Probe is inspecting argo-connectors
component by looking at the state
files that contain simple True/False
designating whether configured connector successfully fetched the data from source or not. It is running on the same host where argo-connectors
component resides and reports back the inspection of service leveraging NRPE Nagios service.
Steps of the probe are:
- contact the POEM's
/api/v2/internal/public_tenants
to find out the list of configured tenants - query local
/etc/argo-connectors/global.conf
to find out where is the root directory with the state files -InputState.SaveDir
- query each individual tenant configuration file to get a list of configured jobs/reports
- check
True/False
content ofInputState.SaveDir/<TENANT_NAME>/{topology-ok, downtimes-ok}
andInputSave.SaveDir/<TENANT_NAME>/<JOB_NAME>/{metricprofile-ok, weights-ok}
state files
Probe reports CRITICAL
status if any of running connectors are failing for last three days in a row and WARNING
if connector fails for previous day.
Default arguments:
$ /usr/libexec/argo/probes/connectors/connectors-probe -h
usage: connectors-probe [-h] -H HOSTNAME
optional arguments:
-h, --help show this help message and exit
-H HOSTNAME SuperPOEM hostname
Example of probe run with different status messages:
./connectors-probe -H poem.devel.argo.grnet.gr
OK - All connectors are working fine
WARNING - Tenant: EGI, Filename: TOPOLOGY not ok for previous day
WARNING - Tenant: EUDAT, Filename: TOPOLOGY not ok for previous day / Tenant: SDC, Filename: DOWNTIMES not ok for previous day