A couple of days ago I submitted the internal-review version of my thesis. The graduation project that had been going for about five months is basically done. From not being able to tell Docker and k8s apart to now preparing for the CKA and CKAD, my understanding of cloud native keeps getting deeper. As for the graduation project I actually made, my verdict is that it's completely unsatisfying. The topic I picked was too big, so no matter what I did it never felt like enough, never felt good enough. At first I always wanted to do a lot, but as the project went on I realized what I could actually do was limited, and then I just had to keep compromising. That reminded me of a course project in my freshman year — it was the same thing, because I didn't understand the project well enough up front. The mistake I made as a freshman, I'm still making as a senior. It feels like standing in place. Sometimes thinking about that made me pretty depressed. Looking more carefully, it's still because I don't have enough project experience, and I haven't absorbed previous lessons enough.
One night I couldn't sleep and finally figured it out: it's normal that what you make isn't perfect. It's the most normal thing in the world. Precisely because you can see the imperfect parts, there are places to improve. Before, whenever I hadn't done well enough I'd rush on to the next task, so I never accumulated enough in any field. That's not right. If it isn't perfect, you should take the lesson and do it again (this is where "Again" from Your Lie in April should start playing: アゲイン). Doing it again will definitely be better than last time, and you'll learn more (provided you've decided to go far in this field — then you should keep thinking more deeply about a problem, and not feel like you already know enough. The last ten li of a hundred-li journey is half the way).
Take the observability problem I ran into on this graduation project: I may have done a lot of work, but there were no concrete metrics I could actually show people. For k8s I only used the dashboard at first, and for packet capture I only used wireshark. Since this experiment didn't observe enough data, next time I should use more professional tools. For packet capture I could consider tcpdump and the like (I haven't looked into it deeply so I'm not totally sure), and for cloud-native monitoring I could use Prometheus and so on.
But the more I learn about cloud native, the more I feel I should still go toward the lower layers first. A simple example: this graduation project used KubeEdge, and the problem I hit in experiments was that when a pod on an edge node wasn't working properly, you couldn't tell from the cloud what had happened. At first I thought it was my proxy config. Then I looked at the related OSPP 2024 topic and realized — oh, they just haven't implemented it yet... Your work sits on top of other people's work, so you get constrained to some degree. Another reason is that top-level applications combine with the technical principles underneath. Because I'm still too noob I won't expand on that (flee). So in grad school I still can't be restless. I should seriously lay the foundation.
Anyway, the graduation project is at a pause. Then I went to a friend's birthday gathering, sang at ktv, and was a bit stiff — felt like people were looking after me (very grateful, and I hope one day I can grow to the point where I don't need looking after). One thing is beyond doubt: everyone there is a really good person, so I actually like that kind of atmosphere. I remembered the first time I went to ktv was with my sister. The only song I sang was the theme of Xuan-Yuan Sword: Scar of Sky, "一吻天荒" (I was super into that show back then, lol).
Being willing to show up in a setting I'm not familiar with — at the end of the day, I think social skills can be learned. As you experience more things, part of a person does change. People have too many sides. In different settings you show different parts of yourself. Sometimes it isn't even pretending; it's just that at that time and in that place, the other sides don't come out easily. So binary classifications like MBTI feel like they limit your behavior and growth. "I'm I* so I might do this and that" — it's not like that. What I do will surprise you.
Then the next day I met up IRL with a friend I'd known for more than half a year on the Fediverse. Finally I could talk in real life about things that happen on the internet. We talked for two hours and it was a lot of fun. We roasted Hideaki Anno together. My friend said getting close to Gundam makes you unlucky, which only made me more curious about Gundam, you know? I also heard about academia from them — learned a lot! What I like most is still talking face to face.
After submitting the thesis, I started setting up my own misskey instance. The domain I chose is neuromansser.tech, neuromancer->neuromansser. Right now it looks roughly like this.
vnil.de, where I used to live, is great. The admin is kind and enthusiastic, and on vnil.de hanging out with friends also felt comfortable and relaxed. Thanks to the friends on vnil.de. If you're still looking for a home and hunting for an instance, maybe give vnil.de a thought~ I'm an adult now. Time to take care of my own server (hey!).
After the instance was more or less up, I also migrated my blog. I just finished the migration last night. There isn't too much extra stuff; overall it's pretty clean (of course not as minimal as the admin's, and js included). I'm reasonably satisfied.
That's my recent life ^ ^ see you~