NR-250703: cluster-autoscaler improvements for GCP #36
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Taint utils taking multiple taints
…wn-after-add-per-ng-poc feat: support `--scale-down-delay-after-*` per nodegroup
Rancher: Fix error messages and expose underlying error.
Existing bucketing is inconsistent. Specifically, the second to last bucket is [100, 1000), which is huge and doesn't allow to differentiate between something that took 2m (120s) and something that took 15m (900s).
Use exponential buckets for function_duration_seconds
fix(kwok): prevent quitting when scaling down node group
Fix VPA e2e test failures
…om/guopeng0/autoscaler into feature/node_group_healthy_metrics
…n_ds_v2 Allow draining when DaemonSet kind has custom API Group
…ealthy_metrics feat:add node group health and back off metrics
Remove unused NodeInfoProcessor
Signed-off-by: Yuki Iwai <yuki.iwai.tz@gmail.com>
CA: Before we perform go test, synchronizing go modules
The grouping should be made by the schedulability equivalence meaning we can introduce optimizations to the binpacking. Introduce a benchmark that estimates capacity needed for 51k pods, which can be grouped to two equivalence groups 50k and 1k.
Add a link to the sample manifest and update the image used in the example. Signed-off-by: Lennart Jern <lennart.jern@est.tech>
Bumps golang from 1.22.1 to 1.22.2. --- updated-dependencies: - dependency-name: golang dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>
Bumps golang from 1.22.1 to 1.22.2. --- updated-dependencies: - dependency-name: golang dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>
Bumps golang from 1.22.1 to 1.22.2. --- updated-dependencies: - dependency-name: golang dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>
Update CAPI docs
…vertical-pod-autoscaler/pkg/recommender/golang-1.22.2 Bump golang from 1.22.1 to 1.22.2 in /vertical-pod-autoscaler/pkg/recommender
…vertical-pod-autoscaler/pkg/updater/golang-1.22.2 Bump golang from 1.22.1 to 1.22.2 in /vertical-pod-autoscaler/pkg/updater
The optimization uses the fact that pods which are equivalent do not need to be check multiple times against already filled nodes. This changes the time complexity from O(pods*nodes) to O(pods).
Refactor estimation
…policy-example docs: precise AWS IAM policy example
Bump default VPA version to 1.1.0
CA: Fix apis vendoring
Fix broken link in README.md to point to equinixmetal readme
Include helm chart version in cluster-autoscaler version matrix
…vertical-pod-autoscaler/pkg/admission-controller/golang-1.22.2 Bump golang from 1.22.1 to 1.22.2 in /vertical-pod-autoscaler/pkg/admission-controller
Add support for label(tags in Azure) based auto-discovery of GCP Managed Instance Groups. The discovery includes figuring out the min and max sizes for the mig pool. Add unit-tests for all the new code added. Add GCP auto-discovery documentation.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add support for label(tags in Azure) based auto-discovery of GCP Managed Instance Groups. The discovery includes figuring out the min and max sizes for the mig pool.
Add unit-tests for all the new code added.
Add GCP auto-discovery documentation .
Auto-Discovery Setup
To run a cluster-autoscaler which auto-discovers instance groups, use the
--node-group-auto-discoveryflag. There are 2 auto-discovery options to choose from.Auto-Discovery by Labels
For example,
--node-group-auto-discovery=label:cluster-autoscaler-enabled=true,cluster-autoscaler-name=<YOUR CLUSTER NAME>will find all the instance groups with instance templates that are tagged with those labels containing those values.NOTE
cluster-autoscaler-name=<YOUR CLUSTER NAME>whencluster-autoscaler-enabled=trueis used across many clusters to prevent Instance Groups from different clusters recognized as the node groups--nodesflags passed to cluster-autoscaler because the node groups are automatically discovered by tagsmin/maxvalues are provided when using this option. cluster-autoscaler will detect the "min" and "max" labels on the Instane Group resource in GCP, adjusting the desired number of nodes within these limits.min/maxlabels on the Instance Group resource, cluster-autoscaler will use the default min/max values of 0 and 1000 respectively.Auto-Discovery by NamePrefix
For example,
--node-group-auto-discovery=mig:namePrefix=test-lemon-peel-mp,min=2,max=10will internally use a Regular Expression to find all the instance groups whose name begins withtest-lemon-peel-mpand set the minimum and maximum number of nodes to 2 and 10 respectively.NOTE
MinandMaxkey/value pairs wheremax > minmust be specified when using this option and will not use any defaults.--node-group-auto-discoveryflag multiple times. Ex: