Le mardi 19 octobre 2010 à 23:49 -0700, Al @ Lab42 a écrit : > Generally yes.
IMHO, monitoring needs a "refresh" to cope with the "new way" servers are operated & built. This is a larger topic that this single thread :) > > So assuming you have monitoring for all of this, is the problem that you > > want visibility of the state right now after a run and not when nagios > > gets round to doing its next checks which might be many minutes? > > Yes, but also I want direct correlation between a puppet run and an > eventual failure. Wouldn't this kind of "instant" monitoring be too overwhelming ? I mean : if you have 500 hosts, checking every 30 minutes you would get a "central service server" checked every 3.6s. How about log correlation, it's not perfect but it can be an acceptable intermediate solution (damn splunk and its crazy pricing) Nico.
signature.asc
Description: Ceci est une partie de message numériquement signée
