Skip to content

Latest commit

 

History

History
55 lines (38 loc) · 3.18 KB

README.md

File metadata and controls

55 lines (38 loc) · 3.18 KB

Healthcheck

This is a sample implementation of various healthcheck modules.

At the beginning, most attention has been given to implement the backend, and there's a very simplistic frontend

This project currently consists of:

  • Commerce Healthcheck adapters
    • An adapter from commerce's CommerceHealthHttpStatus to the Healthcheck interface
  • Demo System relaxed security healthchecks
    • This refactors Olaf's demo-checklist to the Healthcheck interface, implementing relaxed security checks that are suitable for demo systems.
  • Breaking Changes implementations
    • e.g. with some copies from Liferay's portal-impl code (e.g. VerifyProperties, which currently runs only during upgrades)
    • e.g. with a simulated "Have you implemented a deprecated service that's no longer supported?" alert
  • Best Practice implementations
    • e.g. "Do not use default user accounts & passwords"
  • Operational implementations
    • e.g. checks for available memory and redirection configuration

How to build

Clone this repository into a Liferay Workspace's modules/healthcheck directory.

Created with liferay.workspace.product=dxp-7.4-u18, last run with U102.

Reference

Limitations

  • Very basic permission checking (Commerce Health Checks implement this in a better way, for example in CommerceHealthStatusDisplayContext). Health Check UI is only available for Company Administrators.
  • UI is veeeeery barebones right now
  • Only few native health checks implemented: Contribute more, in code or just ideas
  • Assumes it only runs on System level (e.g. first virtual instance) and UI is not available on secondary instances. Even though implemented checks might be instance specific...
  • Speaking of instances: Not much thought has been given to scenarios with multiple instances that might have different - instance specific - configuration.
  • Message locale is determined at time of executing the healthcheck, not at display time. Fine for the kind of quick healthchecks that are currently implemented and can be executed immediately, but would be impractical if there were long running or a very large number of checks.

Ideas for the UI

The UI is - explicitly - very ugly. To make its ugliness even more explicit, it's minimally interactive and 100% built with <table>.

UI Screenshot

You can ignore some healthchecks (successful or unsuccessful). They'll still be executed, but the result will not be shown.

Collecting a few ideas that could go into UI features:

  • Get rid of the based layout to begin with
  • Run healthchecks in background (scheduled) - this enables long running processes to run as well (see remark about locale though)
  • Filter/Sort results by category/result
  • Activate/Deactivate certain categories (e.g. ignore demo-related healthchecks with released security) or pick individual checks to run and blacklist completely.
  • Selectively un-ignore filters (e.g. show ignored ones without unignoring them)
  • Selectively run individual filters or categories (especially when long running checks are implemented, that are meant to run in the background)