Wojciech Tyczynski

Define and implement scheduling latency SLO

15

@ahg-g - if you can provide more details based on our internal work

ClusterLoader: Combine provider specific code in a single place

3

There are multiple measurements that introduce provider specific code, e.g.: https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/common/etcd_metrics.go#L95 https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/common/scheduler_latency.go#L178 Those should be extracted from measurements to a separate single location. @shyamjvs

Prometheus and its operator should be using protobufs when talking to apiserver

3

Currently, they seem to use jsons.

Refactor wait_for_{pods, pvcs, pvc} and {pods, pvcs, pvs} files

4

We're accumulating a lot of code duplication for different types: - pods: https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/util/pods.go https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/common/wait_for_pods.go - pvcs https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/util/pvcs.go https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/common/wait_for_pvcs.go - pvs https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/util/pvs.go https://github.com/kubernetes/perf-tests/blob/master/clusterloader2/pkg/measurement/common/wait_for_pvs.go Those should be refactor to share as much...

Should volume formatting be included in pod startup time?

2

Forked from: https://github.com/kubernetes/perf-tests/pull/657 from @msau42 > This discussion also brought up an interesting point. We are not measuring provisioning time because we said that is bootstrapping behavior that only happens...

Simplify load test

15

With migration to CL2 being roughly done, I would like to simplify and evolve load test a bit to: 1. create a service in a bundle with RC 2. that...

ClusterLoader2 bundle should assume that "bundle is bundle"

4

Context: https://github.com/kubernetes/perf-tests/pull/259/files#r226558746

Expose more information in clusterloader2 logs

6

There are a couple things that we definitely need: - [x] more state about pods from a given controlling object (number of pending, waiting, checking if something was deleted, etc.)....

[WIP] Consistent reads mitigation

1

Ref #2340 /cc @serathius @deads2k

Implement resilient watchcache initialization post-start-hook

7

```release-note NONE ``` /kind feature /priority important-longterm /sig api-machinery