Commit Graph

2549 Commits

Author SHA1 Message Date
Da K. Ma
c16f6611bb Disabled CheckNodeMemoryPressure and CheckNodeDiskPressure predicates if TaintNodesByCondition enabled.
Signed-off-by: Da K. Ma <klaus1982.cn@gmail.com>
2018-03-11 21:26:57 +08:00
Yongkun Anfernee Gui
eba9528753 Add cache comparison for pods and pdbs 2018-03-09 15:10:26 -08:00
Yongkun Anfernee Gui
fda0d07eb6 Scheduler cache comparer
A debug tool that collects resources from api server and compares it
with the scheduler cache. It currently only compares the node list, but
it should be easy to extend. The compare is triggered by signal USER2,
by doing

  kill -12 ${SCHED_PID}

The compare result goes to scheduler log.

Towards #60860
2018-03-09 15:10:22 -08:00
Harry Zhang
5cc841a337 Use inline func to fix deadlock 2018-03-09 10:57:03 -08:00
fisherxu
b49ef6531c regenerated all files and remove all YEAR fields 2018-03-08 17:52:48 +08:00
Harry Zhang
7a7f9dccd0 [PATCH] Use nodename as key 2018-03-07 22:10:47 -08:00
Yongkun Anfernee Gui
73a9836d69 Support snapshotting a scheduler cache
Towards #60860
2018-03-07 16:08:45 -08:00
ravisantoshgudimetla
1c416b1c39 Change to fix logging 2018-03-05 11:15:33 -05:00
Harry Zhang
4e5901f947 Fixe golints of equiv class 2018-03-04 17:12:09 -08:00
Harry Zhang
6380a75484 Update generated files 2018-03-04 14:37:20 -08:00
Harry Zhang
c292af8f7b Use const in equiv class 2018-03-04 14:35:57 -08:00
Jonathan Basseri
f5ab6d5ad4 [PATCH] Fix equiv. cache invalidation of Node condition.
Equivalence cache for CheckNodeConditionPred becomes invalid when
Node.Spec.Unschedulable changes. This can happen even if
Node.Status.Conditions does not change, so move the logic around.

This logic is covered by integration test
"test/integration/scheduler".TestUnschedulableNodes but equivalence
cache is currently skipped when test pods have no OwnerReference.

Add benchmark for equivalence hashing.

Change equivalence hash function.

This changes the equivalence class hashing function to use as inputs all
the Pod fields which are read by FitPredicates. Before we used a
combination of OwnerReference and PersistentVolumeClaim info, which was
a close approximation. The new method ensures that hashing remains
correct regardless of controller behavior.

The PVCSet field can be removed from equivalencePod because it is
implicitly included in the Volume list.

Tests are now broken.

Move equivalence class hash code.

This moves the equivalence hashing code from
algorithm/predicates/utils.go to core/equivalence_cache.go.

In the process, making the hashing function and hashing function factory
both injectable dependencies is removed.

Fix equivalence cache hash tests.

Co-authored-by: Jonathan Basseri <misterikkit@google.com>
Co-authored-by: Harry Zhang <resouer@gmail.com>
2018-03-04 13:02:28 -08:00
Harry Zhang
009699fd22 Update generated types 2018-03-02 17:12:02 -08:00
Harry Zhang
b62d82422d Fix golints in extender 2018-03-02 17:12:02 -08:00
Harry Zhang
71603f2f85 Add preemption in scheduler extender
Add verb and preemption for scheduler extender

Update bazel

Use simple preemption in extender

Use node name instead of v1.Node

Fix support method

Fix preemption dup

Remove uneeded logics

Remove nodeInfo from param to extender

Update bazel for scheduler types

Mock extender cache with nodeInfo

Add nodeInfo as extender cache

Choose node name or node based on cache flag

Always return meta victims in result
2018-03-02 17:12:02 -08:00
Bobby (Babak) Salamat
ebda9584f9 Auto-create system critical prioity classes at API server startup 2018-03-02 16:50:07 -08:00
Shijun Qin
158257473a Fix a grammatical error in a comment
Fix a grammatical error in a comment in scheduler's code. We should use a word's plural form after "one of".
2018-03-02 21:30:44 +08:00
Kubernetes Submit Queue
ae1fc13aee Merge pull request #60386 from k82cn/k8s_60163
Automatic merge from submit-queue (batch tested with PRs 60683, 60386). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Added unschedulabe predicate.

