github jhuckaby/Cronicle v0.7.0
Version 0.7.0

latest releases: v0.9.45, v0.9.44, v0.9.43...
6 years ago

Quite a few changes and fixes in this release:

  • New standard Plugin for making HTTP requests.

    • Can send HTTP GET, HEAD, or POST requests with custom headers.
    • Tracks performance metrics, can also match content for success/error.
  • Much improved handling of detached jobs on Cronicle restart / upgrade.

    • Detached jobs will now seamlessly recover after any server restart (as long as the server wasn't down for dead_job_timeout+).
  • Shell Plugin Improvements

    • New optional log timestamp annotation for Shell Plugin.
    • Improvements to logging and error handling.
    • STDERR is now captured separately, and displayed in its own custom HTML section on the job details page (up to first 32K). Everything is still logged of course.
    • First line of STDERR is now added to final description, in the event of an error, if less than 256 chars.
  • Fixed elapsed time for detached jobs.

    • The elapsed time was being calculated from when the master daemon finished the job, but this may be up to 60 seconds AFTER the job itself completed, for detached jobs, because they send status updates by way of queue files. This corrects that by allowing the child to set the time_end property in the job JSON.
    • Note that this DOES introduce a possible issue with clock sync between servers in the same cluster.
  • Fixed bugs related to missing environment variables on server reboot and init.d.

    • Now always calling Tools.getpwnam() for user info on job launch.
    • This will correct issues like the shell environment missing things like HOME when init.d launches the Cronicle daemon after a server reboot.
  • Bug fix: Increment stats.jobs_failed when a job fails to launch (server unavailable, etc.).

  • Disabled Uglify on client-side JavaScript for now.

    • Makes things easier to debug on production, and the content is gzipped anyway.
  • Increased default dead_job_timeout to 120 seconds.

    • Upgrading servers may take longer than 30 seconds, and don't want detached jobs to be declared dead.
  • Locked some top-level npm package versions, just for a bit of safety (npm-shrinkwrap would be better long term).

Don't miss a new Cronicle release

NewReleases is sending notifications on new releases.