Re: bind2 pmda qa problem

  That is interesting. The hashing root of cause is a good guess but as far
as I remember this is well handled in the code (I will check it when I get
to the computer). More likely it seems to me that the autodetection of
metrics and instances I am reworking now fired again and caused the
instances to get renumbered. Were there any restarts od Pmda or pmcd or a
period of unavailability of the Bind server?

Dne 15. 2. 2017 10:32 napsal uživatel "Ken McDonell" <[email protected]
>:

On 15/02/17 18:49, Lukáš Oliva wrote:

>   Hello,
> Thanks for detection. Canbyou describe the scenario?
>

Lukáš,

This is qa/1087.  The platform is bozo (PCP 3.11.8 x86_64 Ubuntu 16.04).

The metric bind2.zones.serial with the bind-rhel6-stats.xml data file where
this appears to be the relevant part of the data file:

            <zone>
              <name>mydomain.example.com/IN</name>
              <rdataclass>IN</rdataclass>
              <serial>3091</serial>
            </zone>
            <zone>
              <name>mydomain2.examples.com/IN</name>
              <rdataclass>IN</rdataclass>
              <serial>2</serial>
            </zone>
            <zone>
              <name>anotherdomain.example.com/IN</name>
              <rdataclass>IN</rdataclass>
              <serial>27585</serial>
            </zone>

and 1087.out expects something like

bind2.zones.serial PMID: 25.0.133 []
    Data Type: 32-bit unsigned int  InDom: 25.1 0x6400001
    Semantics: counter  Units: count
Help: <empty entry>
    inst [0 or "mydomain.example.com"] value 27585
    inst [1 or "mydomain2.examples.com"] value 3091
    inst [2 or "anotherdomain.example.com"] value 2

which looks wrong (compare the numbers and the instance names)

and I'm seeing

bind2.zones.serial PMID: 25.0.133 []
    Data Type: 32-bit unsigned int  InDom: 25.1 0x6400001
    Semantics: counter  Units: count
Help: <empty entry>
    inst [0 or "mydomain.example.com"] value 27585
    inst [1 or "anotherdomain.example.com"] value 3091
    inst [2 or "mydomain2.examples.com"] value 2

which is also wrong!

So this looks like a code botch in the PMDA that is showing up differently
due to different hashing methods on the platforms.