPSIS
       innwatch [ -l logfile ] [ -t seconds ]

DESCRIPTION
       Innwatch  is normally started by rc.news.  It periodically
       -- every <innwatchsleeptime in inn.conf> seconds --  exam­
       ines  the  load average, and the number of free blocks and
       inodes on the spool partition, as described by its control
       file, innwatch.ctl(5).

       If  the  load gets too high, or the disk gets too full, it
       throttles the server.  When  the  condition  restores,  it
       unblocks  the  server.   In addition, on each pass through
       the  loop  it  will  check   the   logfile   <pathlog   in
       inn.conf>/news.crit  to  see  if it has been modified, and
       send mail to the news administrator if so.

       Upon receipt of an  interrupt  signal  (SIGINT),  innwatch
       will     report     its     status     in     the     file
       <pathrun in inn.conf>/innwatch.status.

OPTIONS
       -l     To specify a log file  to  watch,  other  than  the
              default of news.crit, use the ``-l'' flag.

       -t     To   change  the  period  between  check  from  the
              default, use the ``-t'' flag.

HISTORY
       Written by Mike Cooper <mco...@usc.edu>,  with  modifica­
       tions     by     <k...@munnari.oz.au>,      Steve     Groom
       <st...@elroy.Jpl.Nasa.Gov>   and   Christophe    Wolfhugel
       <wo...@pasteur.fr>.    This   is  revision  1.1.2.1,  dated
       1999/06/12.

SEE ALSO
       ctlinnd(8), inn.conf(5), innwatch.ctl(5), shlock(1).

Fragment man inn.conf o backoffie:

       The following parameters are used only by nnrpd(8) to con­
       trol  high-volume posters via an exponential backoff algo­
       rithm.  These parameters are read at nnrpd(8) run time.

       Exponential posting backoff works as follows. News clients
       are  indexed  by  IP  number (or username, see backoffauth
       below). Each time that a specific IP number posts  a  mes­
       sage, the time of posting is stored (along with the previ­
       ous sleep time, see below).  After a  configurable  number
       of  posts  in a configurable period of time, nnrpd(8) will
       activate posting backoff, and begin to sleep for  increas­
       ing  periods  of  time  before  actually posting anything.
       Posts will still  get  through,  but  at  an  increasingly
       reduced rate.

       The  new sleep time is computed based on the difference in
       time between the last posting  and  the  current  posting,
       assuming that backoff has been activated.

       If  this difference is less than backoffpostfast , the new
       sleep time will be 1 + (previous sleep time * backoffk  ).

       If  this  difference  is  less  than  backoffpostslow, but
       greater than backoffpostfast, then the new sleep time will
       equal the previous sleep time.

       If  this  difference  is greater than backoffpostslow then
       the new sleep time is zero and the number of postings  for
       this IP number is reset to zero.

Którą wersję wolisz? :)

-- 
Gophi * LB#265791 * Mastah Fan of our mother Megan
Phone#: +48503963441 * bash$ [(){ [|[&};](){ [;};]
The shorter is your dick - the longer is your .sig