Isilon Mitrend Data Gathering Procedure

Mitrend is an extremely useful IT Infrastructure analysis service. They provide excellent health, growth and workload profiling assessments.  The service can process input source data from EMC and many non-EMC arrays, from host operating systems, and also from some applications.  In order to use the service, certain support files must be gathered before submitting your analysis request.  I had previously run the reports myself as an EMC customer, but sometime in the recent past they removed that ability for customers and it is now restricted to EMC employees and partners. You can of course simply send the files to your local EMC support team and they will be able to submit the files for a report on your behalf.  The reports are very detailed and extremely helpful for a general health check of your array, data is well organized into a powerpoint slide presentation and raw data is also made available in excel format.

My most recent analysis request was for Isilon, and below are the steps you’ll need to take to gather the appropriate information to receive your Isilon Mitrend report.  The performance impact of running the data gather is expected to be minimal, but in situations where the performance impact may be a concern then you should consider the timing of the run. I have never personally had an issue with performance when running the data gather, and the performance data is much more useful if it’s run during peak periods. The script is compatible with the virtual OneFS Simulator and can be executed and can be tested prior to running on any production cluster. If you notice performance concerns while the script is running, pressing Control + C in the console window will terminate it.

Obtain & Verify isi_gather_perf file

You will need to obtain a copy of the isi_gather_perf.tgz file from your local EMC team if you don’t already have a copy.  Verify that the file you receive file is 166 KB in size. To verify the isi_gather_perf.tgz is not corrupted or truncated you can run the following command once the file is on the Isilon cluster.

Isilon-01# file /ifs/isi_gather_perf.tgz

Example of a good file:

Isilon-01# file /ifs/isi_gather_perf.tgz /ifs/isi_gather_perf.tgz:
gzip compressed data, from Unix, last modified: Tue Nov 18 08:33:49 2014
data file is ready to be executed

Example of a corrupt file:

Isilon-01# file /ifs/isi_gather_perf.tgz /ifs/isi_gather_perf.tgz:
data file is corrupt

Once you’ve verified that the file is valid, you must manually run a Cluster Diagnostics gather. On the OneFS web interface, navigate to Cluster Management > Diagnostics > Gather Info and click the “Start Gather” button. Depending on the size of the cluster, it will take about 15 minutes. This process will automatically create a folder on the cluster called “Isilon_Support”, created under “ifs/data/”.

Gather Performance Info

Below is the process that I used.  Different methods of transferring files can of course be used, but I use WinSCP to copy files directly to the cluster from my Windows laptop, and I use putty for CLI management of the cluster via ssh.

1. Copy the isi_gather_perf.tgz to the Isilon cluster via SCP.

2.  Log into the cluster via ssh.

3. Copy the isi_gather_perf.tgz to /ifs/data/Isilon_Support, if it’s not there already.

4. Change to the Isilon Support Directory

 Isilon-01# cd /ifs/data/Isilon_Support

5. Extract the compressed file

 Isilon-01# tar zxvf /ifs/data/Isilon_Support/isi_gather_perf.tgz

After extraction, a new directory will be automatically created within the “Isilon_Support” directory named “isi_gather_perf”.

6. Start ‘Screen’

 Isilon-01# screen

7.  Execute the performance gather.  All output data is written to /ifs/data/Isilon_Support/isi_gather_perf/.  Extracting the file creates a new directory named “isi_gather_perf” which contains the script “isi_gather_perf”.  The default option gathers 24 hours of performance data and then creates a bundle with the gathered data.

Isilon-01# nohup python /ifs/data/Isilon_Support/isi_gather_perf/isi_gather_perf

8. At the end of the run, the script will create a .tar.gz archive of the capture data to /ifs/data/Isilon_Support/isi_gather_perf/. Gather the output files and send them to EMC.  Once EMC submits the files to Mitrend, it can take up to 24 hours for them to be processed.

Notes:

Below is a list of the command options available.  You may want to change the frequency the command is executed and the length of time the command is run with the I and r options.

 Usage: isi_gather_perf [options]

 Options:
 -h, --help Show this help message and exit
 -v, --version Print Version
 -d, --debug Enable debug log output Logs: /tmp/isi_gather_perf.log
 -i INTERVAL, --interval=INTERVAL
 Interval in seconds to sample performance data
 -r REPEAT, --repeat=REPEAT
 Number of times to repeat specified interval.

Logs:

The logs are located in /ifs/data/Isilon_Support/isi_gather_perf/gathers/ and by default are set to debug level, so they are extremely verbose.

Output:

The output from isi_gather_info will go to /ifs/data/Isilon_Support/pkg/
The output from isi_gather_perf will be /ifs/data/Isilon_Support/isi_gather_perf/gathers/

 

 

 

 

 

 

 

Advertisements

Leave a Reply