Signed-off-by: Da K. Ma <madaxa@cn.ibm.com>

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #60163

**Release note**:
```release-note
None
```
2018-03-02 03:41:50 -08:00
Yang Guo
8d880506fe Support cluster-level extended resources in kubelet and kube-scheduler
Co-authored-by: Yang Guo <ygg@google.com>
Co-authored-by: Chun Chen <chenchun.feed@gmail.com>
2018-02-27 17:25:30 -08:00
jennybuckley
c8dacd8e63 Run hack/update-all.sh 2018-02-26 17:16:14 -08:00
Da K. Ma
dac59e4dd1 Added unschedulabe predicate.
Signed-off-by: Da K. Ma <madaxa@cn.ibm.com>
2018-02-25 14:00:59 +08:00
Kubernetes Submit Queue
755ab974e1 Merge pull request #58835 from ravisantoshgudimetla/critical-pod-with-priority
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Critical pod priorityClass addition

**What this PR does / why we need it**:
@bsalamat - Apologies for the delay. This PR is to ensure that all pods with priorityClassName `system-node-critical` and `system-cluster-critical` will be critical pods while preserving backwards compatibility.

**Special notes for your reviewer**:

- Moved some constants and other data structures to scheduler/api/types.go where other constants are present.
- An automatic assignment of critical priorities to pods based on critical pod annotation for backwards compatibility including some unit tests. 
xref:  https://github.com/kubernetes/kubernetes/issues/57471 

**Release note**:

```release-note
Critical pods to use priorityClasses.
```
2018-02-23 11:22:31 -08:00
Kubernetes Submit Queue
49a1478839 Merge pull request #60263 from tossmilestone/reuse-minNodes
Automatic merge from submit-queue (batch tested with PRs 60106, 59510, 60263, 60063, 59088). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Reuse the `min*Nodes` slices in order to save GC time

**What this PR does / why we need it**:
Reuse the `min*Nodes` slices to save GC time when executing `pickOneNodeForPreemption`.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #59748

**Special notes for your reviewer**:

**Release note**:

```release-note
None
```
2018-02-23 02:59:47 -08:00
tossmilestone
5a083f2038 Reuse the "min*Nodes" slices to save the GC time. 2018-02-23 14:16:19 +08:00
Kubernetes Submit Queue
a195a76151 Merge pull request #59952 from resouer/consts-handler
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Use consts as predicate key names in handlers

**What this PR does / why we need it**:

Per discussion in: https://github.com/kubernetes/kubernetes/pull/59335/files#r168351460

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #59951

**Special notes for your reviewer**:

**Release note**:

```release-note
Use consts as predicate name in handlers
```
2018-02-22 15:45:16 -08:00
Wang Guoliang
6fb1826467 Add test for scheduler:VolumeCountConflicts 2018-02-22 10:41:17 +08:00
ravisantoshgudimetla
68c20ad770 Critical pods priorityClass addition 2018-02-21 20:53:21 -05:00
Bobby (Babak) Salamat
08406c3f6e Make the Unschedulable Queue interface private 2018-02-21 13:53:40 -08:00
Bobby (Babak) Salamat
5a00c42848 Minor improvements to scheduling queue 2018-02-21 12:57:28 -08:00
Kubernetes Submit Queue
fe4b28cdf0 Merge pull request #60062 from bsalamat/sched_q_imprv
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Improve scheduling queue's logic

**What this PR does / why we need it**:
Improves scheduling queue's code based on some recent comments on [the original PR](https://github.com/kubernetes/kubernetes/pull/55109).
This PR does not fix any bugs or make any change of behavior.

