Monitoring

[Antoine Pietri]

• Every service should respond on /__ping
• The services should export some graphed values and possibly alert the orgas
• Weathermap of the network?
• Machine monitoring (https://github.com/prometheus/node_exporter), web interface (https://github.com/prometheus/promdash) and aggregation of data (https://github.com/prometheus/prometheus).

[Rémi Audebert]

For django services: https://github.com/korfuri/django-prometheus

[Rémi Audebert]

Work started at:

• https://bitbucket.org/halfr/sadm/commits/branch/monitoring
• https://bitbucket.org/halfr/prometheus-journald
• https://aur.archlinux.org/packages/prometheus-node-exporter/

[Rémi Audebert]

Work done:

• Services export metrics to be collected by prometheus.

python-lib/prologin/concours/stechec/monitoring.py
python-lib/prologin/hfs/monitoring.py
python-lib/prologin/masternode/monitoring.py
python-lib/prologin/presencesync/monitoring.py
python-lib/prologin/rpc/monitoring.py
python-lib/prologin/workernode/monitoring.py

• The target url are documented in etc/prometheus/prometheus.conf.
• Documented in docs/source/monitoring.rst

Work remaining:

• Update docs/source/monitoring.rst to use grafana instead of promdash. promdash is deprecated.
• Document alertmanager
• Add alerts