Commit Graph

151 Commits

Author SHA1 Message Date
Kubernetes Prow Robot
d73907fc25
Merge pull request #89380 from alculquicondor/perf-spreading
Add perf test case for Topology Spreading
2020-03-28 02:43:52 -07:00
Dave Chen
49283364bf Decouple yaml based integration test from legacy test
- Move utilities or constants out so that both of them should be able
to run independently.
- Rename the legacy test so that it can eventually be deleted when the
perf dash changes is done
2020-03-27 08:45:59 +08:00
Aldo Culquicondor
9b2ff544ed Fix pod affinity performance test configuration
Signed-off-by: Aldo Culquicondor <acondor@google.com>
2020-03-26 15:52:44 -04:00
Aldo Culquicondor
671cd33986 Add perf test cases for topology spreading
Signed-off-by: Aldo Culquicondor <acondor@google.com>
2020-03-25 10:06:09 -04:00
Aldo Culquicondor
c9314dde59 Add support for multiple label values in test cases
They are assigned in a round robin fashion

Signed-off-by: Aldo Culquicondor <acondor@google.com>
2020-03-25 10:06:09 -04:00
Kubernetes Prow Robot
7e7c4d1021
Merge pull request #89272 from alculquicondor/perf-mixed-pods
Add multiple init pods to scheduler perf test cases
2020-03-25 00:29:02 -07:00
Aldo Culquicondor
e902e70d0d Use sqrt(n) chunk size in pod affinity and core scheduler 2020-03-24 10:29:59 -04:00
Kubernetes Prow Robot
2ab6357df0
Merge pull request #88528 from ingvagabund/doc-how-to-extend-scheduler-perf-tests
[doc] scheduler_perf: describe suite configuration in more detail
2020-03-23 17:10:47 -07:00
Aldo Culquicondor
5adc4c41e3 Add multiple init pods to perf test cases
Add test case with several init pods with affinity or antiaffinity.

Signed-off-by: Aldo Culquicondor <acondor@google.com>
2020-03-23 14:55:12 -04:00
Aldo Culquicondor
0e66e56e70 Use b.Fatal instead of klog.Fatal in scheduler perf tests
The test tool doesn't work properly with klog.Fatal

Signed-off-by: Aldo Culquicondor <acondor@google.com>
2020-03-23 14:53:15 -04:00
Dave Chen
eedfb593a9 Respect flags of testing package
`go test -c -o "perf.test"`
`./perf.test --help` doesn't understand "help" flag without
calling `flag.Parse` explicitly.
2020-03-16 10:24:15 +08:00
Jan Chaloupka
b09676921c [doc] scheduler_perf: describe suite configuration in more detail
The configuration file was design as a yaml file on purpose.
To easily extend the test cases without a need to modify
the testing binary. Also, it's possible to extend the configuration
itself to enrich individual test cases.
2020-03-05 11:42:05 +01:00
Jan Chaloupka
5b3b4de972 scheduler_perf: do not override throughput labels
Throughput labels are currently initialized with a "Name" label.
So we need to append to the map instead of creating a new one.
2020-02-28 16:10:50 +01:00
Kubernetes Prow Robot
fe9073b8c1
Merge pull request #88318 from mborsz/bench
Add BenchmarkSchedulingWaitForFirstConsumerPVs benchmark
2020-02-25 07:52:49 -08:00
Maciej Borsz
bd8ed0a2a7 Add BenchmarkSchedulingWaitForFirstConsumerPVs benchmark 2020-02-25 14:41:14 +01:00
Jan Chaloupka
0f3e0b40f6 scheduler_perf: describe how to run BenchmarkPerfScheduling manually 2020-02-24 19:12:00 +01:00
Cong Liu
7f56c753b3 Make MetricCollector configurable for scheduler benchmark tests 2020-02-18 14:02:57 -08:00
Benjamin Elder
3fb7183215 bump pause to 3.2 in test/ 2020-02-14 14:03:19 -08:00
Jan Chaloupka
7b5534021c Collect some of scheduling metrics and scheduling throughput
In addition to getting overall performance measurements from golang benchmark,
collect metrics that provides information about insides of the scheduler itself.
This is a first step towards improving what we collect about the scheduler.

Metrics in question:
- scheduler_scheduling_algorithm_predicate_evaluation_seconds
- scheduler_scheduling_algorithm_priority_evaluation_seconds
- scheduler_binding_duration_seconds
- scheduler_e2e_scheduling_duration_seconds

Scheduling throughput is computed on the fly inside perfScheduling.
2020-02-13 13:32:09 +01:00
Mike Danese
25651408ae generated: run refactor 2020-02-08 12:30:21 -05:00
Mike Danese
3aa59f7f30 generated: run refactor 2020-02-07 18:16:47 -08:00
Mike Danese
38ecb30c58
Revert "Collect some of scheduling metrics and scheduling throughput" 2020-02-06 10:18:00 -08:00
Jan Chaloupka
8a1c4a5a88 Collect some of scheduling metrics and scheduling throughput
In addition to getting overall performance measurements from golang benchmark,
collect metrics that provides information about insides of the scheduler itself.
This is a first step towards improving what we collect about the scheduler.

