Unsolved

This post is more than 5 years old

31 Posts

1363

December 3rd, 2009 08:00

Nice graphs of filesystem load (bytes in/out, IOPS, %r/w, ...)

Hello,

Lately I've been exploring both Celerra Monitor (GUI) and 'server_stats' (Command Line Interface).

I'd like to be able to automatically (hourly, daily, weekly) generate load/utilization graphs à la Celerra Monitor

without needing to log on.

Right now I'm periodically pulling the output of 'server_stats' from the system, with intention to throw

the CSV files at some 2D plotting tool.

Am I re-inventing the wheel here ?

Eric Luyten, Computing Centre VUB/ULB.

Moderator

 • 

285 Posts

December 3rd, 2009 08:00

The way you're doing it is the only way to automate it.  You can cron a script to generate the output you're looking for.

The only other method to create the CSV files is directly off the graph.  You can right-click on the graph and select Export and this will generate the CSV file of the view you have selected.

59 Posts

December 3rd, 2009 10:00

I did this in my company.  Created local cron jobs on cs0 and cs1 (ending script if not primary cs), writing a line to a csv file every minute with perfformance data.  Then ad midnight from my management host I scp'd the files and used gdgraph to create pretty graphs and email them out.  Worked very very well.  I did similar things for file system used, quota usage, and others to trend data long term.  This also allowed me to keep a repsoitory of historical perf data in case of issues or needing to compare with old stats for new issues.

I would say do it, and it would be nice if EMC had some of this capability built into the system.

Sagle

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

December 3rd, 2009 20:00

Bill,

is it possible to query nas database or log files that store historical data (where is it stored anyway) ?  You mentioned that it rolls off every 14 days so if it were possible to grab that data before it gets overwritten and append it to one of my data stores to be plotted later on.

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

December 8th, 2009 13:00

Rainer,

do you know where Celerra keepsthese stats ?

6 Operator

 • 

8.6K Posts

December 8th, 2009 13:00

sorry - I dont - I guess the one's for Celerra Manager and Monitor historical data are stored somewhere in a not really usable format

I think the stats provided by server_stats are pulled from the data mover in realtime

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

December 8th, 2009 13:00

do you have friends in engineering ..can you find out please . I would rather parse XML output of something that's already been stored.

6 Operator

 • 

8.6K Posts

December 8th, 2009 13:00

I would try to get it into MRTG or RRD - no need to re-invent the wheel for the plotting and storage

1 Rookie

 • 

26 Posts

December 9th, 2009 08:00

If you already have an EMC Control Center implementation, you could discover the Celerra's, setup WLA (work load analyzer) collections, and setup report scheduler in Performance Manager. This facilitates the collection and storage of historical stats, and automates the creation of graphs.

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

December 10th, 2009 03:00

i do have ECC but not licensed for NAS agents at the moment. I'll ask my TC for an eval license to see if it has metrics i am interested in ..otherwise it's server_stat and perl for me. Thanks telenoiz

6 Operator

 • 

8.6K Posts

December 10th, 2009 04:00

another alternative would be Data Protection Advisor (DPA)

its gotten a lot of Celerra functionality in the last couple of releases and the Celerra license there are less expensive than ECC

20091023 - DPA 5 5 und Celerra.pdf

Rainer

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

December 10th, 2009 08:00

looks nice, although there has already been made investment into ECC so will try the NAS agent first. Thank you for recommending DPA.

26 Posts

August 11th, 2010 02:00

Hi,

Thanks for the wonderful knowledge sharing.

I would like to know, what are the various options for  server_stats command, to fetch necessary details and then...  we can exporting to excel or csv according to the customer requirements.

would be helpful with various examples on server_stats to find Nework interfaces efficiency etc...

Thanks

regards

VJ

No Events found!

Top