[prev in list] [next in list] [prev in thread] [next in thread] 

List:       netsaint-devel
Subject:    Re: [netsaint-devel] Required enhancement in status change logging
From:       Rene Klootwijk <rene () klootwijk ! org>
Date:       2000-01-27 8:18:42
[Download RAW message or body]

Karl DeBisschop wrote:

> >Well, if NetSaint croaks without writing anything to the log file then
> >there's no easy way to determine the downtime (at least from the
> >log file).  Although I think the initial problem had more to deal with
> >the lag between startup and the first check of a service/host (see
> >original text below).  If NetSaint is able to write a shutdown
> >message to the log file then its easy to calculate the time that it
> >was not running...
>
> Thus, in the original post it was suggested that all service checks be
> logged -- so you could determine when netsaint crashed.  But that
> would quickly swamp the sorting routine, and would take more disk
> space than I want to give to netsaint.  Thus my suggestion that even
> if there is nothing written to the log, a file could still be touched
> when netsaint wakes up to do something else.  Then on restart, the
> stop time could be reasonably well estimated from the age of the file,
> even if netsaint was not able to write a log message as it crashed.
>
> I thought the original post was concerned about both startup lags and
> the interval between the last error and the time netsaint died.  Maybe
> I misunderstood.
>
> --
>
> Karl
>
> _______________________________________________
> Netsaint-devel mailing list
> Netsaint-devel@lists.sourceforge.net
> http://lists.sourceforge.net/mailman/listinfo/netsaint-devel

What I meant in the original posting is what Ethan described, what is most
needed in order to create availability reports is the time between the
startup of Netsaint and the first check of each service (and the results
of this check). This means that only all the results of the first checks
should be logged and from then on only the state changes. Maybe we should
introduce some kind of watchdog signal as Karl described, but could not we
dot it in the log as well? Suggest Netsaint creates a watchdog log line
once every 10 minutes, that would mean 144 lines a day, is not this
acceptable? Use of the watchdog signal could also be make configurable.

--

Rene

[prev in list] [next in list] [prev in thread] [next in thread]