John> When using IP-Monitor/Check Reachability it would be very
John> nice if the group icons would change color when one of its
John> members bscomes unreachable. This way you could have a nice
John> collapsed map monitoring all of the networks, and still be
John> able to tell when there was a problem with one of the
John> machines.
Somone already suggested to be able to define health functions that
convert many status values in a single one that could be used to
select an appropriate color. I have experimented a little with it and
stopped since I do not have a color display in my room...
If someone would like to do this kind of stuff, I could provide you
with what I have done so far.
John> Also, I would like to see IP-Monitor's be able to be applied
John> to groups (in actuality, it would apply to all member of the
John> group, but this makes life easier when the groups are
John> collpased to have a monitor applied recursively to all
John> hosts/groups in the group.
This could be done easily. At least the ip_monitor.tcl script uses a
proc 'extract' that takes the current selection and returns a list of
pairs of tkined ids and their ip adresses. This is done for each NODE
object. It would be easy to add a recursion that expands GROUP objects
to their member nodes.
My reason not to apply these commands recursively is that it would be
too easy to click on a collapsed group by mistake and start a few
hundred monitors without even noticing it (until the phone rings).
John> There should be some way to kill a monitor on a host. Right
John> now you can suspend them, but it would be nice to be able to
John> really delete the monitor.
If you use the default parameter settings (clone nodes == true) you
get a new node shown as a stripchart for every monitored variable.
Select and delete it. The monitor scripts will recognize it and take
appropriate action (e.g. removing the job if no variables are left).
John> Some of the text windows (e.g. IP-Trouble/Ping) come up
John> scrolled all the way to the bottom of the text area. This
John> means that you have to use the slider bar to look at the
John> output from the ping or whatever command you are using.
I do not fully understand why the call to 'update' is needed here, but
the diff below to tkined_objects.tcl seems to fix this problem.
Juergen
*** tkined_objects.tcl-yow Fri May 6 16:06:30 1994
--- tkined_objects.tcl Fri May 20 18:46:18 1994
***************
*** 1188,1193 ****
--- 1191,1198 ----
wm geometry $w.$log +$x+$y
wm deiconify $w.$log
+
+ update idletasks
}
proc LOG::process { log } {