-
Notifications
You must be signed in to change notification settings - Fork 36
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
fix: SASL/PLAIN #255
fix: SASL/PLAIN #255
Conversation
do you think we could extend the end2end tests accordingly? It seems that we already have some tests that try to do that, but it didn't seem to reproduce your issue: Lines 28 to 50 in e65584d
|
I'll look into it. My tests were also against the bitnami image, so it should be possible to reproduce the issue there. |
fc83114
to
e266124
Compare
Turns out you have to set Also the test just seems to hang before this patch, not entirely sure why. I made the test time out after 2 seconds, it passes in way less than that under normal circumstances. I ran
|
Ah, because of the infinite backoff of course. Better solution is to disable that. |
dc3007f
to
39d5608
Compare
The previous fix did not work. This is now definitely verified to be working and also makes coherent sense.
39d5608
to
088209c
Compare
This turned out to be a major hassle to get working. The CI setup didn't configure any connection that allows SASL, so I added that which produced an error complaining about a missing jaas config. Looking into mounting that I found that doing so is essentially impossible in circleci without major changes to the build. Then I found that you were actually using very old versions of the bitnami image, maybe assuming that The fix was to update the kafka images and adding some more environment variables, which seems to work. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Awesome. Thanks a lot for going the extra mile and also improve the testing. I really appreciate that!
The previous fix did not work. This is now definitely verified to be working and also makes coherent sense.
I made a fix in the OAUTHBEARER PR to fix the SASL flow to send a message even when the final state was
Finished(true)
. My solution instead broke the flow in a different way, making it so that we send the final message twice.This should fix it. I didn't catch it before because my kafka configuration was wrong and didn't properly require plain auth, and this particular issue doesn't affect oauthbearer. This time I've verified that it fails with
InvalidSaslState
without this fix, and manages to connect with it.