kafka-jira mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ewen Cheslack-Postava (JIRA)" <j...@apache.org>
Subject [jira] [Created] (KAFKA-6676) System tests do not handle ZK chroot properly with SCRAM
Date Sat, 17 Mar 2018 21:24:00 GMT
Ewen Cheslack-Postava created KAFKA-6676:

             Summary: System tests do not handle ZK chroot properly with SCRAM
                 Key: KAFKA-6676
                 URL: https://issues.apache.org/jira/browse/KAFKA-6676
             Project: Kafka
          Issue Type: Bug
            Reporter: Ewen Cheslack-Postava
            Assignee: Ewen Cheslack-Postava

This is related to the issue observed inĀ KAFKA-6672. There, we are now automatically creating
parent nodes if they do not exist. However, if using a chroot within ZK and that chroot does
not yet exist, you get an error message about "Path length must be > 0" as it tries to
create all the parent paths.

It would probably be better to be able to detect this issue and account for it, but currently
system test code will fail if you use SCRAM and a chroot because while Kafka will create the
chroot when it starts up, there are some commands related to security that may need to be
executed before that and assume the chroot will already be there.

We're currently missing this because while the chroot option is there, nothing in Kafka's
tests are currently exercising it. So given what is apparently a common assumption in tools
that the chroot already exists (since I think the core kafka server is the only thing that
handles creating it if needed), I think the fix here would be two-fold:
 # Make KafkaService ensure the chroot exists before running any commands that might need
 # On at least one test that exercises security support, use a zk_chroot so that functionality
is at least reasonably well exercised.

It would be good to have this in both trunk and 1.1 branches.

This message was sent by Atlassian JIRA

View raw message