Active IQ Unified Manager Discussions

dfm perf data retreive a specific volume

JGPSHNTAP
3,884 Views

so, i want to see statistics from the cli for a specific volume and I'm trying -C volume:volumename:cifs_read_latency but it keeps giving me all the volumes.

 

Is there a way to get just a single volume?

11 REPLIES 11

bobshouseofcards
3,860 Views

You will want to use the "-o" option on the "dfm perf data retrieve" command to specify a specific object whose counters you wish to query.

 

C:\>dfm perf data retrieve help

NAME
    retrieve -- This command allows you to extract the counter data
                with supported statistical calculations on them.

SYNOPSIS
    dfm perf data retrieve { [ -o object-name-or-id ... ] [ -C perf-counter ... ]
                           [ -V view-name ] }
                           [ -d duration ] [ -b start-time ] [ -e end-time ]
                           [ -M month ... ] [ -D weekday ... ] [ -T time-range ... ]
                           [ -m statistical-method [ -P percentile-value ]
                           [ -S data-advance-method ] ]
                           [ -s sample-rate ] [ -x output-format ] [ -R ]

 

Description of options:

    -o Object instance for which data is to be retrieved. Multiple object
       instances can be specified. Specifying a parent will retrieve data
       for all its children.

    -C A counter of the form object-name:counter-name (ex. system:cpu_busy).
       Multiple counters can be specified.

 

 

So in your case, the CLI would be something like "dfm perf data retrieve -o yourvolumename -C volume:cifs_read_latency".  Note that you don't put the volume name in the counter parameter.  You can put multiple "-C" parameters on a single command line if you need multiple counters for the same object, for instance:

 

dfm perf data retrieve -o yourvolumename -C volume:cifs_read_latency -C volume:cifs_write_latency -C volume:cifs_other_latency

 

If you use multiple counters within a single retrieve another option to consider is "-x TimeIndexed" which organizes the output in a tabular format by sample time.  If you specify counters that are updated on un-correlated schedules or if you specify multiple objects that might have uncorrelated collection schedules you may run into some odd results on a TimeIndex report, since DFM tries to make sense of all the data in a single aggregation.

 

Hope this helps you.

 

Bob Greenwlad

Lead Storage Engineer, Huron Legal

Huron Consulting Group

NCDA | NCIE-SAN Clustered Data OnTap

 

 

 

JGPSHNTAP
3,859 Views

That doesn't seem correct.  -o you can't use volume name.. Its telling me there is no object

bobshouseofcards
3,848 Views

Question - 7-mode or clustered Data OnTap?

JGPSHNTAP
3,841 Views

This is a 7-mode environment.  There is CLI for cdot

bobshouseofcards
3,827 Views

To get the form of the volume instance names you need, run the command:

 

dfm perf counter list <controller>:volume:*:*:*:*

 

substituting the name of your 7-mode controller.  If I remember right in 7-mode the volume name form is either controller-name:/volume-name or controller-name:/vol/volume-name, but the counter list command will display the form you need.

 

JGPSHNTAP
3,826 Views

I'm sorry , but these aren't working either..

bobshouseofcards
3,809 Views

Ok - so the really old fashioned steps:

 

Please confirm the version of OnCommand Unified Manager - just for grins.

 

Can you display the output of the "dfm perf counter list" command with what you entered and what it gave you back?

 

JGPSHNTAP
3,804 Views

We are using 5.2.1 - latest for that platform.

 

dfm perf counter list doesn't give us back anything

 

bobshouseofcards
3,773 Views

It gave you nothing? As in just the prompt returned?

 

Literally, to continue I need to see cut/paste of what you typed and what came back.  It should have at least generated an error message.

 

Reference dfm perf cli

JGPSHNTAP
2,441 Views

Ok, Im pretty well versed in dfm, and I appreciate your help, which is leading me to more confusion as to why this doesn't work.

 

 dfm perf data retrieve -o filer1 -C volume:volumeb:cifs_read_latency -C volume:volumebS:nfs_read_latency -s 300 -b
"2015-04-01 16:00:00" -e "2015-04-03 21:30:00" 

 

That returns all the volumes on filer1 cifs_read and nfs_read latency

bobshouseofcards
2,441 Views

Also - please check whether Performance Advisor is enabled (just one of the "obvious" things to verify).

 

C:>dfm option list perfAdvisorEnabled
Option Value
------------------ ------------------------------
perfAdvisorEnabled Enabled

 

If it isn't enabled, definitely turn it on.  Most performance counter data isn't collected if PerformanceAdvisor isn't enabled.

Public