**Release note**:

```release-note
NONE
```

/sig scheduling
2018-02-20 20:00:25 -08:00
Bobby (Babak) Salamat
bba9b12d0c Improve scheduling queue's logic 2018-02-20 17:20:55 -08:00
Kubernetes Submit Queue
6ba46963f8 Merge pull request #59391 from msau42/topology-beta
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Move volume scheduling and local storage to beta

**What this PR does / why we need it**:
* Move the feature gates and APIs for volume scheduling and local storage to beta
* Update tests to use the beta fields
@kubernetes/sig-storage-pr-reviews 

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #59390

**Special notes for your reviewer**:

**Release note**:

```release-note
ACTION REQUIRED: VolumeScheduling and LocalPersistentVolume features are beta and enabled by default.  The PersistentVolume NodeAffinity alpha annotation is deprecated and will be removed in a future release.
```
2018-02-20 13:26:07 -08:00
Kubernetes Submit Queue
5b98dbcfe5 Merge pull request #60008 from k82cn/k8s_54313_2
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Taint node when it under PID pressure.

Signed-off-by: Da K. Ma <madaxa@cn.ibm.com>

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
part of #54313 

**Release note**:
```release-note
If TaintNodesByCondition enabled, taint node when it under PID pressure 
```
2018-02-20 03:13:28 -08:00
Kubernetes Submit Queue
96ec318718 Merge pull request #59842 from ixdy/update-rules_go-02-2018
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

 Update bazelbuild/rules_go, kubernetes/repo-infra, and gazelle dependencies

**What this PR does / why we need it**: updates our bazelbuild/rules_go dependency in order to bump everything to go1.9.4. I'm separating this effort into two separate PRs, since updating rules_go requires a large cleanup, removing an attribute from most build rules.

**Release note**:

```release-note
NONE
```
2018-02-19 22:23:05 -08:00
Kubernetes Submit Queue
6d0b71740f Merge pull request #59968 from kubernetes/revert-59323-nodetaint
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Revert "add node shutdown taint"

Reverts kubernetes/kubernetes#59323

Node becomes unready, but is never removed. I've found the following in [kube-controller-manager.log](https://storage.googleapis.com/kubernetes-jenkins/logs/ci-kubernetes-e2e-gci-gce-autoscaling/6055/artifacts/bootstrap-e2e-master/cluster-autoscaler.log) from test run for one such node:

`E0216 01:14:27.084923       1 node_lifecycle_controller.go:686] Error determining if node bootstrap-e2e-minion-group-01b1 shutdown in cloud: failed to get instance ID from cloud provider: instance not found`

This goes on for the rest of the run (~6h). Looks like the node is stuck in Unready state because of this check: https://github.com/kubernetes/kubernetes/blob/master/pkg/controller/nodelifecycle/node_lifecycle_controller.go#L684. Previously, there was no such check and the node was removed.

Reverting as this would affect all users attempting to resize their node groups on GCE.

```release-note
NONE
```
2018-02-16 20:12:56 -08:00
Da K. Ma
6bda1bec6e Taint node when it under PID pressure.
Signed-off-by: Da K. Ma <madaxa@cn.ibm.com>
2018-02-17 10:55:29 +08:00
Da K. Ma
4df591fc5d Updated comments to correct flag of taint.
Signed-off-by: Da K. Ma <madaxa@cn.ibm.com>
2018-02-17 10:01:25 +08:00
Michelle Au
c7884550b2 Add VolumeNodeAffinity to PersistentVolumeSpec 2018-02-16 17:54:10 -08:00
Jeff Grafton
ef56a8d6bb Autogenerated: hack/update-bazel.sh 2018-02-16 13:43:01 -08:00
Aleksandra Malinowska
2d54ba3e0f Revert "add node shutdown taint" 2018-02-16 12:24:27 +01:00
Harry Zhang
136e5398ed Use consts as predicate name in handlers 2018-02-15 15:18:27 -08:00
Kubernetes Submit Queue
27daaab224 Merge pull request #59323 from zetaab/nodetaint
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

add node shutdown taint

**What this PR does / why we need it**: we need node stopped taint in order to detach volumes immediately without waiting timeout. More info in issue ticket #58635 

