[Check_mk (english)] predictive monitoring in check_mk 1.5.0p6cre

[Rob Walker]

I am trying to get predictive monitoring working on a test site before
I roll it into our main systems.

I am using 1.5.0p6cre.

I have turned on predictive monitoring for a number of hosts, for a
number of services. I ran the following commands as my omd user:


cd ~/var/check_mk/prediction
~/var/check_mk/prediction$ find . | wc -l
982
~/var/check_mk/prediction$ find . -type f | wc -l



My research took me to the file

/opt/omd/versions/1.5.0p6.cre/lib/python/cmk_base/prediction.py , line 52.

Using some creative MKGeneralException calls, I figured out that the
lql there is something close to this:

GET services
Columns: rrddata:m1:load15.max:1496941200:1496944799:1
OutputFormat: python
Filter: host_name = tools00
Filter: description = CPU load


When I run that query, I get a bunch of nothing. I think that this is
what is causing me to have no prediction files, therefore no
prediction values.

$ netcat localhost 6557 < query00
[[None]]

I changed line 68 to have a special string, and that shows up in my
web UI. Therefore I think that cmk is also not getting a response.




A second question I have is about line 248 in that file. There are
calls to logger.verbose, but I cannot find the file that these are
going in to. That would make my debugging a lot easier if I could
call the logger() function instead of messing around inside of the web
UI.

Thanks,
Rob