Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

CMPG Detailed Overview

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

CMPG

Contents

1.0 Introduction
o 1.1 Getting CMPG Input Data Files
2.0 Usage
o 2.1 Supporting Files
o 2.2 CMPG Usage Steps
2.2.1 CMPG/CMDownload Best Practice Install Guide
2.2.2 CMPG Basic Options And Use
2.2.3 Simple Counter Manager Automation
o 2.3 Excel Macro Customizations
2.3.1 Auto zoom Macro
2.3.2 Reset Margins Macro
2.3.3 Aligned LUN Hide Macro
3.0 Customizations
o 3.1 System Graphs
o 3.2 Overview Graph
o 3.3 Detailed Graph
o 3.4 Data Sheet
o 3.5 Additional Counters
3.5.1 Artificial Counters
3.5.2 Composite Counter
o 3.6 Adding FLASH CACHE to the Archiver
4.0 Default Graphs
o 4.1 Overview Graphs
o 4.2 Detailed Graphs
o 4.3 Data Sheets
5.0 Troubleshooting
o 5.1 No performance meta-data

1.0 Introduction
The Counter Manager Performance Grapher (CMPG) is a Microsoft Windows executable
that can be used to produce performance graphs from Counter Manager data sent via
AutoSupport.

The output file is in the form of a Microsoft Excel workbook. The graphs and data sheets within
the workbook can be used for high level performance analysis. The workbook also forms part of
the input into the Performance Consulting Service.
Currently the tool is only available as an executable for Microsoft Windows platforms as it relies
heavily on Windows features to drive the automation of Microsoft Excel. It has been
successfully tested with Excel2007 and Excel2010. Support for Excel2003 has been dropped as
of v4 due to the limitations of this version of Excel.

1.1 Getting CMPG Input Data Files


There are many ways to access the CM data required to generate performance graphs
1. NetApp users with NetApp intranet access can download from the NGS AutoSupport
Search Tool. You must search using either the Hostname or the Sys ID.
2. NetApp users with NetApp intranet access can also use the cmdownload tool which can
download CM data and other supporting files from NetApps AutoSupport repository.
This tool is included in the CMPG download
3. Download directly from a running NetApp controller. For Data ONTAP 7.x systems, the
CM data files are located at /etc/log/.cm_hourly_stats_done , the location for Data
ONTAP 8.x systems is /etc/log/cm_stats_hourly. This file is compressed and doesn't
need uncompressing for use with CMPG. Please remove the dot from the start of this file
before processing with CMPG.
You can also use CMPG to process data from performance archive files. These files must be
downloaded directly from a NetApp controller and are located:

in Data ONTAP 7.3: /etc/stats/archive


in Data ONTAP 7.3.2+: a symlink made from /etc/log/stats/archive to /etc/stats/archive
in Data ONTAP 8.0+: /etc/log/stats/archive

2.0 Usage
CMPG is a Perl based script that is executed from the command line. For example:
cmpg.exe filename CM-HOURLY-STATS.gz

or
cmpg.exe filename stats_archive_XXX_YYY_ZZZ.gz -archive

The following CMPG command line options are available: -

-filename input-file

Specifies the input filename. You can graph longer periods by entering more files. You
can do this by either specifying -f multiple times or using wildcards. An example being:
cmpg -f netapp01\*.gz. This works for either CM data files or performance archive files.
You can easily download multiple CM data files using cmdownload.
-archive [FREQUENCY_1|FREQUENCY_2|FREQUENCY_3|FREQUENCY_4]

CMPG input is in performance archive format and optionally allows you to specify
which sampling frequency within the file to chart. The defaults are FREQUENCY_1 =
once per second, FREQUENCY_2 = once per 5 seconds, FREQUENCY_3 = once per
minute and FREQUENCY_4 = once every 5 minutes. Lower sampling frequencies take
more time to process and require more memory. Note that not all counters are sampled at
the lowest frequencies. The NFS operations histogram is one example.
-details

All detailed graphs are created.


