› Clovertech Forums › Read Only Archives › Cloverleaf › Cloverleaf › Network Monitor GUI issues
Network Monitors running on W2K, W2K3, etc.
Let me preface this by saying… Java sux!
When dragging threads around in the Network Monitor , if you don’t “drop” the thread at just the right time, it jumps up to the top left corner.
Do we know what causes this (other than Java sux!)? Is there something I can do on my system that will keep this from happening? Some video setting, some Java (sux) setting?
It happens when the NetMonitor refreshes.
I have not ever dragged a thread in the netmonitor to a new location and had it bounce back to the upper left corner so perhaps I’m misunderstanding that part of your post.
I have had dragging problems all around when my mouse had gotten worn out and needed replacing.
I have closed the netmonitor and reopend it and had threads go back to the upper left hand corner.
To keep this from happening save the netmonitor changes before exiting, which will save the current view in $HCISITEDIR/views/monitor_dflt.mvw for cloverleaf version 5.2.1P2 running on AIX 5.2.
Another way that I prefer is to make my changes in the netconfigurator and save the view.
Then I sync up the netmonitor view with the netconfg view by doing the following:
rm -f monitor_dflt.mvw
cp config_dflt.view monitor_dflt.view
We made it one of our standards to keep the netmonitor view and the netconfig view the same.
Personally I have found mistakes are reduced when the netconfig and netmonitor have the same view.
Russ Ross
RussRoss318@gmail.com
The workaround I have found is that the thread will jump randomly between the cursor and the upper left corner while you have the left mouse button down. The trick is to release the button when the thread is in the position you want. If the thread is in the corner, just jiggle the mouse a little until it bounces back.
Chris
you describe it exactly, including what I do to get the desired result. After a while, though, my index finger gets tired holding the mouse button down so long! 😆
No repsonse from Michael yet; I think he meant to say that shutting down the monitor daemon helped, but I can’t do that without risking some outage going unnoticed.
I am going to goof around with Russ’ sync suggestions. We, too, try to keep our NetConfig and NetMonitor views as close to identical as possible. His method may work better than the “compare and move” method we are using.
Not sure about the jumping issue in 5.5. I have not seen it but then I do most of my troubleshooting from the command line so not sure if that means it has been fixed or I just have not hit the exact circumstance for it to happen.
But I do have some good news for those considering the move to 5.5. The suggestions for syncing the NetConfig and Netmonitor views is interesting so I thought I would mention that in 5.5 these views are in fact synced for you. Both the NetConfig and NetworkMonitor GUIs use the same view file.
Chris
The only ‘beef’ I have with it is that the window area available for the net monitor is less than the window area available for the netconfig. This is due to the net monitor having the list of groups on the left, list of processes on the right and tabs across the top. So if I use the whole screen for the threads in the netconfig, when I switch the net monitor I need to scroll to see threads on the right hand side.
Scott
But at least it is useful, and related info! Thanks for the info on 5.5, it will help in my bid to management to upgrade.
Scott
I was on vacation.
Yes, Todd, turn the monitor daemon off while moving the threads then turn it back on.
There is no downtime. However, if someone else is running Netmonitor on other pcs, in the same site you are working on, their Netmonitors will show all dead too until you turn it back on.
Generally it doesn’t take that long to shut it down, move the thread and bring it back up.
BTW, I am not talking about shutting down the hostserver here.
We don’t have our operations team or anyone else run the Netmonitor so it isn’t a problem in our case.
Hope this helps,
-mh
My issue with stopping the monitord is the alert system ties off of that. When a major redesign of the Network Monitor layout is required, it takes some time to get things where you want them. Add to that, any time the monitord is started, any alert that is true fires off, and that generates a lot of email to a lot of folks that call me and ask me what happened.
So far, I am liking, and getting used to Russ’ method. It completely removes any custom groups we have set up, but once you get used to the spacing differences between NetConfig icons and NetMonitor icons, it works pretty well. I can get around the custom groups by making them real groups in the NetConfig, too.
Thanks for the clarification
Todd
You’re right, a major redesign is a pain. Sounds like 5.5 is the answer.
As far as firing when true right away…
We resolved this (most of this) by making the true condition stay true for 5 minutes. Generally the conditions will go back to false with in that 5 minute time frame.
hcisitectl -k m -s m -A “a=-cl ‘off.alrt'”
which loads my off.alrt file which has no alerts in it.
This is also usefull to stop additional unecessary alerts once the on-call person is in the site and working on the problem and everyone has already received several alerts but is getting annoyed and doesn’t want anymore alerts.
After the site has been up for a while and gets back to normal then I do
hcisitectl -k m -s m -A “a=-cl ‘default.alrt'”
Doing it this way also allows me to run this script to see which sites have alerts turn on or off.
#!/usr/bin/ksh
site_list=`list_sites.ksh`
for site in `list_sites.ksh`; do
if [ -a $HCIROOT/$site/exec/hcimonitord/pid ]; then
monitord_pid=`cat $HCIROOT/$site/exec/hcimonitord/pid`
cmd_args=`ps -p $monitord_pid -o “%a” | tail -1 | awk ‘{print $3}’`
echo “n Site ( $site ) hcimonitord pid ( $monitord_pid ) command/args ( $cmd_args )”
fi
done
echo “”
Russ Ross
RussRoss318@gmail.com