Outside the window is the classic positive autumn weather typical of St. Petersburg, the conference room at Selectel is warm, with coffee, Coca-Cola, and a sense of almost summer. In the outside world, it is September 5, 2019, and we are on the second day of the Slurm DevOps.
On the first day of the intensive, we covered the simplest topics: Git and CI/CD. On the second day, we prepared Infrastructure as Code and infrastructure testing for the participants—these are multifaceted topics with many nuances and practical assignments.
The speakers on the second day of Slurm were Alexey Stepanenko, an engineer from Selectel's cloud platform department, Nikolai Mesropyan, an administrator at Southbridge, and Vladimir Guryanov, an engineer/team lead at Southbridge.
Alexey Stepanenko is an engineer at Selectel, which not only warmly welcomed us but also provided the venue and server resources. Alexey works on infrastructure tasks related to servicing the OpenStack cloud, specifically monitoring, CI/CD, and configuration management.
Nikolai Mesropyan and Vladimir Guryanov are employees of Southbridge. Vladimir Guryanov has been responsible for monitoring systems since 2015, also serving as the head of the operational management at Komita JSC, and he is the initiator of Komita JSC's transition to a DevOps approach.
By 10:00, participants gradually gathered in the conference room. After the pace of the first day of Slurm DevOps, they appeared somewhat pensive and worn out. No one promised it would be easy. The intensive is designed to load as much practical information into your head as possible in three days at a breakneck pace. It is impossible to master Slurm in three days. The intensive itself is just the beginning of the work. Later, you will need to review the recordings and practice.
On the first day, Alexey Stepanenko touched upon the subtopic of 'IaC: treating infrastructure as code.' On the second day of Slurm, he began with cloud providers as infrastructure suppliers.
— I'll turn the calendar over...
… and it's September third again!
Then Alexey discussed system initialization tools, image building (packer). Following that, he moved on to 'heartfelt stories about Terraform,' , during the topic 'IaC using Terraform.'
And he concluded his talk with questions about configuration storage, collaboration, and application automation.
During the break, we overcame the natural shyness and modesty of the developers and invited everyone to take photos.
Slurm inspires! Sergey Bondarev and Pavel Selivanov keep Dmitry Simonov grounded.
After lunch and a direct infusion of caffeine into the curious minds of the developers, it was time for IaC with an example using Ansible.
Nikolai Mesropyan first addressed the question: 'Why Ansible, in fact?'. The speaker demonstrated creating a small infrastructure using Terraform and working with ansible-vault. Then Nikolai discussed the components of Ansible: playbooks, inventory, roles. He tested the concept of idempotence on cats.
During the presentation, no kittens were harmed.
Nikolai Mesropyan concluded his presentation by assembling and deploying the xpaste application, demonstrating a rolling update and giving a practical assignment: 'Transforming a standalone PostgreSQL into a fault-tolerant cluster.'
— Did we complete the task? The scariest part is behind us. Now hug each other...
In the evening, Vladimir Guryanov spoke on the topic 'Infrastructure Testing'. The participants' energy was already waning. The limit, when information starts slipping between the ears without engaging the brain, was getting closer. Participants got into linguistic curiosities:
Alex GSTC, [Sep 5, 2019, 19:56:29]:
Does it only sound off to me or does it to anyone else?
* repositorY
* defAult
* volt
* ansiblE
* redis
* nAgius
* promEtheus
Nikolay Mesropyan, [Sep 5, 2019, 19:57:09]:
> ansiblE
How's the stress on the soft sign pronounced?)
Lucky SB, [Sep 5, 2019, 19:57:13]:
Also, we were asked why Pavel says scEduler, but I say ansible.Therefore, it was decided to move the infrastructure monitoring with Prometheus to the next, third, and final day of Slurm.
Vladimir Guryanov limited himself to testing and continuous integration with Molecule and GitLab CI, as well as using Vagrant.
The first day of Slurm went without technical issues. It turned out that they were waiting for us on the second day. In the Slurm chat, participants raised the question of recording the event:
Nik Grebnev, [Sep 4, 2019, 21:53:09]: Good evening. As I understand, the recording of today’s event has appeared. But the quality is terrible - 720 lines. How can I get 1080? Everything is so blurry - I thought the stream was poor, but the recording would be 1080. And it turns out that it's still 720 in the recording, which is very disappointing.
Dmitriy Miroshnichenko, [Sep 4, 2019, 21:56:52]: +1, I want to see 1080 in my personal account, since you can't stream more than 720.
Akbarhon Amirkhonov, [Sep 4, 2019, 22:24:40]: There are platform limitations. Unfortunately, it won’t be possible to record in 1080.
Nik Grebnev, [Sep 4, 2019, 22:29:23]: What a pity! That even the phone can't be adjusted for this..... We will suffer with 720p.
Maksim Vasilev, [Sep 5, 2019, 10:19:35]: #support Watching on the speaker console in the recording is just painful. It's very hard to see.We stream through the Facecast platform. Previously, 720 lines were enough for all four Slёrms for participants to normally perceive the video stream. The work laptops of the Slёrm speakers were set up identically and adapted specifically for streaming at 720p. On the second day of the DevOps Slёrm, it turned out that the laptops of the invited speakers were configured differently. And resolving this quickly was not possible — increasing the font size in the consoles broke the tables of the speakers. The small font in the 720p stream turned out to be poorly readable.
The Slёrm is filmed in 1080; it is compressed to 720 by Facecast. So we had the originals in 1080. After a day of communication with Facecast support and accounting, we have a new tariff plan that allows streaming in 1080 and re-uploading the first two days in 1080. So the recordings by viewer requests will also be in 1080.
On the third day of the DevOps Slёrm, they left the most interesting topic: SRE. With the performances of two powerful speakers — Eduard Medvedev, CTO at Tungsten Labs, and Ivan Kruglov, Principal Developer at Booking.com.
In the end, we realized that we made the right decision by moving some topics to the next day — some participants could not keep up with the pace of information delivery and the series of practical tasks, even Windows couldn’t handle it.
Arthur Gordienko, [Sep 5, 2019, 21:45:02] Just came back from the Slёrm.... Does anyone have a recovery flash drive or Windows 10 on it?P.S. To be honest, we hide what is actually happening at the DevOps Slёrm.
You think everything happens like this:
But in reality:
Artyom Galonskiy performs Beethoven's Sonata No. 8 (Pathétique) in C minor. Sergey Bondarev dictates the notes.
Source: habr.com
