[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: livenessProbe

On Wed, Oct 14, 2015 at 11:05 AM, Clayton Coleman <ccoleman redhat com> wrote:
Does your pod support keep alive?  If so, this may be a bug in the
health checker (it's trying to reuse a previously open TCP connection,
which has been closed silently).

The pod is exposing unicorn directly (a http server written in ruby), and it doesn't support keep alive connections indeed.
So that's explain a lot:
- the pod is marked as running, because the health check is ok
- the pod (/containers) is killed because the connection is closed after the 30s timeout.

It's very unclear by the way if the healtcheck is only during startup, or also during normal runtime. I can't find any reference to a regular check in the doc, but I understand in your answer that it's the case. Is the pod checked every "initialDelaySeconds" (why it's named "initial" then)?

Thanks again for your insights Clayton.

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]