Tail of bosh deployments in 5.6 slow

Has anybody else seen in the 5.6 update now we miss the last line in the bosh deployment tail.

Basically it looks like it’s not showing it until it moves on. This is very annoying when doing something like a os-reload on vms where the VM is not completed but the bosh deployment has started work on a VM but not completed it.

Has anybody opened an issue for this. Seems a new problem in 5.6?

But following the deploy in bosh I see this …

I’ve noticed the same thing happening in Ops Manager so it might not be a Concourse specific thing although they might use the same client.

I don’t think this is related. As we are NOT using the bosh specific resources. We are running bosh in a task and that should be showing the output. Why the output is not getting to the web is the issue. Any feedback from the concourse team. This used to work fine before the latest concourse update.

---
jobs:
  - name: job
    public: true
    plan:
      - task: simple-task
        config:
          platform: linux
          image_resource:
            type: registry-image
            source: { repository: busybox }
          run:
            path: /bin/sh
            args:
            - -c
            - |
              #!/bin/sh
              printf "hello..." && sleep 10 && printf " good-bye\n"
              sleep 1
              printf "hello..." && sleep 10 && printf " good-bye\n"
              sleep 1
              printf "hello..." && sleep 10 && printf " good-bye\n"
              sleep 1
              printf "done..."
              sleep 1
              printf "\n"
              sleep 30

This simple pipeline shows the problem.

There must have been a change where lines are not displayed until the \n. This causes issues in bosh deployment when following as bosh will print a line without the new line until it has possibly added the finished timestamp. What could have changed in v5.6 to cause this to no longer work?