For more information regarding the security incident at F5, the actions we are taking to address it, and our ongoing efforts to protect our customers, click here.

Forum Discussion

Robert_Sutcliff's avatar
Robert_Sutcliff
Icon for Nimbostratus rankNimbostratus
Mar 23, 2010

TMM Utilization graph

Hi,

 

What stats are used to generate the TMM Utilization graph (under the Performance section of the Web Interface)? Specifically, how are figures for each TMM process calculated?

 

 

I had thought it would be based on the following SNMP OIDs (by walking the sysTMMStatTable).

 

sysTMMStatTmTotalCycles

 

sysTMMStatTmIdleCycles

 

sysTMMStatTmSleepCycles

 

 

The nearest I can get is to use the following equation :-

 

( (totalCycles - (idleCycles + sleepCycles) ) / totalCycles) * 100 * 5

 

although where the multiplication by 5 comes from I can't explain.

 

 

The reason for this is I want to monitor this via SNMP GET (for use in nagios). We have a number of virtual services that are CMP bound (ie. stick on one tmm process) - so any single tmm process running into high utilization can cause a problem (and the average could still be quite low).

 

 

Any ideas/suggestions?

 

 

Cheers,

 

Rob
management
monitoring

6 Replies

  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Mar 23, 2010
    The answer is going to vary, depending on which version of LTM you're using. The latest versions have better information, but v10 has a different OID than 9.4 IIRC. (Or it might have been 9.4 that introduced the multi-cpu OID's).

     

     

    Anyway. Multi-cpu stats are available under

     

     

    .1.3.6.1.4.1.3375.2.1.7.5.2.1.1.1

     

     

    the SNMP mib is available on your F5 itself, so I won't include it here.

     

     

    H
  • Steven_John_711's avatar
    Steven_John_711
    Icon for Nimbostratus rankNimbostratus
    May 19, 2010
    Hi,

     

     

    I wish you hadn't stopped there to be honest. We've just gone from 6800s 9.4.5 to 6900s 9.4.8 and now want to monitor individual TMM utilisation like we used to do on the 6800 with this Truncate( 100-({sysStatTmIdleCycles}/{sysStatTmTotalCycles}*100), 0). It'd be great if you could point me at the same oids for this multi processor platform if you can please?
  • JRahm's avatar
    JRahm
    Icon for Admin rankAdmin
    May 20, 2010
    my $tmmTotalCycles = '.1.3.6.1.4.1.3375.2.1.1.2.1.41.0';

     

    my $tmmIdleCycles = '.1.3.6.1.4.1.3375.2.1.1.2.1.42.0';

     

    my $tmmSleepCycles = '.1.3.6.1.4.1.3375.2.1.1.2.1.43.0';

     

  • tony_100112's avatar
    tony_100112
    Icon for Nimbostratus rankNimbostratus
    Jul 19, 2010
    I generate the TMM Utilization graph with the equation of ( (totalCycles - (idleCycles + sleepCycles) ) / totalCycles) * 100 * 5 via ZABBIX

     

    But sometimes the value was negative number or greatly exceeded the true value

     

    What can I do for this situation？
  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Jul 23, 2010
    How are you getting the stats? If via iControl, how are you converting the hi/lo 32 bit representations to 64 bits? (Some compilations of perl require different maths to doit due to the sign extension of unsigned numbers).

     

  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Jul 23, 2010
    Actually, that should be 'How does ZABBIX read the data'? (Sorry, never used it).

     

     

    H