current scale. Two important ones are when missing metrics might
change the direction of scaling, and when the recommended scale is
within tolerance of the current scale.
The way that ReplicaCalculator signals it's desire to not change the
current scale is by returning the current scale. However the current
scale is from scale.Status.Replicas and can be larger than
scale.Spec.Replicas (e.g. during Deployment rollout with configured
surge). This causes a positive feedback loop because
scale.Status.Replicas is written back into scale.Spec.Replicas,
further increasing the current scale.
This PR fixes the feedback loop by plumbing the replica count from
spec through horizontal.go and replica_calculator.go so the calculator
can punt with the right value.
Nit: remove capitalization of preferred
Remove line from kubelet and add to separate PR for easier merge
nit: dependency added to separate PR
Add check to ensure strict policy cannot be set without feature gate enabled
Topology Manager runs "none" policy by default.
Added constants for policies and updated documentation.
Adds a go app which runs the e2e tests with ginkgo.
- Supports all the existing env vars of the bash script
- Improved flow control to avoid and better report issues
regarding the process PID
- Adds flags for modifying where to find the test binary and
ginkgo binary so that you can run it locally
- Adds 3 flags for specifying extra args before the double-dash,
extra args after the double-dash, and the seperator to use between
values in those env vars. This allows setting arbitrary, complex
values for use on the command such as flags which include spaces
or other characters.
In coreutils LC_ALL=C is required to make sorting stable across
platoforms and locales.
$ echo "a-c\nabc" | LANG=C LC_ALL=C sort
a-c
abc
$ echo "a-c\nabc" | LANG=C sort
abc
a-c
libcni 0.7.0 caches ADD operation results and allows the runtime to
retrieve these from the cache. In case the user wants a different
cache directory than the defaul, plumb that through like we do
for --cni-bin-dir and --cni-conf-dir.
Previous commit "Use ip address from CNI output" introduces
ability to run pod which can havn't eth0. But also it
add problem: after kubelet restart, if we have already started
pod w/o eth0, kubelet can't find proper interface (it's
normal for vhostuser type of cni plugin when eth0 doesn't exist)
and kubelet restarts "broken" pod.
Fix of this issue requeres new feature of libcni - caching
results.
Looks like new libcni requires cniVersion in CNI output.
This patch specifies version both for CNI conf and CNI output.
Signed-off-by: Alexey Perevalov <a.perevalov@samsung.com>
