-
Notifications
You must be signed in to change notification settings - Fork 7.3k
Performance with multiple pods in kubernetes #39304
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
@chuckbutkus could you provide the show-config output and / or the full Pod yaml to make sure we're not missing anything in attempting to reproduce this? |
|
Pod YAML:
|
Here is the timing output against that specific instance (3 pods were running):
|
thank you @chuckbutkus we'll try to reproduce this. If possible can you try @ahus1 suggestion from #39023 to enabled tracing? |
Please follow #39023 for a resolution. Closing this one as a duplicate. |
Uh oh!
There was an error while loading. Please reload this page.
Before reporting an issue
Area
core
Describe the bug
We previously ran version 26.1.1 and had no performance issues. After upgrading to 26.1.5 we saw our app pause occasionally. After investigation we found it was Keycloak after the upgrade. We tried the latest version (26.2.1) and it has the same pauses, so we had to revert back to 26.1.1.
Investigating further it appears to be something with how multiple pods interact. When we reduce the number of pods to one, everything is fine on all versions.
We install Keycloak via the bitnami helm chart, so it deploys as a statefulset.
Version
26.2.1
Regression
Expected behavior
Accessing Keycloak to get token or user data should not occasionally pause for several seconds.
Actual behavior
Accessing Keycloak to get token or user data occasionally pauses for several seconds.
How to Reproduce?
Run Keycloak (any version from 26.1.5 onward) with multiple pods (not sure if it matters, but we install via the bitnami helm chart). Then run the following python program:
import time
Against our install, we will often see long pauses (see below). On version 26.1.1 or with only a single pod we don't see these pauses.
Anything else?
This seems related to this issue: #39023
The text was updated successfully, but these errors were encountered: