Even though RollingUpdate is the default updateStrategy, we need to
specify it explicitly here because otherwise updating from
extensions/v1beta1 to apps/v1 doesn't change the updateStrategy.
This is the 2nd attempt. The previous was reverted while we figured out
the regional mirrors (oops).
New plan: k8s.gcr.io is a read-only facade that auto-detects your source
region (us, eu, or asia for now) and pulls from the closest. To publish
an image, push k8s-staging.gcr.io and it will be synced to the regionals
automatically (similar to today). For now the staging is an alias to
gcr.io/google_containers (the legacy URL).
When we move off of google-owned projects (working on it), then we just
do a one-time sync, and change the google-internal config, and nobody
outside should notice.
We can, in parallel, change the auto-sync into a manual sync - send a PR
to "promote" something from staging, and a bot activates it. Nice and
visible, easy to keep track of.
It is expected that nodes with extended resources attached will be
tainted with the resouce name, so that we can create dedicated nodes.
If ExtendedResourceToleration admission controller is enabled, pods
requesting such resources will automatically tolerate such taints.
nvidia-gpu-device-plugin daemonset doesn't request such resources but
still needs to run on such nodes, so it needs this toleration.
- Runs as system critical pod
- Makes resource limits to match its resource requets
- Modifies test/e2e/scheduling/nvidia-gpus.go to cope with the recent
change of running the device plugin as a system addon.
- The resource settings of the addon is based on the test results
from 8 nvidia-tesla-k80 gpus.
Rohit Agarwal