Dustin Decker

I had started implementing this a while back as a goroutine that periodically lists and deletes pods if the given annotation is older than a given duration: https://github.com/cruise-automation/k-rail/compare/exec-taint-label#diff-a6ad059d9beaf8a3c7bdcd2b41ab23b5R18 It's largely...

Oh, I just noticed that the WIP is using Delete rather than the Eviction API: https://kubernetes.io/docs/tasks/administer-cluster/safely-drain-node/#the-eviction-api We should use the eviction API because it respects PodDisruptionBudgets and some other things.

One workaround we've seen is to put the contents of the env file in the value field in one secret. You can then drop that secret's value as a file...

Interesting, thanks for the info and the quick response. What would it take to replace that with a pure go implementation? There have been a number of git RCE vulns...

Hey @calvin-f, why did you close this? I also tried to solve this race with https://github.com/go-git/go-git/pull/175 Did you find something out that made it unnecessary?

I see. Could you re-open this then? I'm trying to get it fixed upstream and prefer this fix over mine.

Thanks George, that was quick! I'll give it a shot soon and comment on the PR.

I never got around to testing it and am no longer using Kafka in my current position. Care to give https://github.com/uber-go/kafka-client/pull/55 a try, @Gillani0 ?

Can you do a `docker service ps rl90o2zc17b14r294p0a4dr7v --no-trunc` ? Or if that no longer exists, do it for whatever service is named during when this message appears again: ```...

@EliRibble, glad that helped, but I agree that there is something more wrong. I'm seeing some similar messages in one of my 17.05 clusters and I'm stuck as well. I'll...