Issue with nmon analysing

Dear Experts,

Am using nmon analyser version 3.3g, and when i tried analyzing my current nmon file it was done successfully. But the problem is all my old nmon files are compressed. So i uncompressed and analyzed using the same analyzer. but got this error

no valid input data! nmon run may have failed
Is it because of the nmon file size as it is of more than 1.5 MB ? The file which was successful was of 989KB. If yes then can anyone please tell me how to analyze such a big files like even 4MB.
I even tried nmon analyzer V 3.4 which is the latest one but ended up with same result.
Im lost in here... can any one please help me out....

Thanks all in advance..:):):):)

I doubt it has to do with a file size but with the structure of the data in the files.
Try to install an older version with that it worked. Maybe there is some incompability between versions.
You could also ask in the IBM DeveloperWorks Performance Tools forum, where the author of nmon Nigel Griffiths (nickname "nagger" iirc) is often present. If he doesn't know a solution he can for sure give a hint.

With nmon analyzer - it can be file size as Excel has its own set of limits.

re: developers. Nigel might be able to help, but he developed the tool, not the analyzer.
And the initial "analyzer developer" has retired as I understand it.

Now, shorten yoru nmon files (to test for size issues) - you should be able to cut off the tail of "unsorted" nmon files.

The data collection starts with the line T0001 - which is a timestamp/keyword; BBBP is the last line of "static configuration information"

AAA,progname,topas_nmon
AAA,command,/usr/bin/topas_nmon -X -youtput_dir=/home/michael/x054 -ystart_time=05:59:16,Nov05,2012
AAA,version,TOPAS-NMON
AAA,build,AIX
AAA,disks_per_line,150
...
BBBP,707,netstat,""
BBBP,708,netstat,"Route Tree for Protocol Family 24 (Internet v6):"
BBBP,709,netstat,"::1%1              ::1%1             UH        1        56 lo0      -      -   "
BBBP,710,emgr -l
ZZZZ,T0001,05:59:28,05-NOV-2012
CPU01,T0001,27.4,59.3,1.3,11.9
PCPU01,T0001,0.50,0.36,0.01,0.12
CPU_ALL,T0001,27.4,59.3,1.3,11.9,,1 
PCPU_ALL,T0001,0.50,0.36,0.0,0.12,1.00
MEM,T0001,15.6,97.8,1433.9,500.9,9216.0,512.0
...
TOP,1376396,T0119,1.75,0.00,1.75,17,584,8,584,0,0.025,0,syncd,Unclassified
TOP,6488224,T0119,0.59,0.40,0.20,1,5832,756,5248,0,0.254,0,named8,Unclassified
ZZZZ,T0120,06:58:59,05-NOV-2012
CPU01,T0120,1.6,1.1,0.1,97.2
PCPU01,T0120,0.02,0.01,0.00,0.97
...
DISKAVGRIO,T0120,0.0,0.0,0.0
DISKAVGWIO,T0120,0.0,0.0,0.0
TOP,6488224,T0120,0.61,0.41,0.20,1,5832,756,5248,0,0.254,0,named8,Unclassified
BBBP,711,ending df -m
BBBP,712,ending df -m,"Filesystem    MB blocks      Free %Used    Iused %Iused Mounted on"
BBBP,713,ending df -m,"ddev/hd4         384.00    227.36   41%     4695     9% /"
BBBP,714,ending df -m,"ddev/hd2        2304.00    151.01   94%    42846    53% /usr"
...
BBBP,841,ending lparstat -H,"add_logicl_lan_buf        0            0.0           0.0          0           0"
BBBP,842,ending lparstat -H,"n-------------------------------------------------------------------------------"
BBBP,843,ending vmtune -a
BBBP,844,ending emgr -l
BBBP,845,ending uptime
BBBP,846,ending uptime,"  06:59AM   up 17 days,  10:49,  2 users,  load average: 2.49, 2.66, 2.66"

You can probably delete all the BBBP lines, and find your halfway mark and delete the second half. No idea how the analyzer will work/not work with the first half missing.

So, from my example of 120 samples, find the start of T0061, and delete...

...
DISKWIO,T0060,0.2,0.0,0.0
DISKAVGRIO,T0060,0.0,0.0,0.0
DISKAVGWIO,T0060,0.0,0.0,0.0
TOP,6488224,T0060,0.73,0.47,0.25,1,5832,756,5248,0,0.254,0,named8,Unclassified
ZZZZ,T0061,06:29:29,05-NOV-2012
CPU01,T0061,2.0,3.2,0.2,94.6
PCPU01,T0061,0.02,0.03,0.00,0.95
...

becomes

...
DISKWIO,T0060,0.2,0.0,0.0
DISKAVGRIO,T0060,0.0,0.0,0.0
DISKAVGWIO,T0060,0.0,0.0,0.0
TOP,6488224,T0060,0.73,0.47,0.25,1,5832,756,5248,0,0.254,0,named8,Unclassified

Hope this helps

Thanks MichaelFelt..

Ill surely try it out now.. Thanks for your replay.

The "replay" is all yours.

FYI: I recommend
nmon -c 240 -s 360 ... whatever else you are doing ...

for 24 hour reports - so that the consolidator and trend analyzer can also be used easily. 12 hour reports (day/night collections, e.g. start at 0600 and 1800, or 0700 and 1900) I use nmon -c 240 -s 180 ...

Have Fun!

The key here, to keep from having issues with excel and "larger than necesssary files" is to keep the count variable -c at 240 or less

Hi MichaelFelt,

 I tried deleting the BBBP lines but still getin the same issue. 

This is what we are using .

/usr/bin/topas_nmon -s 300 -c 288 -f -youtput_dir=/var/tmp/

Thanks & Regards,
B. Jayadeava

The -s 300 -c 288 should work okay. I still recommend changing to 6 minute intervals and 240 entries. Works much better with nmon_analyzer and consolidated/trend views (excel and/or analyzer does not like more than �250 entries and averages them to get number of entries below 250 - so 288 becomes 144 "values".

When using the "bare analyzer" you can have more.

OK. What version of AIX (with nmon included) - or are you still using a downloaded version - and are they different between the archived versions (that you had compressed) and the ones that are working - or are none working?

It will be 24 hours, but I'll try deleting the BBBP lines and see what happens with my reports.

Never got around to testing that - but I did remember a different tool I sometimes use - when stuck: pGraph. Also reads the binary format files, lslparutil from HMC, etc..

Take a look: developerWorks: Wikis - Systems - Other Performance Tools
pgraph specifically: developerWorks: Wikis - Systems - Other Performance Tools - pGraph

I guess this will solve the issue. Ill give it a try for pgraph.. Thanks a lot Michael for your valuable inputs. Ill get back to you checking this.