It seems that it broke all kubemark suites (which are blocking now).

The problem why we didn't catch this in presubmit is that in presubmit we are using debian not GCI. The change is already merged: kubernetes/test-infra#945

Jenkins unit/integration failed for commit 765f7b9. Full PR test history.

The magic incantation to run this job again is @k8s-bot unit test this. Please help us cut down flakes by linking to an open flake issue when you hit one in your PR.

@k8s-bot kubemark test this issue: #IGNORE
Check if it works.

Merging to unblock the build.

How is kubemark different from other e2es? This PR has passed node and cluster e2es.

That's a very good question - I don't know. In kubemark we just create ordinary Pods and Secrets for HollowNodes (this is the only part that can be affected). It's really just a very simple use case of our basic objects.

When I tried kubemark, I see CPU availability to be an issue. Could that be
the root cause? Can we increase the node sizes?

On Fri, Oct 28, 2016 at 2:39 PM, Marek Grabowski [email protected]
wrote:

That's a very good question - I don't know. In kubemark we just create
ordinary Pods and Secrets for HollowNodes (this is the only part that can
be affected). It's really just a very simple use case of our basic objects.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#35763 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AGvIKG6E_npducLYn7HjACWy6HogjlmNks5q4muIgaJpZM4KjIIe
.

If it reduced the amount of available CPU/mem on the machine it can certainly cause kubemark failure - but how exactly this PR causes that (assuming that I understood correctly what you wrote).

When I tried the kubemark tests, many of the pods were unable to run due to
lack of CPU. I suspect kubemark tests are packing the nodes with a lot of
pods and thereby causing CPU starvation for the entire node. I was unable
to login to the nodes once the tests started running.

On Mon, Oct 31, 2016 at 2:51 AM, Marek Grabowski [email protected]
wrote:

If it reduced the amount of available CPU/mem on the machine it can
certainly cause kubemark failure - but how exactly this PR causes that
(assuming that I understood correctly what you wrote).

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#35763 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AGvIKOvf-TWtXc8gwFLHA0Nwv8pLMCV2ks5q5bofgaJpZM4KjIIe
.

OK, but how this PR affects that?