Metrics in question:
- scheduler_scheduling_algorithm_predicate_evaluation_seconds
- scheduler_scheduling_algorithm_priority_evaluation_seconds
- scheduler_binding_duration_seconds
- scheduler_e2e_scheduling_duration_seconds

Scheduling throughput is computed on the fly inside perfScheduling.
2020-02-02 18:02:49 +01:00
notpad
6511b603c9 Define workloads specs by YAML 2019-12-19 09:08:11 +08:00
Abdullah Gharaibeh
a051c591a1 Fix inter-pod affinity scheduler benchmarks 2019-12-10 10:00:28 -05:00
Abdullah Gharaibeh
da67bcbd35 added benchmarks for preferred (anti)pod affinity 2019-12-05 10:50:00 -05:00
Abdullah Gharaibeh
e9c400aabc update scheduler benchmarks to be more representative 2019-12-02 13:19:14 -05:00
Cong Liu
8995c1e030 Update bucket for scheduler framework latency histograms. 2019-11-13 16:35:59 -05:00
Abdullah Gharaibeh
134baa50b9 Convert multiple node label predicates to be a single filter plugin. 2019-11-05 10:34:51 -05:00
draveness
e5a23f8dda fix: scheduler perf test with pod informer 2019-10-15 13:54:45 +08:00
Ahmad Diaa
4448a1cea9 move factory package to scheduler 2019-10-12 23:03:07 +02:00
hwdef
321b690f62 fix(test): delete unused unused var or const 2019-09-20 14:41:53 +08:00
draveness
54e0f4773c feat(scheduler): use api server to watch scheduled pods 2019-09-17 11:24:59 +08:00
Kubernetes Prow Robot
3b393e3d17
Merge pull request #82330 from draveness/feature/remove-additional-cost-in-the-scheduler-benchmark
fix(scheduler): remove the defer function cost
2019-09-11 15:26:23 -07:00
draveness
4105fea6eb fix(scheduler): remove the defer function cost 2019-09-05 10:57:11 +08:00
draveness
6cfec4d919 feat: remove factory.Config from mustSetupScheduler 2019-08-21 19:54:28 +08:00
draveness
6d77624bc3 feat: use scheduler.New instead in createSchedulerConfigurator 2019-08-21 19:09:57 +08:00
Aldo Culquicondor
d935b1054c Remove Configurator interface
The Configurator has been used as a holder for listers that tests need,
which is not its purpose. By making the tests obtain listers from more
appropriate places, such as informers, there is no need for various
accessors to the Configurator.

Also, FakeConfigurator is not being used anymore, so there's no need for
an interface instead of a plain pointer.

Signed-off-by: Aldo Culquicondor <acondor@google.com>
2019-08-01 08:36:38 -04:00
Jan Safranek
f6430c0159 Add benchmark for scheduling of pods with PVs 2019-06-26 14:07:17 +02:00
yameiwang
8d07b2641e fix typo in /test/integration 2019-05-30 03:47:43 +08:00
xichengliudui
0b184d35a1 fixgo lint failures test/integration/...
update pull request

update pull request

add files

update pull request
2019-02-19 04:41:50 -05:00
xichengliudui
641f883b82 fix shellcheck in test/integration/... and test/kubemark/... 2019-02-18 01:16:39 -05:00
Kubernetes Prow Robot
b50c643be0
Merge pull request #73540 from rlenferink/patch-5
Updated OWNERS files to include link to docs
2019-02-08 09:05:56 -08:00
Davanum Srinivas
b975573385
move pkg/kubelet/apis/well_known_labels.go to staging/src/k8s.io/api/core/v1/
Co-Authored-By: Weibin Lin <linweibin1@huawei.com>

Change-Id: I163b2f2833e6b8767f72e2c815dcacd0f4e504ea
2019-02-05 13:39:07 -05:00
Roy Lenferink
b43c04452f Updated OWNERS files to include link to docs 2019-02-04 22:33:12 +01:00
AdamDang
f870382587
Fix some error in returned message
Fix some error in returned message:
a interval -> an interval
to small -> too small
2019-01-29 12:09:37 +08:00
Krzysztof Siedlecki
04c4bef206 adding 5000 nodes test to scheduler benchmark 2018-12-10 13:54:53 +01:00
Davanum Srinivas
954996e231
Move from glog to klog
- Move from the old github.com/golang/glog to k8s.io/klog
- klog as explicit InitFlags() so we add them as necessary
- we update the other repositories that we vendor that made a similar
change from glog to klog
  * github.com/kubernetes/repo-infra
  * k8s.io/gengo/
  * k8s.io/kube-openapi/
  * github.com/google/cadvisor
- Entirely remove all references to glog
- Fix some tests by explicit InitFlags in their init() methods

Change-Id: I92db545ff36fcec83afe98f550c9e630098b3135
2018-11-10 07:50:31 -05:00
Wei Huang
5259d09c38
add more scheduler benchmark testcases
- add benchmark test for PodAffinity
- add benchmark test for NodeAffinity
- add 1000-nodes test for PodAntiAffinity/PodAffinity/NodeAffinity
2018-10-24 11:06:32 -07:00
k8s-ci-robot
94306c12f5
Merge pull request #69057 from denkensk/create-a-new-scheduler-constructor
create-a-new-scheduler-constructor
2018-10-11 13:45:02 -07:00