Millet Porridge

Because HackerNews is all in English and I wanted to read it but didn't want to read English, this website came to be — mainly to see new stories and comments as fast as possible. Later, because translation APIs were really too expensive, after much consideration I decided to translate it myself.

Coming to the company on Monday, I found my usual freerdp couldn't connect to company services. I'll introduce my investigation and solution — not necessarily applicable to all situations, just sharing here.

Since the Kubernetes integration work is basically done, I've been considering adding an independent CI system.

I don't know whether readers of my blog normally use Ansible. As a full-featured ops tool, Ansible can handle most problems I meet at work. Recently my main focus has been Kubernetes. In this blog I describe the problems I solved with Ansible, hoping that when you use Kubernetes in the future you'll consider managing clusters with Ansible.

An earlier blog introduced cron job usage, but during use I found a task scheduled for 8 a.m. executes in the afternoon.

I was recently considering using Nginx's keepalive parameter and originally wanted to explore how Nginx implements keepalive, but ended up mainly studying the reload process — quite a detour. Just the right chance to share this part of the code with everyone.

Following 'How a Kubernetes-based PaaS platform provides ssh services (research)': after I implemented this feature in our production environment, users reported that they disconnect immediately after logging into ssh, with no shell available. This post records the cause and solution, hoping Kubernetes users meeting related problems can avoid them.

Because of the pneumonia outbreak everyone had to work remotely; after the new year the company uniformly updated Windows' remote connection feature, so we had to switch connection tools. Taking this blog as the chance, I'll share a few rdp tools.

This may be the last episode of the Docker series, mainly introducing swarm mode usage. I touch swarm mode rarely at work. For personal projects I deploy applications single-node via swarm; I'll briefly share using docker stack for deployment and upgrades after swarm initialization, and finally share swarm's limitations.

Some services I self-host. The server is a VM on a home computer; unless the machine reboots it generally doesn't go down. If a service has problems, just contact me directly.