Description of problem: Given a scaling app with a disable_auto_scaling marker added, the app will automatically scale down after scaling up via REST API. Currently, the following cartridges are covered: python-2.7 python-3.3 will append the list in comments if find others. Version-Release number of selected component (if applicable): devenv_3880 How reproducible: always Steps to Reproduce: 1. create a scaling app 2. create a disable_auto_scaling marker 3. push the marker to its remote repo 4. scale up the app via REST API 5. check the scaled-up gear and the haproxy-status page 6. wait for sometime (in my test, it was around 10 mins) 7. repeat step 5 8. check the scale_events.log 9. check if the disable_auto_scaling marker is at presence Actual results: In Step 5, the app was scaled up successfully, and the scaled-up gear was accessible. However, in Step 7, only the local gear was shown in the haproxy-status page, and it returned error 503 when accessing the scaled-up gear. Moreover, in the scale_events.log, the sscaled-up gear was scaled-down: I, [2013-10-10T04:36:54.139800 #14505] INFO -- : GEAR_DOWN - capacity: 0.0% gear_count: 2 sessions: 0 remove_thresh: 40.5% but the disable_auto_scaling marker was in the app's repo Expected results: The scaled app should not be scaled down with the disable_auto_scaling marker at presence. Additional info:
We now require that haproxy be restarted to pick up the disable_auto_scaling marker since haproxy isn't restarted automatically on a git push anymore.
After restarting the ha cart, the auto_scaling was indeed disabled. Move this bug to VERIFIED.