-metadata meta-data-file

On some occasions meta-data is missing from CMDATA files and this prevents
meaningful data from being extracted and graphed. As of 3.0.2 the script will
automatically use the file metadata.xml. This should suffice for most situations, but this
behaviour can be overridden, if needed, using this option.
-storage [storage-file]

The filename supplied with this option contains output from the filer's storage command.
If you do not specify the file name then the script will attempt to locate a file ending with
'STORAGE' or 'storage.txt' in the same directory as the data files. This is only applicable
for CMDATA files.
-raid [sysconfig-r-file]

The filename supplied with this option contains output from the filer's sysconfig -r
command. If you do not specify the file name then the script will attempt to locate a file
ending with 'SYSCONFIG-R' or 'sysconfig.txt' in the same directory as the data files.
This is only applicable for CMDATA files.
-logo image-file

This option specifies the filename of a image that will be inserted into the top left corner
of the graphs
-xml xml-config-file

The name of a custom configuration file to use for producing graphs. By default CMPG
uses -x cmpg.xml
-perfInfo

Causes CMPG to dump out the performance information contained within the CM data
file. The output file is based upon the input file with -info.xml.

-noPrompt

Deletes an output file if it already exists


-noDataAverage

Prevents data points from being averaged out when multiple performance files are
supplied.
-gmtOffset [+|-]hhmm

Timestamps for CM data files are recorded in GMT. The data files, however, do not say
what time zone the controller is located in. Use this option to specify the controller's
GMT offset to make the charts accurate. The local time zone is used by default.
-analysis

Normal legends are replaced with selectable ones and standard deviation figures are
added to the data sheets.
-uncompThreads [number_of_threads]

Enables multithreaded environment for uncompressing downloaded archive files. Any


integer value representing number of parallel threads can be specified as argument to this
option.
o
o

If this option is specified without any argument then the number of


threads default to the number of processors.
If this option is specified with an integer argument then the numbers of
threads is the minimum of the two values i.e. integer argument or number
of processors in the system.
If this option is not specified then the number of threads defaults to the
value 1.

-output [output_directory_path]

Specifies the output directory to which the final output file after processing the data
should be placed. Currently only local and mapped drives are supported. If the path
contains directories with names having blank spaces in between then the whole path
should be enclosed in double quotes.
-version

Print the version information


-help

Print the help information


All options can be shortened as long as no ambiguity exists. For example: -filename can be
shortened to -f.

2.1 Supporting Files

The CMPG tool is distributed with these supporting files:


1. cmpg.xml An XML file containing the default set of graphs that are created by the tool
2. perfInfo.xsl An XML style sheet used to view counter information produced by the p
options
3. metadata.xml The meta-data used when none can be located in a CMDATA file
4. cmdownload.exe Used to download CMDATA files and other files from NetApps
AutoSupport repository
5. alignment.xml Used with the -x options to create alignment output
6. archive.xml Used with the -x option to create archive output. Also used with the archive option
7. cmpg.exe The main CMPG executable
8. df-trend.xml Used with the -x option to create space trending output
9. df.xml Used with the -x option to create space output
10. trend.xml Used with the -x option to create long term trending output

2.2 CMPG Usage Steps


CMPG/CMDownload Best Practice Installation Steps
Download CMPG into a unique directory on your PC that has a short path, such as the c:\cmpg
directory.
Extract the .ZIP file into this directory.
Find My Computer on your desktop, and right click it, go into Properties, Advanced tab, and
then Environment Variables
In the System Variables window, select Path, and then click on Edit.
We want to grow this variable, using a semicolon to add a new directory path to the end of the
existing statement. Please add your cmpg directory, eg., ;c:\cmpg. Hit save, and exit these
windows. Dont forget to add the semi-colon between path entries.
Open a new- DOS prompt/window, and enter the command CMPG. You should see CMPG
command output as a result. DOS windows that exist before these steps, will not have
cmpg/cmdownload in its path and will not work.

