[
https://issues.apache.org/jira/browse/KAFKA-4264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15796432#comment-15796432
]
Ewen Cheslack-Postava commented on KAFKA-4264:
----------------------------------------------
Is this actually true? The kafka-server-start.sh script has {{kafka.Kafka}} in
it as that is the class it executes:
https://github.com/apache/kafka/blob/trunk/bin/kafka-server-start.sh#L44 I
agree this approach isn't ideal (a pid file would probably be a better
solution), but it seems like this currently works ok. Is it possible the
`kafka.Kafka` output is omitted in your case due to a long command line that
pushes it past the length of output provided by {{ps}}? I've just tested
locally and it is working fine here.
> kafka-server-stop.sh fails is Kafka launched via kafka-server-start.sh
> ----------------------------------------------------------------------
>
> Key: KAFKA-4264
> URL: https://issues.apache.org/jira/browse/KAFKA-4264
> Project: Kafka
> Issue Type: Improvement
> Components: tools
> Affects Versions: 0.10.0.1
> Environment: Tested in Debian Jessy
> Reporter: Alex Schmitz
> Priority: Trivial
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> kafka-server-stop.sh greps for the process ID to kill with the following:
> bq. PIDS=$(ps ax | grep -i 'kafka\.Kafka' | grep java | grep -v grep | awk
> '{print $1}')
> However, if Kafka is launched via the kafka-server-start.sh script, the
> process doesn't include kafka.Kafka, the grep fails to find the process, and
> it returns the failure message, No Kafka server to stop.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
