Slurm-2: the beginning. Work plan Docker: device, Dockerfile, docker-compose
Yesterday started Slurm-2, intensively on Kuburnetes. First performance: Pavel Selivanov tells how the participants of the Slurm will master Cubernets, and then give a lecture on Docker. In the chat, they complained that "Docker in an hour is far beyond the limits", to which Pavel answered: "Our task is not to study Docker, but to agree on points of view on Docker so that there is no ambiguity in future work."
RedSlerm went smoothly, and we already relaxed. Started Slurm-2. At 13:21 the broadcast fell off, which went through a dedicated channel from a local provider. After 5 minutes, the broadcast was transferred to the backup channel (4G from Tele2). After another 3 minutes Tele2 fell. After a couple of minutes, the server Selectel stopped pinging , where we are practicing (thanks again to Selectel for the power).
Passed information that there was an accident on the main channel. We followed the lunch break, and after an hour the problems with communication disappeared, the accident was eliminated.
Colleagues complain that I am writing about problems, not about achievements. Achievements are simple: by the beginning of the second day, 77 participants created their own cluster, and another 10 hastily complete it.
There are still a couple of places on MegaSlurm , advanced intensively for those who are already working with Cuvernethes.