You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Mr TheSegfault (JIRA)" <ji...@apache.org> on 2019/07/30 13:49:00 UTC

[jira] [Commented] (MINIFICPP-959) Review librdkafka thread safety

    [ https://issues.apache.org/jira/browse/MINIFICPP-959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16896144#comment-16896144 ] 

Mr TheSegfault commented on MINIFICPP-959:
------------------------------------------

[~le.nghia] This I think would be easier to verify if we actually had tests on kafka's usage. PublishKafka was a quick implementation and we really need to beef tests under it. I think we need to link this to MINIFICPP-976, because docker will be a very easy environment to pull down a docker image with a kafka broker. I don't think we should create integration tests in our normal test suite

> Review librdkafka thread safety
> -------------------------------
>
>                 Key: MINIFICPP-959
>                 URL: https://issues.apache.org/jira/browse/MINIFICPP-959
>             Project: Apache NiFi MiNiFi C++
>          Issue Type: Bug
>            Reporter: Daniel Bakai
>            Assignee: Nghia Le
>            Priority: Major
>
> I don't think librdkafka is doing what it is supposed to around the KafkaLease and the atomic spinlock combined with mutexes.
> It is thread safe in a way that resources are only accessed by one thread at a time, but the behaviour otherwise is I think not what was intended (multiple threads can initialize a KafkaConnection after each other, for example).



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)