As an operator I can observe the lifecycle of events as they are processed by the backend.
Current Behavior
The logs generated by the backend expose some but not all of the activities/phases in the event lifecycle.
Possible Solution
Implement a trace command in sensuctl which allows operators to specify one or more patterns for matching based on entity names, subscriptions and/or check names.
When the backend processes events matching one or of these patterns, the event is reported to the operator in such a way that the event ingress from agent, mutation, filtering and handling can all be understood without having to resort to debugging via log messages.
Context
Customers
1.
2.
3.
4.
Your Environment
Sensu version used (sensuctl, sensu-backend, and/or sensu-agent):
Rough thoughts/reaction: Fix/improve logging around many of the event actions. Build upon a logger abstraction to allow sensuctl/API to observe log events matching certain criteria.
Add mechanism to create a unique ID [UUID] for a given event. Attach backend hostname as a label.
Backend and Agents Events' API, it creates a unique ID on event submission.
Log event field name: event_uuid
Going through the whole system and updating any sort of log events to reference that event ID for context (this will allow us to 'thread the needle' for a particular event)
**** This will be ONLY for backend and agent logs. To trace, we will need to know which backend or agent.
@cwjohnston commented on Mon Feb 25 2019
Expected Behavior
As an operator I can observe the lifecycle of events as they are processed by the backend.
Current Behavior
The logs generated by the backend expose some but not all of the activities/phases in the event lifecycle.
Possible Solution
Implement a
trace
command in sensuctl which allows operators to specify one or more patterns for matching based on entity names, subscriptions and/or check names.When the backend processes events matching one or of these patterns, the event is reported to the operator in such a way that the event ingress from agent, mutation, filtering and handling can all be understood without having to resort to debugging via log messages.
Context
Customers
1. 2. 3. 4.
Your Environment
@portertech commented on Wed Feb 27 2019
Rough thoughts/reaction: Fix/improve logging around many of the event actions. Build upon a logger abstraction to allow sensuctl/API to observe log events matching certain criteria.
@portertech commented on Wed Jul 03 2019
We need to keep it simple at first, adding the concept of an event id for tracing through the system would be hugely beneficial.
@annaplotkin let's line this one up 👍
@annaplotkin commented on Wed Jul 03 2019
@portertech 👍
@annaplotkin commented on Tue Jul 09 2019
Proposal:
**** This will be ONLY for backend and agent logs. To trace, we will need to know which backend or agent.
@palourde commented on Tue Jul 09 2019
Issue moved to sensu/sensu-go #3125 via ZenHub