shutdown api has been changed on mesos 0.24.1

[asdfsx]

Mesos 0.24.0 added experimental support for v1 Scheduler HTTP API. and I cannot shutdown the framwork by using the following command:

curl -d "frameworkId=YOUR_FRAMEWORK_ID" -X POST http://YOUR_MESOS_URL:5050/master/shutdown

Now I use the following cmd to shutdown the framework

curl -X POST -H "Content-Type: application/json" http://YOUR_MESOS_URL:5050/api/v1/scheduler -d@teardown.json

Content of the teardown.json:

{
  "framework_id"    : {"value" : "framework id"},
  "type"            : "TEARDOWN"
} 

[christobill]

This also works:

curl -d 'frameworkId=YOUR_FRAMEWORK_ID' -X POST http://YOUR_MESOS_URL:5050/master/teardown 

[asdfsx]

@christobill maybe I got something wrong before -_- ,I'll close this issue!

[asdfsx]

@christobill I found if I boot up the hdfs by docker, I can not stop it ~~I'm confused by this

docker run -it --rm -v "$PWD"/hdfs-mesos-0.1.5:/hdfs-mesos-0.1.5 --net host --workdir /hdfs-mesos-0.1.5 mesos_jdk:0.24.1_1.7 bin/hdfs-mesos  

mesos_jdk is a image I build for execute hdfs-mesos, Dockerfile as follow

FROM centos:7

RUN rpm -i http://repos.mesosphere.io/el/7/noarch/RPMS/mesosphere-el-repo-7-1.noarch.rpm && \
yum -y install mesos-0.24.1
RUN yum -y install java-1.7.0-openjdk
RUN yum -y install which

[christobill]

@asdfsx what I meant is: you are absolutely right, doc/README.md is not up-to-date. You should reopen the issue.

[christobill]

README.md should be changed like this:

2 Shutdown the hdfs framework in Mesos: curl -d "frameworkId=YOUR_FRAMEWORK_ID" -X POST http://YOUR_MESOS_URL:5050/master/teardown (or curl -d "frameworkId=YOUR_FRAMEWORK_ID" -X POST http://YOUR_MESOS_URL:5050/master/shutdown for mesos<0.24.1)

[christobill]

@asdfsx concerning your docker issue. Stop the docker container. Get the framework id in mesos UI. curl -d "frameworkId=YOUR_FRAMEWORK_ID" -X POST http://YOUR_MESOS_URL:5050/master/teardown (or curl -d "frameworkId=YOUR_FRAMEWORK_ID" -X POST http://YOUR_MESOS_URL:5050/master/shutdown for mesos<0.24.1) With zkCli, rmr /hdfs-mesos

I am just quoting the doc (appart from the shutdown/teardown change)

[asdfsx]

@christobill I guess this depends on the situation. If you have succescfully started the frame work, you can use the old style to shutdown the framework. If you failed to start the framework, you should use "teardown" to kill all the executors. "teardown" always work, no matter whether the framework was started successfuly.