Cancel
Start searching
This search is based on elasticsearch and can look through several thousand pages in miliseconds.
Learn more
This search is based on elasticsearch and can look through several thousand pages in miliseconds.
Learn more
The DevOps Tool Bazaar at Blue Yonder on February 20, 2018, was a refreshing change from the otherwise single-topic DevOps meetups. The diverse range of topics not only sparked discussion but also provided insight into the speakers’ respective fields and the challenges they face.
Jürgen Hermann from 1&1 focuses on packaging Python software into platform-native packages using tools such as dh-virtualenv and fpm. Using Sentry—a system for centralized logging and management of application exceptions— as an example, he explains how these tools work and what pitfalls to watch out for. You can find more details in his slides.
In his Meetup announcement, Sebastian Neubauer of Blue Yonder already points out that he intends to deliberately provoke with his talk—something the title of his talk already hints at: “Throw Away All Your Wrapper Bash Scripts!” By putting this challenge to the test himself, he also asks the audience whether most wrapper scripts aren’t actually unnecessary.
Leaving aside the often poor quality and maintainability of the scripts, Sebastian’s central argument is that the supposedly noble goal of shielding users from complexity should be viewed negatively. In his opinion, wrapper scripts obscure the user’s view of the necessary fundamental ops know-how. Therefore, he argues, it is of greater value to write good documentation that enables a deeper understanding of the underlying technologies in the operations field.
To explain: Wrapper scripts are those scripts—rarely short and often written in Bash—that are designed to make it easier for users to work with complex technology. Everyone knows them; many have written them themselves; few want to maintain them.
Using the company’s own blog —which is built on the Hugo blogging tool—Johannes Graf von Synyx demonstrates how GitLab CI and containerized test instances can drastically simplify blogging for employees and thereby boost their motivation.
Hugo generates static pages from Markdown files, which are then uploaded to the target system. This lays the foundation for fast and easy blogging. To give employees a preview of the blog post within the context of the blog site, GitLab CI uses a branch to generate a complete, fully functional preview instance of the entire blog in a Docker container. Within this instance, functionality and content can be checked and corrected as needed. After a successful merge into the master branch, the preview instance is cleaned up.
Michael Ströder specializes in the important but often neglected topic of security. His fundamental principle for improving security is to change keys frequently and increase the number of security layers. Unfortunately, however, when it comes to public-private key pairs, there is no inherent technical solution to enforce regular key rotation. Employer policies are a rather cumbersome and impractical tool for guaranteeing higher security. Michael is therefore working on a method to make practical use of the SSH certificates already implemented in OpenSSL. The basic idea is to have a secure instance based on the public-private key pair generate short-lived certificates that are validated on the target system.
No one could avoid hearing about Meltdown and Spectre in recent weeks. Most articles on relevant IT news sites often only mentioned that the cause lies in modern processor architectures and that kernel programmers for various operating systems must now find ways to fix the vulnerabilities created by pipelining, branch prediction, and the like. Anyone who really wants to understand Meltdown must delve deeper into this thoroughly complex subject. Patrick M. Hausen from punkt.de has done just that and, in his presentation, uses clear examples in C to demonstrate how processor behavior can be exploited through timing measurements, caching, and shadow registers.
