On the second day of Slurm, the participants were already feeling slightly tired. It’s good that Alexey Gladkikh, a specialist in conferences and mass gatherings like the Battle of Kulikovo and the “Bloody Wedding” in Game of Thrones, helped us. In both cases, the movement became active, no one stayed too long, the blood did not stagnate.
On the second day of Slurm, Pavel Selivanov raised the topics of advanced Kubernetes abstractions: DaemonSet, StatefulSet, RBAC, Job, CronJob, Pod Scheduling, InitContainer.
After the coffee break, Marcel Ibraev came in great and with him DNS in the cluster, and at the same time the publication of services and applications - NodePort vs LoadBalancer vs Ingress.
“But the Tsar is not real!”
Here you try to quickly grab your right ear with your left hand, and your nose with your right hand. After a second, clap your hands. And immediately grab your left ear with your right hand, and your nose with your left hand. And repeat. After that, raising ceph to work with Kubernetes will be just bullshit.
Then Pavel Selivanov grabbed the audience with a tenacious hand. He started by working with Helm, writing his own charts.
And then that same Ceph appeared under the command of Sergei Bondarev: installation in the “do as I do” mode, connecting volumes to pods and of course sc, pvc, pv - practice using Ceph as an example.
There were timid cries in the hall that there were problems. Several hands went up. Sergei Bondarev replied: “Now we’ll come and solve the problem.” Then a voice came from the depths of the hall: “No person, no problem.” The number of hands has sharply decreased. Effective bug management.
- Are you the one who said you have a problem? - No, no, no, I’m fine and everything works!
The technical support demonstrated miracles of professionalism. Not a single question, not a single problem passed them by.
At such moments, each participant understands that he will not be left alone. Moreover, the third day of Slurm is ahead - the most difficult, most interesting and invigorating. Practical work, dockerizing the application and launching it into a cluster.
Source: habr.com