Restart Scheduler

During our installation phase we had the ToRC-scheduler running in a minimal mode. The main purpose of the minimal mode is to have a DNS server running during the installation of the compute nodes. Now that we have all compute nodes up and running and connected to Mesos we can re-start ToRC using the regular configuration.

Check that all nodes are up

Use Mesos Admin UI to verify that all slaves (compute & wedge) are up and connected before you proceed.

mesos-all-slaves

Verify Scheduler Configuration

Log in to the wedge and edit the scheduler configuration file.

root@wedge:~# cd ~/torc_scheduler/config/
root@wedge:~/torc_scheduler/config# vi config.yml

Verify that the compute node settings are correct. By default they match our demo setup.

nodes:
- name: wedge
  ip: $MASTER_IP
  type: master
- name: bladerunner1
  ip: 10.250.3.21
  type: slave
- name: bladerunner2
  ip: 10.250.3.22
  type: slave
- name: bladerunner3
  ip: 10.250.3.23
  type: slave
- name: bladerunner4
  ip: 10.250.3.24
  type: slave

Rebuild the docker image for the ToRC Scheduler. The config files are part of the container image.

root@wedge:~# cd ~/torc-scripts/deploy
root@wedge:~/torc-scripts/deploy# ./docker_build_scheduler_master.sh

Kill torc-scheduler and restart with standard configuration. Replace 10.250.3.20 with the IP of your wedge.

root@wedge:~/torc-scripts/deploy# ./kill_scheduler_master.sh
root@wedge:~/torc-scripts/deploy# ./run_scheduler_master.sh 10.250.3.20

Log file can be checked using following script. Press ctrl-c to stop the log output.

root@wedge:~/torc-scripts/deploy# ./log_scheduler_master.sh

Use Mesos Admin to check that all ToRC core services are up and running.

mesos-all-slaves

Next, let’s explore the functionality offered by the ToRC core services.