-
-
Notifications
You must be signed in to change notification settings - Fork 545
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
KIP-601: Change the authentication timeout to 30 seconds #1332
Comments
What's the property of the corresponding Java client that governs this? I can't find any reference to it in the docs, nor could I find anything for librdkafka. |
Kafka java clients does not have specific authentication timeout. Instead we have connection setup timeout (includes tcp connection setup as well SSL/SASL handshake). default is 10 seconds with the connection setup timeout increasing to a maximum of 30 seconds. See:
librdkafka also has similar config socket.connection.setup.timeout.ms Line 44 in 4246f92
|
|
Yes. I think we can start by bumping that up. If you think implementing a deferred retry policy is easy, we can start with a smaller timeout of say 5 seconds and then gradually increase it up to 30 seconds. But if a static retry policy is easier, let's do 30 seconds. Wdyt @Nevon? |
I actually already bumped it to a static 10s, and forgot to reference this issue. My reasoning is in #1340, but essentially if we set it to 30s now it'll be more difficult to implement a retry policy later, because we'll need to lower the initial timeout value which will could cause issues for some folks. We also have to factor in the I'm gearing up to release v2.0.0, so I didn't want to introduce a major change at this point, like a retry policy with exponentially longer timeouts, so this seemed like a reasonable compromise. Do you agree that this issue can be closed with that change? |
@Nevon I think you can close this issue given the increase from 1s to 10s. |
Hi @Nevon , We are using kafkajs with Confluent cloud. The have some cluster rolls activities where they might take some broker out of the cluster for some maintenance. During one of such cluster rolls we saw a lots of following errors
As per the mechanics of a roll, the connections to the broker being bounced were dropped We are using latest version 2.2.4 with default values of connectionTimeout=1sec and authenticationTimeout=10sec. Since in kafkajs we have different timeouts for connection and authentication, what values of connectionTimeout and authenticationTimeout do you suggest for overall connection establishment timeout of 30 sec. Looking forward for your suggestion here. Thanks in advance :) |
You guys should update the docs to reflect the updated information. |
Describe the bug
The default SASL authentication timeout is set to 1 second. This is not a good default since it'll result in a lot of reconnects until the Broker can service the auth request in under 1 second. Can we bump this up to 30 seconds instead? This is more in-line with the official Kafka clients.
To Reproduce
Expected behavior
Observed behavior
The text was updated successfully, but these errors were encountered: