Skip to content

Observe and export Choria Network metrics to Prometheus

License

Notifications You must be signed in to change notification settings

choria-io/tally

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

25 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Choria Fleet Tally

In large dynamic fleets it's hard to keep track of counts and versions of nodes. This is a tool that can observe a running network and gather versions of a specific component. The results are exposed as Prometheus metrics.

$ tally --component '*' --port 8080 --prefix choria_tally

For this to work it uses the normal Choria client configuration to connect to the right middleware using TLS and listen there. The certificates needed does not need to match the .+.mcollective pattern, meaning tally will never make RPC requests to your fleet - it passively listens for advisories.

This will listen on port 8080 for /metrics, it will observe events from the server component and expose metrics as below:

Metric Description
choria_tally_good_events Events processed successfully
choria_tally_process_errors The number of events received that failed to process
choria_tally_event_types The number of events received by type
choria_tally_versions Gauge indicating the number of running components by version
choria_tally_processing_time The time taken to process events
choria_tally_nodes_expired The number of nodes removed during maintenance runs

Additionally, this tool can also watch Choria Autonomous Agent events, today it supports transition events and exec watchers:

Metric Description
choria_tally_machine_transition Information about transition events handled by Choria Autonomous Agents
choria_tally_exec_watcher_success Machine exec watcher success runs
choria_tally_exec_watcher_failures Number of exec watcher executions that failed
choria_tally_exec_watcher_runtime Exec watcher runtime

If you have any Choria Governors the tool can listen to the events these emit and report on those.

Metric Description
choria_tally_governor Events and their types seen per Governor and per Component

Clustered Deployment

Clustered deployments are problematic since from the perspective of Prometheus the stats will be multiplied by the number of running instances.

When ran with the --election TALLY flag it will activate leader election against Choria Streams.

Regardless all metrics will have a active label set which will be 1 on the active node (or only node when leader election is not configured).