[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Detecting failed clients

From: Jerry Christopher
Subject: Detecting failed clients
Date: Wed, 06 Jun 2001 14:39:37 -0700

cfengine 1.6.3
solaris 7, 8

I read Mike Lang's post 2001-04-09 regarding this subject and the two
responses given.  I don't believe the responses (call cfrun & cfengine
module using database connection) fit well for our situation, and I'm
interested in hearing how others are approaching the same issue, if at

We have several sites using cfengine on solaris 7 & 8.  Our architecture
is 3-tiered: a "gold" or master distribution server, a couple of
"distrib" or slave servers at headquarters, and a "distrib" server on
the LAN for each remote site.

Distrib servers and clients run cfengine periodically from cron and a
new log is written locally on the client each time.  I'd like to know if
& when a client has not run cfengine and/or the date and time of the
last successful run.  

Example:  I make a change to cfengine.conf on the "gold" server to add a
new link to the links section.  The distrib servers pick up the change
on the next run and add the link on the run after that.  The 3rd tier
clients pick up the change from the distrib server on their next run &
make the link on the run after that.  A user on a client at "Site X"
discovers that he doesn't have the link that his peers have on their
workstation.  The user calls the helpdesk, submits a ticket or in
someway makes this problem known.  It may be some time before I actually
hear about the missing link and then I can go out and investigate the

I'm looking for ways to proactively "know" when clients fail to run
without serially calling every local and remote client with cfrun.
(if this is a valid method by just parsing, I'd appreciate comment)

Some thoughts being tossed around are touching a file on an NFS mount or
scripting a check of the date/time stamp on the log file, collecting
this and the uname -n of the client, and getting this information "up
the chain" back to the master server somehow.  A manual check of the
file would find hosts that have not run cfengine "lately" and also might
be compared against a layer 1 inventory of suns (using getethers? not
sure here) to find a client that has never run cfengine for some reason.

Thoughts and comments appreciated!

Thank you,


Jerry Christopher
UNIX Systems Administrator
Applied Micro Circuits Corporation
6290 Sequence Drive
San Diego, CA 92121 USA

Work 858-535-4245
eFax 208-441-7898

reply via email to

[Prev in Thread] Current Thread [Next in Thread]