Xander Neben  8 hours ago

https://jira.mongodb.org/issues/?jql=project%20%3D%20BF%20AND%20summary%20~%20powercycle%20AND%20issueFunction%20in%20linkedIssuesOf(%22project%20%3D%20SERVER%20AND%20(component%20IS%20EMPTY%20OR%20component%20!%3D%20'Testing%20Infrastructure')%22)%20and%20assignee%20not%20in%20(%22mikhail.shchatko%40mongodb.com%22%2C%20%22robert.guo%40mongodb.com%22)%20and%20created%20%3E%20%222020%2F01%2F01%22%20ORDER%20BY%20key%20DESCHere are the powercycle failures in the last 2 years. I had Juan label these as “infra-failure”, “not-unique-to-powercycle”, and “unique-to-powercycle”. I think there are a few legit failures in here but would like you to do a quick passover. I’m hoping you will think that the one real failure is not worth maintaining powercycle. Let me know what you think.
----New
 
Max Hirschhorn  21 minutes ago

I think https://jira.mongodb.org/browse/SERVER-52680 is a bug we should have been able to catch outside of what powercycle is doing. All of what powercycle did which was interesting was choose to start mongod on a different port and running replSetReconfig
 
Max Hirschhorn  20 minutes ago

None of the ones listed here make me feel like powercycle is finding things and so I think if we're having problems related to the infrastructure then starting by removing Windows support feels non-contentious
 
Max Hirschhorn  19 minutes ago

If removing altogether on Linux is part of the question then I feel like it's a question of "insurance policy we don't miss something" than there being a chance we write some durability bug which wouldn't surface any other way