2.2.2 CMPG Basic Options and Usage Examples


1) Download and graph 4 weeks of performance data using cmdownload
(requires NetApp intranet access)

Downloads CM-STATS.gz and configuration metadata for the last four weeks to
subfolder that matches the storage controllers name
cmdownload -f <filername or sysid> -w 4

Or if you are using serial numbers


cmdownload -s <serialnumber> -w 4

Uncompress the .gz downloaded files before using thwm with cmpg.
cmpg -f

<filername>\*.stats -nodataaverage

Note: the -nodataaverage stops data points being averaged and peaks being hidden
cmpg f <filername>\*.stats -details

The -details flag is needed to include individual loop charts, and expose performance
charts per volume and per lun
cmpg -f <filername>\*.stats -output Z:\cmpg\resultDirectory

The -output flag can be used to place the result file into the specified destination
directory, which can be a network path mapped as Win
Note: Internet Explorer has a habit of automatically uncompressing the GZIP file downloaded from
ASUP search and leaving the .gz extension. If you have problems uncompressing the file then just open it
in a text editor and see if you can see plain text.

2) Download and graph the last 24 weeks performance data


cmdownload -f <filername or sysid>

-w 24

Or if you are using serial numbers


cmdownload -s <serialnumber> -w 24
cmpg f <filername>\*.stats -x trend.xml

3) Download and graph the last 24 weeks space utilization

cmdownload -f <filername or sysid> -w 24 -c DF c DF-A -c DF-S -c DF-R


-zip

Or if you are using serial numbers


cmdownload -s <serialnumber> -w 24 c DF c DF-A -c DF-S -c DF-R -zip
cmpg f <system\system.zip>\*DF* -x df.xml

4) Download and project storage growth for aggregate and volumes for the last
52 weeks
cmdownload f <system> -w 52 -c DF c DF-A c DF-S c AGGR-STATUS zip
cmpg f <system\system.zip> -x df-trend.xml

5) Download and graphs LUN alignment for a system


cmdownload -f <filername or sysid>
cmpg f <filername>\*.stats x alignment.xml

This can produce a large number of charts and it can be laborious to search for bad
alignment. Please see section 2.3.3 for the excel macro that can be used to hide all the
charts with good alignment leaving only charts that indicate bad alignment.
Note: the recent slides on LUN Misalignment are here
6) Get the Performance Archiver data from a filer running 7.3.x
All of the examples here can be run with these files using the archive option at
the end of CMPG.
In ONTAP 7.3: /etc/stats/archive/
In ONTAP 7.3.2+: a symlink made from /etc/log/stats/archive to /etc/stats/archive
In ONTAP 8.0+: directory moved from /etc/stats/archive to /etc/log/stats/archive
Note: the performance archiver does not contain processor domain, LUN or
FlexScale counters.
2.2.3 Simple Counter Manager Automation
This batch file will save the users having to open a DOS prompt and simplify cmdownload /
CMPG use. Please download and save this 'scma-batch.bat' file to the CMPG working directory
and create a shortcut to it on your desktop.

2.3 Excel Macro Customizations


There are a number of macros that generate nicely formatted CMPG workbooks. The following
instructions show how to add these to your default Excel instance, so these Macros are loaded by
default when Excel is opened. These instructions are for Excel 2007.
1. Enable the developer tools in Excel. Ribbon button>Excel Options. Then enable
Developer Tools.

2. Now, record a macro.

3. Store the macro in your Personal Macro Workbook.

4. Stop the recording.

5. Right-click the sheet tab and go to the Visual Basic editor.

6. Double click on Module1 and overwrite the macro with one of the macros below on the
right-hand side.

2.3.1 Auto Zoom Macro

This macro allows you to automatically zoom all worksheets in the workbook to 75% and
addresses viewing and screen resolution issues. Copy and paste the following using the
instructions above: Sub SetZoom()
' Declare the variables
Dim ac As Excel.Chart
Dim c As Excel.Chart
' Record the active chart
Set ac = ActiveChart
' Stop Excel updating stuff as it just slows the macro down
Application.ScreenUpdating = False
Application.Interactive = False
' Select each chart and set the zoom level
' Cannot use charts.select for some reason
For Each c In ActiveWorkbook.Charts
c.Select
ActiveWindow.Zoom = 75
Next
' Make sure the original chart is now selected
ac.Select
' Set these back
Application.ScreenUpdating = True
Application.Interactive = True
End Sub

