DEV Community: Rondo

Record of Site Issues #1

Rondo — Fri, 19 Jun 2026 01:42:21 +0000

Environment And Situation

Control room of an apartment
Number of installed product : 1 (PC-based NVR, dual-LAN supported)
Remote support : O

Reported Issue

When the user connects LAN cables to both NIC on the product(one NIC is connected to a private network and the other one is connected to internet)simultaneously, internet is unreachable.
Monitor connection keeps disconnected after system booted. It happens in random minutes(mostly between 1~5) - The colleague in the site told me he's not sure whether it's just about monitor disconnection or the entire system down.

Diagnosis And Assumptions

I asked to change the monitor but the monitor issue kept happening. -> The monitor was innocent.
If it was the case that the entire system was down, it might be because of kernel panic. -> I needed to check kernel messages using dmesg and journalctl
Every time the monitor issue happened, fans on the product started running faster.
I was debugging via Anydesk, but when the monitor issue happened, Anydesk remote screen stopped.
This site was where I'd already resolved other issues. But back then, there was no kind of this issue.

-> Suspicion on hardware(GPU or CPU heat problem, power supply issue, etc.)

-> Suspicion on environment(Based on the fact that the monitor issue was not there in the site back in the day) - The user might have connected other products on the same power strip

-> Suspicion on the product itself(hardware related malfunction caused by physical shock).

First Solution

-> Recall the product and check it in company.
-> After check the product in the company, send it to hardware vendor.

Second Diagnosis After Recall the Product

The monitor issue didn't appear again.
I checked the recording data -> !!Recording data was well!! == The system didn't go down entirely. It was almost surely about only VGA issue. -> The user might have connected other products on the same power strip or been about another electronic issue.

Final Conclusion

VGA malfunction caused by electronic issue(site dependent)

Solution

After all, the issue didn't happen again.

So we decided to let the product run for a day and if there's no other issue then get it back to the site and check the overall environment(with installation technicians)

About Network Issue

-> Even though I couldn't check the network issue properly, I was pretty sure that the issue was related to routing metric(similar issues happened multiple times in other sites)

Keep This in Mind

When system related issue happens, check recording data first.
Even if an electronic issue occurs, it does not necessarily affect the entire system. In this case, only VGA output appeared to be affected.

Additional Assumption

VGA output might have been affected first because of its sensitivity to power-related issues(not verified)

Software Experience - for Who?

Rondo — Sun, 31 May 2026 14:31:54 +0000

Context

I've been working in physical security field and in charge of a PC-based NVR(Network Video Recorder) product.

The Hidden Stakeholders

Last few weeks, I've dealt with a bunch of site issues.

Sometimes I resolved the issue remotely on my own.
Sometimes I supported the CS team in site to treat issues when I couldn't connect to the site(isolated network)
And sometimes I helped installation technicians.

Over the course of these experiences, I realized something.
There were more people related to my product than I used to think.
And each of them priortizes different things.

Stakeholder	What They Care About
Installation technicians	Easy setup of network, cameras(codec, resolution, bitrate, etc.) and other initial configurations
CS team	Access to logs, settings, and troubleshooting information to diagnose and resolve site issues
Customers	A simple, reliable, and easy-to-use experience, especially for monitoring and playback

What I Missed

Among them, I usually focused on improving only customer's experience. Why? user experience is always the top prior. There's no doubt.

And it was my mistake.
I should've focused on installation technicians and CS team as well.

The idea came from a simple thought - why were there so many issues that I, the engineer, must've dealt with?

The answer was simple - the software was not good enough for them(installers and CS) to resolve the issues themselves.

The software might satisfy customer experience, but not the other stakeholders.

CS team : If the software does not support enough features for the to resolve the issue, they will end up asking me to deal with the issue.
Installation technicians : If the software is inconvenient to set up initial settings(network, camera, etc.) they will also end up asking me about that. Even worse, they may accidentally set the product wrong.

Initial installation is especially important. Customers usually don't change the initial settings and don't want to touch that. It's related to installation technicians' experience after all.

So simply, improving their experiences is also for myself.

Finding Common Ground

At this point, I had a question.

How can I satisfy all of them? That's too much.

Of course it may sound tiring. And it might not be able to satisfy all of stakeholders.

Here's a trick : before you get down to your tasks, find a common point first. For exmaple, in my case, easy network setting is important for both the CS and installation technicians, so I priortized it first.

Since we're always pressed for time, negotiating with yourself and finding a common point will help you satisfy your stakeholders.

Conclusion

I used to think improving customer experience was enough.

Now I think software experience should include everyone who interacts with the product, even if they are not the end users. Because when their experience improves, mine improves too.

Thank you for reading this article.

Tools should Not Be the Goal

Rondo — Mon, 18 May 2026 14:44:29 +0000

Lately, I've been learning about agents, especially OpenCode.

It was really exciting expericence. Since beforehand, I'd only used a single agent, like Gemini-cli. Orchestrating different models for the same goal was a complete shock.

I made few simple tools using multi agents, such as todo-list web app, schedule manager, and so on, just for fun.

Speaking of that, I've spent most of my free time searching and learning about AI things. I've read articles like 'which model is better?', 'Using LM Studio in Ubuntu' and watched Youtube videos like 'how to set a strict harness for your agent'.

And that was it.
I realized I was collecting tools, not solving problems.

Now I got a sense of orchestrating agents simultaneously. Now I know how to setup LM Studio in Ubuntu. I can make some tools using such things.

But I still don't know what's the real problem that people have and I should resolve.

