#This project is deprecated. The mesos offer/subscription mechanism doesn't work well dealing with workload managers. You should use our cloudify HPC plugin instead.
The HPC framework for Mesos runs tasks on different HPC machines. It is in heavy development (see issues) and for now should not be considered as estable.
- Framework sending only dummy tasks. REST API to send tasks to the framwork to be developed.
- Only HPCs using Slurm are supported (only an Slurm executor developed).
- Compile and install mesos:
$ ./scripts/build-dependencies.sh
- Compile the framework and executor:
$ ./scripts/build.sh
- Create the executor credentials:
$ cd ./slurm_executor/etc
$ cp config.json_sample config.json
- Finally edit the file created to add the SSH credentials to the HPC machine.
To run the framework a mesos instance running somewhere is needed. The exact command arguments to run the hpc framework it is going to depend on this configuration. Some examples:
- Mesos with only one master not using authorization mechanism on 192.168.56.10 and listening on default port.
$ cd ./hpc_framework/Release
$ DEFAULT_PRINCIPAL= ./hpc-framework --master=192.168.56.10:5050
- Mesos running with zookeeper listening on default port and not using authorization mechanism.
$ cd ./hpc_framework/Release
$ DEFAULT_PRINCIPAL= ./hpc-framework --master=zk://node1:2181/mesos