Re: Making a box "scream bloody murder" if certain jobs aren't running -- any ideas?

I use the message every day approach and I've been known to be busy or pre-occupied to notice its absence...but I have also caught things going on when i DID notice that it didn't come through.

Rob's "keep alive" would work if you had a reliable second box that could script the timeout.

And it isn't "base OS", but using something like HTTPAPI given the right web service would let you know something was wrong.

Of course, if you figure out the monitor, it is pretty easy to effect the cure in most cases.




________________________________
From: James H. H. Lampert <jamesl@xxxxxxxxxxxxxxxxx>
To: Midrange Systems Technical Discussion <midrange-l@xxxxxxxxxxxx>
Sent: Monday, March 18, 2013 1:11 PM
Subject: Re: Making a box "scream bloody murder" if certain jobs aren't running -- any ideas?

On 3/18/13 10:37 AM, rob@xxxxxxxxx wrote:

James,

In theory you could write a sender, and a receiver, for the IBM smtp
stuff.  Send one every x minutes, if the receiver doesn't find it then use
a sockets program to active that fence charger wired to your chair.

Or, we paid a little for a slightly more full featured account than the
free ones at yahoo.  Our IBM i sends off an email to yahoo, it forwards it
on to our email receiver (Domino).  If the Domino doesn't receive one
within x minutes all hell breaks loose.  My first suggestion doesn't test
if, for some reason, your sender gets blacklisted, comm drops, etc.  The
second method is a much more robust end-to-end.

Well, I'm more concerned with how to "activ[ate] that fence charger
wired to [my] chair [and the boss's]," or to make "all Hell break loose."

For verification of function, the boss and I are both thinking of
sending a test email as part of the IPL process, after giving TCP, SMTP,
and MSF time to come up, and sending one at a fixed time every day, but
I'm thinking that an alarm is more likely to be noticed than the lack of
a verification.

--
JHHL