Databases Reference
In-Depth Information
where objectType=\'GRIDDISK\' and name=\'${met}\' \
and metricValue \> ${MBMIN}"
done
for met in $IORQ
do
echo "Listing ${SCOPE} for ${met} when #requests > ${IORQMIN} IO requests/second"
dcli -g ${CELLGROUP} "cellcli -e list ${SCOPE} \
where objectType=\'GRIDDISK\' and name=\'${met}\' \
and metricValue \> ${IORQMIN}"
done
for met in $LAT
do
echo "Listing ${SCOPE} for ${met} when IO latency > ${LATMIN} microseconds"
dcli -g ${CELLGROUP} "cellcli -e list ${SCOPE} \
where objectType=\'GRIDDISK\' and name=\'${met}\' \
and metricValue \> ${LATMIN}"
done
[oracle@cm01dbm01 source]$ ./lst13-06-griddisk-iostats.sh metriccurrent
Listing metriccurrent for GD_IO_BY_R_LG_SEC when IO rate > 1 MB/second
... Output omitted
Listing metriccurrent for GD_IO_RQ_R_LG_SEC when #requests > 10 IO requests/second
... Output omitted
Listing metriccurrent for GD_IO_TM_R_LG_RQ when IO latency > 10000 microseconds
... Output omitted
[oracle@cm01dbm01 source]$
In Listing 13-7, we are reporting on grid disk metrics in which the I/O bandwidth exceeds 1 MB/second, the
number if I/O requests exceeds 10 I/Os per second, and when the average I/O latency exceeds 10 milliseconds, or
10,000 microseconds. These thresholds are defined at the beginning of the script—you can change these values to
report on different values as needed.
How It Works
The example in this recipe uses the
list metriccurrent
or
list metrichistory
command on a variety of grid disk
metrics using a
cellcli where
condition based on a specified set of modifiable thresholds. In Listing 13-7, we used
the
where metricValue >
clause to generate a report for all situations that crossed the thresholds hard-coded in the
variable declaration section at the top of the script.
■
as an alternative to the script in listing 13-7, it may be worthwhile to establish grid disk metric thresholds.
please see recipe 13-4 to learn how to configure metric thresholds for your exadata cells.
Note
Search WWH ::
Custom Search