That's what I recently realized-I was focusing on AI related things too much, meaning I was focusing on making tools when I didn't know what's the problem that the tool was meant to resolve.

It was complete nonsense.

I think tools should not be the goal. The true goal should be solving problems regardless of where they're. They can be in your workflow, your daily life, and so on.

That's why I try to focus on inconvenience in my daily life these days.

For example, when I need to make a document about a product(You know, documentation is borrrring!) I am working on, I think 'Can't I automate this documentation process?' before I start the task. Even though in most case, I fail to find the solution, I think the process itself is worth.

I am not claiming that there's no need to learn about new AI agents and such things.

There are a lot of great AI agents out there and it's really exciting to just learn them itself!

What I'd like to say is that we sometimes need to reflect and ask ourselves 'why am I learning this?', 'what should I make with this?'.

Ubuntu NetworkManager Fooled Me with Routing Metrics

Rondo — Wed, 06 May 2026 15:19:35 +0000

Recently, I was testing a PC-based NVR that supports Dual-LAN and uses Ubuntu.

The testing environment was very simple: One NIC was connected to an isolated network that has a switch connected to a lot of cameras. And the other NIC was connected to internet.

And not surprisingly, I could clearly see the camera screens via embedded VMS viewer. Everything seemed working well.

But I soon realized that I couldn't use internet at all. That didn't make any sense for me since I used internet about just few minutes ago.

Since connecting to internet is very important for the product(It allows us debug remotely when issues happen in sites.), I had to deal with this issue.

First, I tried the simplest method : I disabled network interface that was connected to internet and enabled it. Result? I could access internet again. But only for about 4 minutes.

I repeated the same treatment several times, but the results were the same. I could access internet only for a few minutes. Every single time. I completely had no idea. I thought it was because of hardware, but my guess was completely wrong.

The key was routing metrics. I knew the concept, but I didn't know that Ubuntu's NetworkManager can change the metrics on its own. For some reason, NetworkManager kept assigning a lower metric to the isolated network than the internet interface.
(For context: a lower metric means higher priority.)

So I googled a command to control the metrics by hand. I could do this using nmcli

sudo nmcli connection modify 'connection_name' ipv4.route-metric (value)

After I fixed the metric value 999(high enough value) for the isolated net, I could finally connect to internet without disconnection.

To release this product, I needed to make sure that the product can always be connected to internet. My solution was to attach stickers to let users know which NIC is for internet and which is not and add some code to execute the command above.(It was not easy since we're using C and connection_name is necessary.)

My point is that if you're having trouble with dual-LAN hardware over Ubuntu, check your routing metrics before blaming the hardware.
Start by checking : ip route show

Nobody Wanted to Write Release Note. So I Made a Git-Log Searching Tool Using AI Agent at Work.

Rondo — Mon, 13 Apr 2026 14:00:10 +0000

TL;DR

Inefficient workflow
QA had trouble with tracking issues(They didn't know what changed)
Developers didn't update release note for QA. We were busy dealing with our own tasks. (I know, it was just an excuse.)
Sales kept asking the same question again and again
So I made a git log searching tool(webpage).

Every firmware release triggered the same cycle:
“What changed?”
“Which features are included?”
“Does this affect our product? At which OEM?”

The reason why I couldn't do this manually

I was assigned a task to deal with the problem. But:

Too short deadline
Other ongoing tasks

I had to finish the task as soon as possible.

I don't usually use AI agents at work to deal with my tasks, unless the tasks that meet three personal criteria:
1. The task must be clear and pinpointed.

2. The task must be irrelevant to our base code. (I don't want AI agents mess up with the huge code base.)

3. The task must be what I can do on my own(So that I can fix problem later on.)
Fortunately, the task I was given met all the criteria.
So I thought it was the best time to try an AI agent at work.

The Tool to Make

A webpage that allows other teams to browse and search git logs(Of course, they're pre-filtered.)

Since I know a bit about Django, I decided to use that.
The setup process looks like this:

Clone repository on the server where the tool is running.
Save the path of the repository using django-admin.
Get git logs based on the saved path.

How I Built It

I decided to use Gemini CLI with iterative loop:

Ask Gemini-CLI to implement an exact feature.
Accept the output.
Run the code and check issues.
If any issues found, then let Gemini-CLI know the issues and let it fix the issues.
If Gemini-CLI keeps failing in finding the issues, then I fix them.

Example prompt:

Add a dropbox to limit shown logs on each page. The range is from 10 to 100 and gap is 10. The dropbox must be located next to filter box.

I also asked questions below multiple times during the process:

'Is the project really going well?'
'Are there any other ways to make it better?'

These kinds of meta-questions helped the agent stay on the right track.

One little trick that (might have) helped a lot:
I asked the AI to maintain a task log file to track progress and check it after every single task. Already know this trick? Sorry.

The Result

The final tool:

Select repository and branch
Filter by target product
Search commit messages (version, tags, etc.)
Pagination

Built with:

Django (and Django template)
SQLite
Local repositories + git log

Effects:

QA and Sales stopped asking repetitive questions.
Satisfied boss
Happy developers

It's not perfect.
Repositories have to be updated manually (git pull), but believe it or not, it's intended(sometimes you need to intentionally hide information.)

What I Learned

Using an AI agent didn’t remove the need to keep thinking.
Actually, I think it only made one skill more important:

The ability to define problems clearly and break them into small steps.

AI can generate code fast, of course.
But it won't decide what should be built and why.

Closing

This was a small tool, but a meaningful experiment.
I don’t think AI will replace developers anytime soon.
But that's only for those who won't stop thinking, I guess.