On 11.03.2016 04:55, Nick Coghlan wrote: > On 10 March 2016 at 19:28, Stef Walter <[email protected]> wrote: >> This is a heads up that docker 1.10 breaks API in various ways. It broke >> kubernetes among other things. > > Huh, would a plausible symptom of that be "kubectl get node" reporting > a Docker 1.10 node as Not Ready, even though "curl -L > http://127.0.0.1:10248/healthz"; on the node itself reports "ok"? > > $ rpm -qa docker kubernetes > kubernetes-1.2.0-0.13.alpha6.gitf0cd09a.fc23.x86_64 > docker-1.10.2-5.git0f5ac89.fc23.x86_64 > > I'd been assuming there was something wrong with my node config, > rather than looking for bugs or incompatibilities.
The kubernetes + docker 1.0 breakage I was referring to is limited to kubectl exec and kubectl logs. https://bugzilla.redhat.com/show_bug.cgi?id=1311861 However this kubernetes package regression seems suspiciously similar to what you're talking about: https://bugzilla.redhat.com/show_bug.cgi?id=1316424 Both solved by later kubernetes RPM builds. Thanks to Jan. By the way the Cockpit integration tests is the thing that has been finding all these issues. So we end up being a clearing house for regressions on the APIs that Cockpit touches :S Cheers, Stef
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Container-tools mailing list [email protected] https://www.redhat.com/mailman/listinfo/container-tools
