Taskscaler removes instances too early
This problem is observed by @sge70 while using "pwsh" with the "fleeting-plugin-aws": https://gitlab.com/gitlab-com/account-management/emea/mbition/scalable-runners-implementation/-/issues/3#note_1344173084
We see that the runner does delete instances which were not idle for the set idle time. But what makes it for an issue for us. The runner decides in the the middle of a running job to delete the instance
Update:
This problem was a result of removing instances too early based on their idle time and is fixed with gitlab-org/fleeting/taskscaler!25 (merged).
Edited by Arran Walker