Logging and monitoring

Why to log/monitor?

Because it helps in:

• knowing what's happening in the applications.
• getting clear details while debugging
• error reporting,
• getting warning conditions when users have encountered something unexpected
• tracing user actions through a common trace ID

Logging

• Logs are the messages emitted by applications for certain events.
• Real-time logging allows watching logs stream in from users on the platform.
• Searchable logs:
  • Using Tags
    • These are special markers assigned to each log as it come in.
    • e.g. environment:production, severity:error, service:accounts.
    • Helps to keep track of what type of log it is and where it came from.
  • Using Attributes
    • These are the details within each log message.
    • e.g. @event.outcome:failure , @http.useragent.device:desktop , or @trace_id:2fb45a.
• Pre-processing logs
  • It can be used to get some metadata from logs into actionable tags/attributes.
  • Sends logs through a pipeline where the logs can be formatted and tagged as required.

Monitoring

• Alerts
  • In regular interval of times (say in every 1 hour)
  • Only when it is requried (like when there are more than 10 messages of this type)
  • Send alerts to Slack/Mail.
• Keeping track of metrics within the application.
  • e.g. logs(status:error)
• Real User Monitoring (RUM)
  • Track clicks & response times
• Application Performance Management (APM):
  • tells how long your requests take.
  • shows waterfalls where the processing is slowest.

Tools

• Datalog