Network Interface

In the process of implementing a 1/5/15 CPU Load monitor for Nagios, I was confronted with an LPAR problem. Using the uptime command, the load values are for all of the CPUs combined – which makes it impossible to use a standard value for different partitions, because they’ll have a different number of CPUs assigned. The solution is to divide the load values by the number of CPUs – but how do you find ou the number of CPUs? IF you are using SMT, you have to use either smtctl, vmstat, or something similar. Only the root user has execute access to the smtctl command, so that’s not very useful. Use the following grep command: vmstat | grep lcpu | cut -d” ” -f3 | cut -d”=” -f2 .
That simple .

Having a customer running AIX 6.1, one of the things I regularly have to deal with is the failure of most open source packages to identify the OS out of the box. Mostly because most sane people don’t bothering compiling open-source on AIX I imagine, but it’s much easier now that it used to be. In any case, just edit the configure file, and replace any instances if ‘aix5*’ with ‘aix6*’, and then check config.guess for an AIX check – usually if you search for the string ‘AIX’ you’ll see a check immediately after for the version number – wher you see ‘[45]‘ just replace it with ‘[456]‘. You’ll run into the same problem in the aclocal.m4 file if that’s the method being used.