Container logs terminal lack of output can be confusing

[abuehrle]

I tried to get the logs on several containers (kube-api server, rabbitMQ and others), but when I try to open the terminal, it doesn't display anything.

Note that the logs do show up on Pods, but not in containers. I see that those terminal buttons both do different things: one has a tooltip for logs (in pods) and the other 'attaches' (in containers). What am I supposed to see in the attached terminal - should I see the Docker logs in there?

Here's a recording:

http://recordit.co/oRRdey5SMV

[abuehrle]

I guess what's going on here is that there is no load on the app, and so there are no messages to display. It might be a good idea to include a message like: 'There are no logs' or something to that effect, so that it shows it's doing something.

[2opremio]

It doesn't show the existing logs but the upcoming logs, but I guess we could display a message until there is done output.

(CC @davkal @bowenli )

[bowenli]

Something like "Attached, waiting for output..." would be fine. Note that a spinner would not be appropriate in this case since the connection is already established.

[rade]

but why is there a "Connection Closed" popping up? I thought that is what you were complaining about, @abuehrle?

[abuehrle]

I think there are a couple of things going on here:

• lack of feedback for the user when there are no Docker logs
• an intermittent connected/not connected message when you attempt to view container logs - it's hard to tell if we are connected or not

There was also a warning and no logs on viewing the Pod logs of Weave Net, but I will log that one for you separately as it's unrelated: a container name must be specified for pod...

[lukemarsden]

It would be great if this feature could show the latest 40 log lines or so, and/or be reworked to show "Logs" versus "Attach". "Attach" is less helpful than "Exec" now that "Exec" is a thing. cc @rade

[rade]

see also #2234.

[rade]

Note that Scope's 'Get logs' button on the pod detail panel does actually fetch the proper pod log via the k8s API, including history. So this is only an issue for container logs.