[Access-tokens][ClusterLoader2] Add standalone test #1021
Conversation
k8s-ci-robot
added
the
cncf-cla: yes
|
Hi @jprzychodzen. Thanks for your PR. I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
k8s-ci-robot
added
needs-ok-to-test
| @@ -0,0 +1,138 @@ | |||
| # Stress testing access token validation | |||
| # | |||
| # Targeting 2 000 tokens with 20 000 total QPS | |||
There was a problem hiding this comment.
Ideally, the test should scale with the cluster size (as load test does). In such case we'll be able to run it along our CI/CD tests at 100, 500 and 5k node scale.
There was a problem hiding this comment.
Added linear scaling of QPS. Actually, we could get more QPS for smaller clusters, but this sets reasonable requirements and expectations what can be delivered.
jprzychodzen
force-pushed
the
at-add-config
branch
2 times, most recently
from
6b07414
to
fa776e5
Compare
|
Decreased QPS target number to match real-world expectations |
| steps: | ||
| - name: Starting measurements | ||
| measurements: | ||
| - Identifier: APIResponsivenessPrometheus |
There was a problem hiding this comment.
Please also start the TestMetrics measurement (you can see how it's done in density or load test).
It's a bundle measurement producing a lot of useful data (e.g. profiles, restart checks, resource usages, etc.)
| # Stress testing access token validation | ||
| # | ||
| # Targeting 2 000 tokens with 5 000 total QPS for 5k node cluster, so it's 2.5 QPS per token. | ||
| # For smaller cluster, we want to scale down lineary QPS per token to 2.5 * (Number of nodes)/(5 000). |
There was a problem hiding this comment.
Could you make it more explicit that this test always creates 2K tokens and 80 pods and only the QPS depends on the cluster size?
There was a problem hiding this comment.
Added comment and calculations.
jprzychodzen
force-pushed
the
at-add-config
branch
from
fa776e5
to
f68953d
Compare
|
/ok-to-test |
k8s-ci-robot
added
ok-to-test
| # Targeting 2 000 tokens with 5 000 total QPS for 5k node cluster, so it's 2.5 QPS per token. | ||
| # | ||
| # Number of tokens does not scale with number of nodes. | ||
| # By default, those 2 000 tokens are are assigned to 80 service accounts, with 25 tokens each. |
There was a problem hiding this comment.
Sorry for nitpicking, but I'd also add a line about number of pods.
IIUC there will be always 80 pods, which means that this test can be run even on a 1 node cluster (assuming the node is big enough). If we had more pods then we'd run into 110pods/node limitation.
Add standalone test for ClusterLoader2 that allows benchmarking and testing access tokens thresholds. This test is based on Mateusz's code from github.com/mm4tt/k8s-util.
jprzychodzen
force-pushed
the
at-add-config
branch
from
f68953d
to
68a1d78
Compare
|
/lgtm |
k8s-ci-robot
added
the
lgtm
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: jprzychodzen, mm4tt The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
the
approved
Add standalone test for ClusterLoader2 that allows benchmarking and
testing access tokens thresholds.
This test is based on Mateusz's code from github.com/mm4tt/k8s-util.