This playbook deploys a full Apache Mesos stack. The access to mesos (agent and master) need credentials. The default one is "marathon:marathon".
- CentOS 7 (deprecated) or Debian/Ubuntu:[focal|jammy]
- Ansible Galaxy collection: community.general
- Ansible version:
$ ansible --version
ansible [core 2.14.2]
python version = 3.10.9 (main, Dec 6 2022, 18:44:57)
jinja version = 3.1.2
libyaml = True
- These playbook will install cronjobs to cleanup unused docker images, containers and local volumes. To besure that local volumes will not be deleted, add a tag called "mesos" as label. As example:
docker volume create --label tag=mesos volumename- These playbook will not restart services. If you apply new configuration and want to restart the service, use the tag "restart"
ansible-playbook -i ../inventory/inventory/mesos plays/server-config.yamlansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --skip-tags installansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags updateansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags restartansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags dnsansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags workeransible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags manageransible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags pluginansible-playbook -i ../inventory/inventory/mesos plays/server-config.yaml --tags ssl| Software version | Role | Install type |
|---|---|---|
| Mesos 1.11.0-0.3.0 | Mesos Masters | RPM/DEB |
| Zookeeper 3.8.1 | Zookeeper cluster | RPM/DEB |
| Mesos-DNS 0.9.0 | Service Discovery for Mesos Tasks | RPM/DEB |
| Software version | Role | Install type |
|---|---|---|
| Mesos 1.11.0-0.3.0 | Mesos Agent | RPM/DEB |
| Docker CE latest | Docker engine | RPM/DEB |
| DNSMasq 2 | Container DNS | RPM/DEB |
For persistent container storage, we recomend to use our ansible-docker-volume playbook to deploy S3 docker-plugins as systemd service.
If weave cannot ping between containers at different hosts, you have to change the following configuration:
sysctl net.bridge.bridge-nf-call-iptables=0Do not forget to make it persistent.
If you get the error message "To many open files" but file-max looks fine, then it could be helpful to resize max user instances:
sysctl fs.inotify.max_user_instances=8192