2.3.2 Reset Margins Macro


This macro automatically reset the margins on all worksheets in the workbook to 'narrow' as seen
in the toolbar under Page Layout. This macro improves CMPG chart generation because it no
longer cuts off the right axis of the charts. Copy and paste the following using the instructions
above:
Sub SetMargins()
' Declare the variables
Dim ac As Excel.Chart
Dim c As Excel.Chart
' Record the active chart
Set ac = ActiveChart
' Stop Excel updating stuff as it just slows the macro down
Application.ScreenUpdating = False
Application.Interactive = False

' Select each chart and set the margins


' Cannot use charts.select for some reason
For Each c In ActiveWorkbook.Charts
ActiveSheet.PageSetup.LeftMargin = Application.InchesToPoints(0.25)
ActiveSheet.PageSetup.RightMargin = Application.InchesToPoints(0.25)
ActiveSheet.PageSetup.TopMargin = Application.InchesToPoints(0.75)
ActiveSheet.PageSetup.BottomMargin = Application.InchesToPoints(0.75)
ActiveSheet.PageSetup.HeaderMargin = Application.InchesToPoints(0.3)
ActiveSheet.PageSetup.FooterMargin = Application.InchesToPoints(0.3)
Next
' Make sure the original chart is now selected
ac.Select
' Set these back
Application.ScreenUpdating = True
Application.Interactive = True
End Sub

2.3.3 Aligned LUN Hide Macro


This macro will hide all the LUN charts that show alignment, and show all the misaligned LUNs
that need closer attention. Copy and paste the following using the instructions above: Sub ShowBadAlignment()
' Don't care if we error
On Error Resume Next
' Stop the screen updating, might make little difference here
Application.ScreenUpdating = Fals
' Limits
Dim AlignLimit As Double
Dim OpsLimit As Double
' Only show charts if
AlignLimit = 0.8 ' Correctly aligned are less then this percentage
OpsLimit = 10
' Operations per second is greater than this
' Set the variables
Dim c As Excel.Chart
Dim s As Series
' Loop through each chart in the current workbook
For Each c In Application.ActiveWorkbook.Charts
' Hide the sheet to start with
c.Visible = xlSheetHidden
Dim Align As Double
Dim Ops As Double
' Get the 0 bucket for the histogram

Set s = c.SeriesCollection("0")
' Is this a valid object?
If IsObject(s) Then
' Get the average value for the series
Align = Application.WorksheetFunction.Average(s.Values)
End If
' Get the operations for the histogram
Set s = c.SeriesCollection("Ops.")
' Is this a valid object?
If IsObject(s) Then
' Get the average value for the series
Ops = Application.WorksheetFunction.Average(s.Values)
End If
' Does this chart match the criteria?
If Align < AlignLimit And Ops > OpsLimit Then
c.Visible = xlSheetVisible
Debug.Print c.Name
End If
Next
End Sub

3.0 Customizations
CMPG relies upon an external XML file for configuration. The configuration tells CMPG what
graphs to output and what counters should be used. CMPG can output two different types of
graphs and one type of data sheet. Three default system graphs that are included in every run of
CMPG and these default graphs cannot be configured.
3.1 System Graphs
The default graphs Protocols, Inbound Throughput, and Outbound Throughput are always
included. The principal reason is that additional coding is required to place the CPU utilization
area graph on the secondary axis and this cannot yet be specified in the XML configuration.
The following are examples of the three default system graphs: -

Figure 1: Protocols Graph

Figure 2: Inbound Throughput

Figure 3: Outbound Throughput

3.2 Overview Graph


The overview graph is used to display counters from multiple objects of the same type on a
single graph. For example, this allows for comparison between NIC throughput or volume
latency. It is a good practice to keep the number of counters to a minimum since there can be
many objects of a single type within a system.

The overview graph is specified in the XML configuration by using the <overview-graph> tag.
For example: <config>
<overview-graph title="Top 5 LUN Reads" max-series="5" remove-series=1 />
<counters lun:read_data=""/>
</overview-graph>

<config>

The following options can be specified for the overview graph: title : This will be the title of the graph. This will also be used to form the name of the
graph
axis-title : The title of the left axis. This is automatically generated using the
performance meta-data and is the units of the counter being used. Where
multiple counters are used the predominant units will be used. Any title
specified here will be prepended to the automatic title.
max-series : Specifies the maximum number of series to show on the graph. The series
are ordered and then the least significant ones are removed until maxseries is reached
The valid values for this option is any whole number above 0
remove-series : With this option specified the scale of the graph is recalculated to ignore
the first 7 timestamps. This should prevent the regular disk scrubbing
process from skewing the scale
The valid values for this option are 0 and 1
The <overview-graph> must contain a <counters> tag. The counters tag itself must contain at
least one pair of counter name and counter title pairs. The counter title is used to generate the
name for a series in a graph. If counter title is empty then the series name will contain only the
instance name for the counters.