**Which issue(s) this PR fixes** 
Fixes #58635

**Special notes for your reviewer**:

**Release note**:
```release-note
NONE
```
2018-02-15 09:52:10 -08:00
Kubernetes Submit Queue
4949b63e0a Merge pull request #59335 from resouer/equiv-vol
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Fixes volume predicate handler for equiv class

**What this PR does / why we need it**:

Per discussion in #58797 , we are missing some predicate handler in factory.go.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #58797

Ref #58222

**Special notes for your reviewer**:

Kindly ping @msau42

**Release note**:

```release-note
Fixes volume predicate handler for equiv class
```
2018-02-14 20:57:08 -08:00
Kubernetes Submit Queue
01bd3c4b74 Merge pull request #59734 from mlmhl/format_imports
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Format some import statements in scheduler pkg

**What this PR does / why we need it**:

As the title says, apply `goimports` on some files under `pkg/scheduler` pkg.

**Release note**:

```release-note
NONE
```
2018-02-13 08:04:15 -08:00
Kubernetes Submit Queue
ba791275ce Merge pull request #59671 from bsalamat/sched_queue_perf
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Improve performance of scheduling queue by adding a hash map to track all pods with a nominatedNodeName

**What this PR does / why we need it**:
Our investigations show that there is a performance regression in the new scheduling queue which is not enabled by default and is enabled only if "priority and preemption" which is an alpha feature is enabled. This PR is an important performance improvement for those who want to use priority and preemption in larger clusters.
The PR adds a hash table to track nominated Pods so that finding such Pods will be faster.
Other than improving performance, we don't expect this PR to change behavior of scheduler.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #

ref/ #56032
ref/ #57471 

**Special notes for your reviewer**:

**Release note**:

```release-note
NONE
```

/sig scheduling
2018-02-13 00:07:58 -08:00
Kubernetes Submit Queue
821cf9234d Merge pull request #59246 from huangjiuyuan/scheduler/add-tests-for-schedulercache
Automatic merge from submit-queue (batch tested with PRs 59479, 59246). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Add tests for schedulercache

**What this PR does / why we need it**:
Add tests for `node_info.go` under `schedulercache` package.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #

**Special notes for your reviewer**:

**Release note**:
```
NONE
```
2018-02-12 17:14:31 -08:00
Kubernetes Submit Queue
ab2e1cb02a Merge pull request #59479 from tossmilestone/avoid-ecahe-update-race
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Avoid race condition when updating equivalence cache

**What this PR does / why we need it**:
Lock the ecache to update the ecache on each predicate running, to avoid race condition.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fix #58507 

**Special notes for your reviewer**:
None

**Release note**:

```release-note
None
```
2018-02-12 16:38:07 -08:00
Bobby (Babak) Salamat
df5fc09411 compare Pods by UID, not by name and namespace 2018-02-12 10:13:13 -08:00
mlmhl
b3fff71161 format some import statements in scheduler pkg 2018-02-12 09:04:00 +08:00
Kubernetes Submit Queue
74089bc4bb Merge pull request #58737 from NickrenREN/fix-scheduler-ephemeral-storage
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

Subtract local ephemeral storage resource from NodeInfo when removing pod

**What this PR does / why we need it**:
When we are removing pods, we need to subtract local ephemeral storage resource from NodeInfo

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #

**Special notes for your reviewer**:

**Release note**:
```release-note
NONE
```

/kind bug
/sig storage
/sig scheduling

/assign @jingxu97  @bsalamat
2018-02-11 13:43:01 -08:00