Description of problem: If you start a job in the imagefactory (say, a build job), and then stop the imagefactory, it will leave resources lying around. For instance, in the case of an upload build, it will leave the virtual machine that is doing the install still running. Similar problems happen if you are doing snapshot builds, in that the snapshot instance will be left running. The factory should clean this stuff up when it gets a SIGTERM.
making sure all the bugs are at the right version for future queries
https://www.aeolusproject.org/redmine/issues/2446
We have made an initial cut at this with the following commit: https://github.com/aeolusproject/imagefactory/commit/67659edd86a1b5a02bc0f8550c8a00bf5947a2e1 This should ensure that no EC2 instances are left running if the factory is shut down mid-build. We have created a redmine feature to capture the larger effort to make this progressively more robust: https://www.aeolusproject.org/redmine/issues/2595 This does not, as of yet, terminate running local VMs. Am putting in ON_QA. If you would like this to remain open to also cover the cleanup of local VMs that is fine.
Verified that ec2 instances are shutdown mid-push. Reopening based on no local cleanup (comment 4)
adding to sprint tracker
local builds.. on the host.. builds should get shutdown.. same in ec2.. vsphere and rhevm will stop pushing.. and error out.. on_qa
Local builds for RHEV-M and Vsphere do not get stopped. They continue running even though imagefactory is no longer running. Setting back to NEW.
This appears to have been auto-approved because flags weren't flipped prior to moving to 1.1.0?. This is will be considered for a future release.