Figure 4: Overview Graph

3.3 Detail Graph


The detail graph can be used to display multiple counters for object instances of a single type.
Each instance will be plotted on a single graph; allowing for more counters to be specified
without complicating the output.
The detail graph is specified in the XMLconfiguration by using the <detail-graph> tag. For
example: <config>
<detail-graph title="Vol %s"/>
<counters
volume:read_data="Reads"
volume:write_data="Writes"
volume:avg_latency="Average Latency" />
<secondary-axis title="Latency" counters="volume:avg_latency"/>
</detail-graph>

</config>
Note: Detailed graphs are only created if the -detail option is specified in the command line.

The following options can be specified for the overview graph: -

title : This will be the title of the graph. This will also be used to form the name of the
graph
title : The title on the top of each chart created. One chart per volume in the case. The %s
in this example specifies cmpg to label the worksheet volume:<vol_name>
The <detail-graph> tag must contain a <counters> tag. The counters tag itself must contain at
least one pair of counter name and counter title pairs. The counter title is used to generate the
name for a series in a graph. If counter title is empty then the series name will contain only the
instance name for the counters.
There is also an optional <secondary-axis> tag for the detail graph. This option allows for
counters with different measurement units or drastically different scales to be plotted on the same
graph. The following options can be specified for this tag: title : The title of the right axis. This is automatically generated using the performance
meta-data and is the units of the counter being used. Where multiple
counters are used the predominant units will be used. Any title specified
here will be prepended to the automatic title.
counters : This option specifies the comma separated list of counters that should be
graphed on the secondary axis.

Figure 5: Detail Graph

3.4 Data Sheet


A data sheet will output minimum, maximum and average values for specified counters for each
instance of an object type in a tabular format.
The data sheet is specified in the XML configuration by using the <data-sheet> tag. For
example: <config>
<data-sheet title="ProtocolStats">
<counters

/>
/>
system:fcp_ops="FCP" />
system:iscsi_ops="iSCSI" />
system:http_ops="HTTP" />
system:cpu_busy="CPU" />
system:cifs_ops="CIFS"
system:nfs_ops="NFS"

</data-sheet>

</config>

The following options can be specified for the overview graph: title : This will be the name of the worksheet. This is limited to the 32 characters and
cannot contain certain special characters.
As with the graphs; each <data-sheet> tag must contain a <counters> tag. The counters tag itself
must contain at least one pair of counter name and counter title pairs.

Figure 6: Data Sheet

3.5 Additional Counters


Executing the tool with the filename and printInfo options will produce an XML file that
contains all the information regarding the counters available within the CM data file. This file
can then be opened in a browser. First copy the CMPG included file perfinfo.xsl into the
directory where your data resides, and then execute: -

cmpg.exe f CM-HOURLY-STATS.gz -perfInfo


Will produce an XML file that will resemble the following when opened in a browser: +system object
+disk object
-processor object
-processor_busy
Description: Percentage of elapsed time that the processor is executing non-idle processes
Properties: percent
Units: percent
+ifnet object
+nfsv3 object
+target object
+lun object
+volume object
+cifs object
+fcp object
+iscsi object
+aggregate object
+qtree object

Figure 7: CM Data Counter Information


Any of the counters listed for a specific object are available to be graphed. Overview graph with
counters from different objects in the same worksheet can also be plotted. End user can modify
the configuration XML accordingly to achieve the mentioned capability
Diagnostic mode counters are not normally available and are therefore normally hidden in the
XML document. The known exception is the domain_busy counter which is available with
ONTAP versions of 7.2.4 despite being labeled a diagnostic counter.
Counters that consist of an array of values are accessed by following the counter name with a
colon and then the label name of the sub counter. For example, the diagnostic counter
nfsv3_op_count contains an array of operations.
-nfsv3 object
-nfsv3_op_count
Description: Array of select NFS v3 operation counts
Properties: delta
Units: none
Type: array
Labels: null, getattr, setattr, lookup, access, readlink, read, write, create, mkdir, symlink, mknod, remove, rmdir, rename, link,
readdir, readdirplus, fsstat, fsinfo, pathconf, commit

