Help with a custom test script
list Q
Hello, I am new to writing custom scripts and looking for advice on how to write a test to monitor the status of a machine in our render farm. Each machine is broken down into a number of CPU cores. We have a command that displays the status of each core in the format per line of "<machine name>_<core number> in|out <partition name>" (i.e. testbox_0 in testpartition). So here is the part I an trying to work out. A machine with 4 cores will produce 4 lines each with it's own status. I would like to to include all cores in one test. I would like to produce a YELLOW status if any core is labeled as out and RED if they are all listed as out. I would also like to make rrd graphs that shows the in and out time for each core. Any help would be greatly appreciated. Q
list Henrik Størner
▸
On Sun, Aug 26, 2007 at 03:39:21PM -0700, Q wrote:
I am new to writing custom scripts and looking for advice on how to write a test to monitor the status of a machine in our render farm. Each machine is broken down into a number of CPU cores. We have a command that displays the status of each core in the format per line of "<machine name>_<core number> in|out <partition name>" (i.e. testbox_0 in testpartition). So here is the part I an trying to work out. A machine with 4 cores will produce 4 lines each with it's own status. I would like to to include all cores in one test. I would like to produce a YELLOW status if any core is labeled as out and RED if they are all listed as out. I would also like to make rrd graphs that shows the in and out time for each core. Any help would be greatly appreciated.
This doesn't do the graph, but it will give you the status you want:
#!/bin/sh
# Change this command to the one you use to collect the data
getthestatus >$BBTMP/corestat.$$
OUTCOUNT=`grep -c " out " $BBTMP/corestat.$$`
if test "$OUTCOUNT" = "4"
then
COLOR=red
elif test "$OUTCOUNT" -gt 0
then
COLOR=yellow
else
COLOR=green
fi
# Note: This command spans multiple lines
$BB $BBDISP "status $MACHINE.cores $COLOR `date`
$OUTCOUNT cores out
`cat $BBTMP/corestat.$$
"
rm $BBTMP/corestat.$$
exit 0
Regards,
Henrik
list Q
Thanks Henrik, That did the trick for status and helped me write a few more testes. . Your help is very much appreciated. I still need to work out how to do rrd data I am starting to do research on that but if anyone has any suggestions I would be glad to see them. Q
▸
Henrik Stoerner wrote:On Sun, Aug 26, 2007 at 03:39:21PM -0700, Q wrote:I am new to writing custom scripts and looking for advice on how to write a test to monitor the status of a machine in our render farm. Each machine is broken down into a number of CPU cores. We have a command that displays the status of each core in the format per line of "<machine name>_<core number> in|out <partition name>" (i.e. testbox_0 in testpartition). So here is the part I an trying to work out. A machine with 4 cores will produce 4 lines each with it's own status. I would like to to include all cores in one test. I would like to produce a YELLOW status if any core is labeled as out and RED if they are all listed as out. I would also like to make rrd graphs that shows the in and out time for each core. Any help would be greatly appreciated.This doesn't do the graph, but it will give you the status you want: #!/bin/sh # Change this command to the one you use to collect the data getthestatus >$BBTMP/corestat.$$ OUTCOUNT=`grep -c " out " $BBTMP/corestat.$$` if test "$OUTCOUNT" = "4" then COLOR=red elif test "$OUTCOUNT" -gt 0 then COLOR=yellow else COLOR=green fi # Note: This command spans multiple lines $BB $BBDISP "status $MACHINE.cores $COLOR `date` $OUTCOUNT cores out `cat $BBTMP/corestat.$$ " rm $BBTMP/corestat.$$ exit 0 Regards, Henrik