Testing the nomad orchestrator by running some things
- virtualbox
- go
Single nomad agent on single host
# run vagrant
$ vagrant up && vagrant ssh
# start nomad
$ sudo nomad agent -dev | grep ERROR
# run the server job in /vagrant
$ nomad job run server.nomad
# inspect the job
$ nomad status server
# cleanup when done
$ vagrant destroy
Nomad cluster on single host
$ mkdir -p /tmp/{server1,client1,client2}
$ cd /vagrant
$ nohup nomad agent -config conf/server.hcl | grep ERROR > log/server.log &
$ nohup sudo nomad agent -config conf/client1.hcl | grep ERROR > log/client1.log &
$ nohup sudo nomad agent -config conf/client2.hcl | grep ERROR > log/client2.log &
$ tail -f log/*
# 1. Build-, and package the binaries on your host (requires go)
$ ./build.sh
# 2. commit and tag changes and make a new release on github
# 3. update source and checksum in *.nomad
- run arbitrary binary w exec driver
- scale out using dynamic ports
- use artifact to fetch remote resource
- nomad agent cluster
- inspect cgroups for binary jobs
- try nomad alloc signal|exec
- disaster recovery
- try bridge network
- register service w consul
- service discovery w consul
- add checksum to build step