The sub counters can be accessed by the format nfsv3_op_count:getattr


Note
A feature in Mozilla Firefox 3.0 prevents the XML file from displaying properly. Either use Microsoft Internet
Explorer or, in Firefox, navigate to the about:config page and change the option
security.fileuri.strict_origin_policy to false

3.5.1 Artifical Counters


diskloop:user_read_throuhgput
The user_read_througput is calculated as the total blocks read from each disk in a loop
multiplied by the average chain length for the disk IO and then multiplied by 4. The figure gives
throughput in kilobytes per second. Fabric MetroCluster systems will show throughput per
switch port and not per FC initiator.
diskloop:user_write_throuhgput
The user_write_througput is calculated as the total blocks written to each disk in a loop
multiplied by the average chain length for the disk IO and then multiplied by 4. The figure gives
throughput in kilobytes per second. Fabric MetroCluster systems will show throughput per
switch port and not per FC initiator.
diskloop:cp_read_throughput
The cp_read_througput is calculated as the total blocks read from each disk in a loop in order to
satisfy user writes multiplied by the average chain length for the disk IO and then multiplied by
4. The figure gives throughput in kilobytes per second. Fabric MetroCluster systems will show
throughput per switch port and not per FC initiator.
system:input_ratio
The input ratio is calculated as disk writes divided by the sum of network and fibre channel
writes. Input ratio should vary between 0 and 1 with 1 being the ideal.
system:output_ratio
Output ratio is the disk reads divided by the sum of network and fiber channel reads. Output
ratios should range from 1 upwards with higher numbers being ideal as they indicate higher
cache hits.
3.5.2 Composite Counter
Composite Counters enables complex math to be done on counters and graphed as normal
counters
<composite-counter name="concurrency" object="volume" unit="none">
<function type="add" counter="volume:total_ops"/>
<function type="product" counter="volume:avg_latency"/>

<function type="divide" value="1000"/>


</composite-counter>

3.6 Adding FLASH CACHE to the Archiver


Please follow these steps:
1. On 7.3 copy default file to a new file call custom1.gz in the same directory.
\etc\stats\archive\.preset
2. On 8.0.1X24 the file location will be /mroot/etc/log/stats/archive/.preset
3. Extracted custom1.gz so you have custom1 file in the directory.
4. Edit this file and add the following at some point. You dont have to gzip it up again.
<object name="ext_cache_obj">
<counter name="accesses" archive="FREQUENCY_1">
</counter>
<counter name="hit" archive="FREQUENCY_1">
</counter>
<counter name="hit_percent" archive="FREQUENCY_1">
</counter>
<counter name="hit_normal_lev0" archive="FREQUENCY_1">
</counter>
<counter name="hit_metadata_file" archive="FREQUENCY_1">
</counter>
<counter name="hit_directory" archive="FREQUENCY_1">
</counter>
<counter name="hit_indirect" archive="FREQUENCY_1">
</counter>
<counter name="miss" archive="FREQUENCY_1">
</counter>
<counter name="miss_metadata_file" archive="FREQUENCY_1">
</counter>

<counter name="miss_directory" archive="FREQUENCY_1">


</counter>
<counter name="miss_indirect" archive="FREQUENCY_1">
</counter>
<counter name="inserts" archive="FREQUENCY_1">
</counter>
<counter name="evicts" archive="FREQUENCY_1">
</counter>
<counter name="usage" archive="FREQUENCY_1">
</counter>
</object>
5. On the console do the following.
filer> options stats.archive.enable off
Fri Sep 11 08:12:22 GMT [perf.archive.stop:info]: Performance archiver
stopped.
filer> options stats.archive.preset.level custom1
filer> options stats.archive.enable on
Fri Sep 11 08:16:32 GMT [perf.archive.start:info]: Performance
archiver started. Sampling 21 objects and 203 counters.

