Conceivably, a user might see a "FATAL" message written to a log or stdout and assume that the agent has halted, crashed, or otherwise exited.
In the Ruby agent, however, "FATAL" is used to convey that there is a high likelihood of normal behavior being interrupted or corrupted but the agent will still do its best to continue.
For example, invalid YAML syntax found in the newrelic.yml config file will produce "FATAL" messages but the agent will continue to operate and source its configuration from other sources (defaults, environment variables).
On a case-by-case basis, we should either have each "FATAL" message accompanied by an effort to halt the agent or downgrade the message severity to "ERROR".
workato-integration[bot]
commented
2 weeks ago
Conceivably, a user might see a "FATAL" message written to a log or stdout and assume that the agent has halted, crashed, or otherwise exited.
In the Ruby agent, however, "FATAL" is used to convey that there is a high likelihood of normal behavior being interrupted or corrupted but the agent will still do its best to continue.
For example, invalid YAML syntax found in the newrelic.yml config file will produce "FATAL" messages but the agent will continue to operate and source its configuration from other sources (defaults, environment variables).
On a case-by-case basis, we should either have each "FATAL" message accompanied by an effort to halt the agent or downgrade the message severity to "ERROR".