QryRTT resolvercounter are not calculated anymore with latest bind version

[ybrock]

Hi,

I was using this template and I loved the Query Round Trip Time graph.
Since the last update of Bind (on EL7 here), the QryRTT* counters have all 0 in the JSON file /tmp/bindstats.json, but the counters are Ok in the XML file

In the JSON File :
"QryRTT10": "0",
"QryRTT100": "0",
"QryRTT500": "0",
"QryRTT800": "0",
"QryRTT1600": "0",
"QryRTT1600+": "0",

In the HTML/XML file 👍 110065252
30438688
3003133
45595
9004
21

The version of Bind has been updated from 9.9.4 to 9.11.4 (bind-9.11.4-9.P2.el7.x86_64 actually).

There is probably a small bug in bind-stats.py but I can't find it !

Cheers
Yvan

[ybrock]

  <counter name="QryRTT10">110065252</counter>
  <counter name="QryRTT100">30438688</counter>
  <counter name="QryRTT500">3003133</counter>
  <counter name="QryRTT800">45595</counter>
  <counter name="QryRTT1600">9004</counter>
  <counter name="QryRTT1600+">21</counter>

Sorry, the XML was interpreted by github. This is what is found in the HTML/XML file

[ybrock]

Between Bind 9.9 and Bind 9.11 the statistics version have changed :
Version 2.2 for Bind 9-9
Version 3.8 for Bind 9.11

The processing of Version 3 statistics is quite different in the python script.
I have narrowed down the issue to a view problem :
resqtype (outcounter), resstats (resolvercounter) and cachestats (cache) statistics are all wrong, because they are calculed from the view named "_bind", instead of beeing computed from the view "_default", this is why the values appears with a wrong value, often "0".

I'm not really good with Python XML parsing, I'm still trying to fix, from line 130, to be sure we're in the _default view, like it was done in v2 stats.

[zceigler-n]

@ybrock any luck with this? Currently running into the same issue with views