Nomad-Vagrant Cluster
This cluster is used primarily to test Nomad.
Prereqs
What is Nomad exactly?
"HashiCorp Nomad is a single binary that schedules applications and services on Linux, Windows, and Mac. It is an open source scheduler that uses a declarative job file for scheduling virtualized, containerized, and standalone applications."
A single region Nomad cluster might look like:
A distributed multi-region cluster might look:
Getting started
Once you've cloned the repo, the Vagrantfile will bootstrap three virtual-machines; a Nomad Server and two Nomad Clients.
Run:
vagrant up
After creating the virtual machines, vagrant status will tell you if everything is working.
christopherblack$ vagrant status
Current machine states:
nomad-server running (virtualbox)
nomad-client-one running (virtualbox)
nomad-client-two running (virtualbox)
This environment represents multiple VMs. The VMs are all listed
above with their current state. For more information about a specific
VM, run `vagrant status NAME`.
Setting up the cluster
Currently we have to manually bootstrap the client & server agents. The next iteration of this project will use a service like upstart to automatically bootstrap the services...for now, simple vagrant ssh into the nomad-server, and both clients, and run:
nomad agent -config /vagrant/server.hcl on the server.
nomad agent -config /vagrant/client.hcl on the clients.
nomad status will indicate if they're running.
Once we've bootstrapped the cluster, we're ready to start running jobs!
Running your first job
vagrant ssh nomad-server will connect you to the Nomad Server running locally.
$ nomad job init will create a sample job.
nomad job run example.nomad will run the job.
We should see output that looks something like
==> Monitoring evaluation "dbf3b537"
Evaluation triggered by job "example"
Allocation "2ebc1e19" created: node "fe34f75c", group "cache"
Evaluation within deployment: "706c024d"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "dbf3b537" finished with status "complete"
nomad status should return
ID Type Priority Status Submit Date
example service 50 running 2019-03-18T16:06:49Z
The ID is the canonical ID of the job itself.
If/when we wanted to stop the job; nomad stop example would do the trick. However, we want the job to keep running to exercise a few commands.
Running nomad status followed by the job ID will return even more information about the job...including the allocation ID.
nomad status example
An allocation represents a relationship between a "job" and an "executor", that is to say a unit of work. Each allocation is unique. An allocation is the resultant of an evaluation, which takes place every time there is a change in state...i.e. a job is ran. An allocation represents a single execution context between the server(s) and a client with available system resource to process the job.
To get even more granular, nomad alloc status 885958a4 with the alloc ID as a parameter will return the job id, eval id, status, and other important pieces of information. Nomad node id represents the client in which the job was processed. This can be very helpful if say a job keeps failing because it is being dispatched to an unhealthy client.
With alloc status, we can see the system resources that were allocated to the corresponding job.
This is by no means an exhaustive overview of Nomad. I hope that this helps provide a high-level overview.
Additional Resources
https://github.com/hashicorp/nomad-guides https://github.com/oribaldi/nomad-cluster https://github.com/jamesdmorgan/nomad-cluster