Hi Colin, Thank you for your review. I removed the "absolute path need to be provided" line from the KIP, and will modify the code to get the absolute path to the config files using some bash in the kafka-server-start file. For your second question, I've added a line in the KIP: "If both parameters are provided, the value for node-id parameter will take precedence, i.e, the process with node id specified will be killed, no matter what's the process role provided."
What do you think? Thanks, Hailey On Thu, Nov 9, 2023 at 4:03 PM Colin McCabe <cmcc...@apache.org> wrote: > Hi Hailey, > > Thanks for the KIP. > > It feels clunky to have to pass an absolute path to the configuration file > when starting the broker or controller. I think we should consider one of > two alternate options: > > 1. Use JMXtool to examine the running kafka.Kafka processes. > I believe ID is available from kafka.server, type=app-info,id=1 (replace 1 > with the actual ID) > > Role can be deduced by the presence or absence of > kafka.server,type=KafkaServer,name=BrokerState for brokers, or > kafka.server,type=ControllerServer,name=ClusterId for controllers. > > 2. Alternately, we could inject the ID and role into the command line in > kafka-server-start.sh. Basically add -Dkafka.node.id=1, > -Dkafka.node.roles=broker. This would be helpful to people just examining > the output of ps. > > Finally, you state that either command-line option can be given. What > happens if both are given? > > best, > Colin > > > On Mon, Oct 23, 2023, at 12:20, Hailey Ni wrote: > > Hi Ron, > > > > I've added the "Rejected Alternatives" section in the KIP. Thanks for the > > comments and +1 vote! > > > > Thanks, > > Hailey > > > > On Mon, Oct 23, 2023 at 6:33 AM Ron Dagostino <rndg...@gmail.com> wrote: > > > >> Hi Hailey. I'm +1 (binding), but could you add a "Rejected > >> Alternatives" section to the KIP and mention the "--required-config " > >> option that we decided against and the reason why we made the decision > >> to reject it? There were some other small things (dash instead of dot > >> in the parameter names, --node-id instead of --broker-id), but > >> cosmetic things like this don't warrant a mention, so I think there's > >> just the one thing to document. > >> > >> Thanks for the KIP, and thanks for adjusting it along the way as the > >> discussion moved forward. > >> > >> Ron > >> > >> > >> Ron > >> > >> On Mon, Oct 23, 2023 at 4:00 AM Federico Valeri <fedeval...@gmail.com> > >> wrote: > >> > > >> > +1 (non binding) > >> > > >> > Thanks. > >> > > >> > On Mon, Oct 23, 2023 at 9:48 AM Kamal Chandraprakash > >> > <kamal.chandraprak...@gmail.com> wrote: > >> > > > >> > > +1 (non-binding). Thanks for the KIP! > >> > > > >> > > On Mon, Oct 23, 2023, 12:55 Hailey Ni <h...@confluent.io.invalid> > >> wrote: > >> > > > >> > > > Hi all, > >> > > > > >> > > > I'd like to call a vote on KIP-979 that will allow users to > >> independently > >> > > > stop KRaft processes. > >> > > > > >> > > > > >> > > > > >> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-979%3A+Allow+independently+stop+KRaft+processes > >> > > > > >> > > > Thanks, > >> > > > Hailey > >> > > > > >> >