6. Run this CMPG command


cmpg -f <file> -archive -details

4.0 Default Graphs


CMPG in its default configuration will produce a number of graphs which are explained below.

4.1 Overview Graphs


Graph

Description

Protocols

Shows the average activity of the different protocols on the storage controller. This information is super
imposed on top of overall CPU utilisation which is plotted on the secondary axis.

Operations
Outbound

Shows the average activity of the outbound Ethernet and Fibre Channel interfaces against disk reads. This
graph is also overlaid on top of average utilisation.

Operations
Inbound

This graph is the opposite of the Operation Outbound graph and shows inbound activity against disk writes.

All
Processors

Shows the average utilisation of each installed processor

V Filers

Shows the average processor time that is allocated to each virtual filer configured on the storage controller.
The availability is this graph is dependant upon the configuration of the storage controller.

Domain
Usage

Shows the average time that each domain has been executing on any processor. The axis scale will vary
according to the number of processors in the storage controller. The availability of this graph is dependant
on the version of ONTAP installed on the storage controller.

All Disk IOps

Shows the average number of IO operations per disk in the system. This graph may not show all disks
installed in the system due to limitations within Microsoft Excel.

NIC Receives

Shows the average amount of data received by each Ethernet interface in the storage controller. Interfaces
that are part of a virtual interface or marked as down are not displayed.

NIC
Transmits

Shows the average amount of data transmitted by each Ethernet interface in the storage controller.
Interfaces that are part of a virtual interface or marked as down are not displayed.

FC Target
Receives

Shows the average amount of data received by each fibre channel target interface in the storage controller.
Series with no data are removed to improve clarity of the graph.

FC Target
Transmits

Shows the average amount of data transmitted by each fibre channel target interface in the storage
controller. Series with no data are removed to improve clarity of the graph.

Top 5 Vol
Writes

Shows the average write activity for the 5 busiest volumes for writes

Top 5 Vol
Reads

Shows the average read activity for the 5 busiest volumes for reads

Top 5
Volume
Latencies

Shows average latency for the 5 busiest volumes for latency

Top 5 LUN
Writes

Shows the average write activity for the 5 busiest LUNs for writes

Top 5 LUN
Reads

Shows the average read activity for the 5 busiest LUNs for reads

4.2 Detailed Graphs


Graph

Description

ISCSI

Shows average iSCSI reads and writes against iSCSI operations

LUNs

Shows the average read, write and average latency for each configured LUN on the storage controller

Volumes

Shows the average read, write and average latency for each volume on the storage controller

4.3 Data Sheets


Graph

Description

LUN Stats

Shows the minimum, average and maximum values for reads, writes and average latency

Volume Stats

Shows the minimum, average and maximum values for read operations, write operations and average latency

Input Stats

Shows the minimum, average and maximum values for Ethernet receives, Fibre Channel receives and disk
writes

Output Stats

Shows the minimum, average and maximum values for Ethernet transmits, Fibre Channel transmits and disk
reads

Protocol Stats

Shows the minimum, average and maximum values for the CIFS, HTTP, iSCSI, NFS and FCP protocols

5.0 Troubleshooting
5.1 No performance meta-data
Symptom
CMPG reports that the performance meta-data is missing from the data file
Explanation
Occasionally the performance meta-data section of the CM stat file is missing. If the
uncompressed file is opened using a text editor then the start of the file will look like: <perf-info></perf-info><perf-data>...

Resolution
The workaround is to find a CM stat file from a system with the same version number and
replace the perfinfo section of the problem file with that of the good file.
Below are some links to data required if you see the issues above. Simply insert the contents of
this file in between the perfinfo sections. <perf-info> PutTheDataHere </perf-info>
7.0.3
7.2.5.1
7.2.6.1
7.3.1.1
Another way to fix this is to copy one of the text files from below, place a <perf-info> block at
the front and a </perf-info> block at the end. Next open a command prompt window and then
type the following: type <perfInfoFile> > temp.txt
type <uncompressedCMSTATS> >> temp.txt
ren temp.txt <somethingMoreUseful>

You might also like