Originally posted by: Pilotage_SLS_RO

Hello all

I have an AIX 7.1 , IBM,8202-E4D

In crontab everyday is executed a script  , and the command itself is :

/usr/bin/nmon -d -fT -s 30 -c 2878

Now what i am interested in are the top processes consuming CPU at a given time , thus I believe i should be looking at these lines :

TOP,+PID,Time,%CPU,%Usr,%Sys,Threads,Size,ResText,ResData,CharIO,%RAM,Paging,Command,WLMclass
TOP,7602428,T1721,2.11,0.14,1.98,9,40184,4168,30940,5816108,0,12,dsmc,Unclassified
TOP,7602428,T1722,1.79,0.12,1.67,9,40184,4168,30940,5076779,0,5,dsmc,Unclassified
TOP,7602428,T1723,1.81,0.12,1.69,9,40184,4168,30940,5502985,0,0,dsmc,Unclassified
TOP,7602428,T1724,3.94,1.82,2.12,9,40184,4168,30940,6180025,0,0,dsmc,Unclassified
TOP,7602428,T1725,1.82,0.28,1.54,9,40184,4168,30940,4750274,0,0,dsmc,Unclassified
TOP,7602428,T1726,4.09,1.57,2.52,9,40184,4168,30940,8158211,0,0,dsmc,Unclassified
TOP,7602428,T1727,2.37,0.15,2.22,9,40184,4168,30940,7327329,0,0,dsmc,Unclassified
TOP,7602428,T1728,0.93,0.07,0.86,9,40184,4168,30940,2527529,0,4,dsmc,Unclassified
TOP,12451856,T1762,0.11,0.10,0.01,47,71124,128,71048,14,0,0,java,Unclassified
TOP,2228306,T1860,0.19,0.19,0.00,35,3712,0,3712,0,0,0,j2pg,Unclassified

So T1726 , T1727 and T1728 should be fine , but what are  "T1762" and "T1860"  , how does this AIX do the counting ?

I have seen the thread about a similar problem  , but that was  AIX 5.x and this is 7.1

Is "T1860" equal to "T1700" , and is "T1762" equal to "T1802" ?? if that is then all is correct  ,and i can interpret the numbers as they are , just tell me if it is so .

Originally posted by: luverofpeanuts

I'm not quite understanding what you're questioning?  

The "T" numbers are time interval labels.. so it doesn't make sense to me why you would ask if "T1860" equals "T1700"  ??  They are different interval records.  

In your list,  T1726, T1727, and T1728 are interval records for the same PID.    

The records for T1762 and T1860 are interval records for different PIDs.   If those PIDs do not have interval records labeled T1726, etc... then those PIDs didn't exist during those intervals, or they didn't meet the CPU % threshold requirements to be recorded in the nmon file. 

That is the way I would interpret your list of TOP records.

Originally posted by: Pilotage_SLS_RO

Hello @luverofpeanuts 6cc324f9-4f1d-4c6e-93d8-96cf5af5a5b1

So from what you are saying T1234 sould mean the seconds spent from the time the PID started , in this case is there a way to know when the PID started ?

Lets take for example PID 12451856 :

TOP,   +PID,             Time,     %CPU,  %Usr,   %Sys,    Threads,  Size,       ResText,    ResData,   CharIO,   %RAM,   Paging,  Command,  WLMclass
TOP,   12451856,   T0006,   0.17,       0.16,       0.01,    46 ,           66900,   128,            66824,        17,            1,            0,              java,             Unclassified

UARG,T0006,12451856, 1,java , 46 , root, system,/usr/java6_64/bin/java -Djava.util.logging.config.file=/usr/local/tomcat6/tomcatm/conf/logging.properties -Djava.util.logging.manager=org.apache.juli.ClassLoaderLogManager -Xms256m -Xmx256m -Djava.endorsed.dirs=/usr/local/tomcat6/tomcatm/endorsed -classpath /usr/local/tomcat6/tomcatm/bin/bootstrap.jar:/usr/local/tomcat6/tomcatm/bin/tomcat-juli.jar -Dcatalina.base=/usr/local/tomcat6/tomcatm -Dcatalina.home=/usr/local/tomcat6/tomcatm -Djava.io.tmpdir=/usr/local/tomcat6/tomcatm org.apache.catalina.startup.Bootstrap start

BBBP,1232,ipcs -a,"m         3 0xffffffff D-rw-------     root   system     root   system      1 268435456    12451856   15925432   23:33:52   7:47:58     23:33:52"

Should "7:47:58" be the starting hour of the PID 12451856 , and shoud "T0006" mean it has spent 6 seconds form start ?

Originally posted by: luverofpeanuts

Hello Pilotage....

>>>Should "7:47:58" be the starting hour of the PID 12451856 , and shoud "T0006" mean it has spent 6 seconds form start ?

No.  The T0006 is indicating that the line of data is for the 6th nmon collection interval since the nmon collection began.  It does not indicated a specific time.  

Here is how nmon is started, from you original post. 

     /usr/bin/nmon -d -fT -s 30 -c 2878

You have instructed nmon to collect data every 30 seconds   2878 times.    That means, in your nmon file, there should be "T----" values ranging from T0002 to T2878.    Why no T0001?  Because T0001 is the start...   so the first data recorded is saved at T0002, after the first 30 seconds.    

Have you looked at the 'ZZZZ' records in your nmon file?    Do this....

      cat <insert your nmon file name here> | awk '/^ZZZZ/||/^UARG/||/^TOP/  {print}

Seeing the ZZZZ, UARG, and TOP records together in the output might help make it more clear how the "T" value is used.    Using more advanced awk commands you can create quite a useful and readable output from the raw nmon file.

I've recently been doing the same post processing of the nmon output file for the same reason you are... to see the processes active for each time interval and how much CPU the were using...and to see the more complete command arguments as well.  

I hope that helps