Gluster containers!
- 2. Gluster Meetup June 2016, Bangalore, India Mohamed Ashiq Liyazudeen Associate Software Engineer @redhat @ IRC:ashiq on freenode, github: github.com/MohamedAshiqrh, mailto: [email protected] Humble Devassy Chirammal Senior Software Engineer @redhat @ IRC : hchiramm on freenode, github : github.com/humblec, mailto : [email protected] website : humblec.com
- 3. Gluster Meetup June 2016, Bangalore, India Agenda ● GlusterFS ● Docker ● Why containers? ● GlusterFS Container ● Q&A
- 4. Gluster Meetup June 2016, Bangalore, India GlusterFS ● Open-source scale-out distributed file system. ● Aggregates various storage servers over network interconnects into one large parallel network file system. ● User space filesystem with no dedicated metadata server that helps us achieve scalability.
- 5. Gluster Meetup June 2016, Bangalore, India Docker ● Docker offers a simple way to package an application and its runtime dependencies into a single container. ● Docker Container uses shared kernel concept and are much more efficient than hypervisors in system resource terms. ● Docker is both a development tool and a runtime environment.
- 6. Gluster Meetup June 2016, Bangalore, India Container Architecture Source: Docker Inc.
- 7. Gluster Meetup June 2016, Bangalore, India Docker Concepts ● Docker Image is a static specification of what the container should be in runtime, including the application code inside the container and runtime configuration settings. They are Read-only.
- 8. Gluster Meetup June 2016, Bangalore, India Docker Concepts ● Docker container always starts with an image and is considered an instantiation of that image.
- 9. Gluster Meetup June 2016, Bangalore, India Why Container? ● Shared Kernel ● Environment Regardless ● Isolation ● Minimizes System Drift ● Shipping ● Cleaner updates and Manageable state change ● Storage Hyper-converged Environment Support
- 10. Gluster Meetup June 2016, Bangalore, India GlusterFS Container ● Systemd is used in GlusterFS container because gluster needs to run more than one process in the container. ● GlusterFS configuration is made persistent as containers writable layer is not persistent in case of re- spawn. ● Even though GlusterFS can operate on any networking model, we prefer host networking (--net=host) of Docker for better performance on storage network.
- 11. Gluster Meetup June 2016, Bangalore, India Steps Breakdown ● Pull the image ● Prepare the node for GlusterFS configuration persistence(create directories /var/lib/glusterd, /var/log/glusterfs and /etc/glusterfs). ● /var/lib/glusterd – volume management files ● /var/log/glusterfs – gluster log files ● /etc/glusterfs – glusterd management files ● GlusterFS prerequisites(like opening the required port). ● Start GlusterFS container.
- 12. Gluster Meetup June 2016, Bangalore, India GlusterFS Container ● # docker pull gluster/gluster-centos:gluster3u7_centos7 ● # mkdir -p /etc/glusterfs /var/lib/glusterd /var/log/glusterfs ● # modprobe dm_snapshot (in case of snapshot) ● # docker run --name gluster -d -v /etc/glusterfs:/etc/glusterfs:z -v /var/lib/glusterd:/var/lib/glusterd:z -v /var/log/glusterfs:/var/log/glusterfs:z -v /sys/fs/cgroup:/sys/fs/cgroup:ro --net=host --privileged=true -v /dev:/dev gluster/gluster- centos:gluster3u7_centos7
- 13. Gluster Meetup June 2016, Bangalore, India GlusterFS docker run As seen before its a long command, lets check why all the options are required. ● As Gluster needs to run more than one service in a container, It needs systemd. For systemd to run in a container “--privileged=true”. ● Host Networking (--net=host) of Docker for better performance on storage network.
- 14. Gluster Meetup June 2016, Bangalore, India Network options ● Docker Network “none”, adds a container to a container-specific network stack. ● Docker Network ”bridge”, represents the bridge network (default docker0). ● Docker Network ”host”, adds a container on the hosts network stack.
- 15. Gluster Meetup June 2016, Bangalore, India Persistence of Gluster configuration ● -v /etc/glusterfs:/etc/glusterfs:z -v /var/lib/glusterd:/var/lib/glusterd:z -v /var/log/glusterfs:/var/log/glusterfs:z ● The above ensures that the gluster configuration are persistent, In the command. ● As we mention “:z” - SELinux labels are automatically reset to svirt_sandbox_file_t so that the container can interact with the Host directory.
- 16. Gluster Meetup June 2016, Bangalore, India Gluster Peer Probe
- 17. Gluster Meetup June 2016, Bangalore, India Trusted Storage Pool
- 18. Gluster Meetup June 2016, Bangalore, India References ● http://humblec.com/building-glusterfs-in-a-docker-container/ ● http://humblec.com/gluster_containers_in_kubernetes_clust ● http://humblec.com/glusterfs-containers-docker-kubernetes- ● https://hub.docker.com/r/gluster/ ● https://github.com/gluster/docker ● https://github.com/gluster/glusterfs-kubernetes-openshift
- 19. Gluster Meetup June 2016, Bangalore, India Thanks ! Q & A