Ahh I see. For the record, I like Linux.

I go back and forth. I don’t remember what I said here.

no I was making a joke that was apparently not well received here. Aside from rust in the kernel, I didn’t think this was one of those sacred-cow communities.

Sorry I’m new to lemmy, is there a preferred service we use here?

This had nothing to do with this distro. Its the filesystem mount timing out during an filesystem format upgrade that tried to happen during the mount by systemd that did this.

ext4 doesnt do tiered storage. I could make an LVM and have it pool things into one storage volume but I wanted to learn this. bcachefs is simpler and cleaner, but it’s still young and very volatile. Also, ext4 does have bugs still, even today.

I was just thinking about taking a more recent backup when I ran this update and thought against it. I’m going to have to verify this thing against the old backup somehow tomorrow. :/

I get that, but its a cool new toy to play with, and the fsck code is very effective so long as it doesnt run out of memory.

Here’s the thing. Bcachefs is still under development, and Kent is really careful with his filesystem. This happens to me every now and then if I havent rebooted in a long time or theres a kernel update with filesystem changes it doesn’t like. The trick is to skip the userspace fsck code and pass the -k flag so it uses the kernel fsck code which is much farther along. I’ve never lost anything on this filesystem and its messed up in lots of bizzare ways.

I’m still skeptical. The licensing model is certainly friendlier and I think that’s why more people are willing to give this a chance and put dev time into it. But the cost is still high and performance is trash.

I’m most interested in seeing what that team that splintered off intel to go all in on RISC-V come up with.

I put it back on heimdallnet and tried downloading fedora linux. Status is firewalled and theres no DHT but the download works and there are no errors in the log. I looked at an i2p torrent I was playing with too and its showing the trackers are non responsive now, but theres no error in the logs.

I think I am not understanding how PublishPort= works in quadlets. That was enough to have it pass those ports to the container before I added the network= string in the quadlet. What is different now? Its like it passes the webUI port and not the others

No, that would be manjaro.

The Qualcomm stuff is actually coming along, as is Apple. They’re on pace with pretty much any non-raspberry pi SBC.

u-boot loads off the SD card because there is nowhere else to put it on this board. I want to put alpine there too. Alpine and u-boot can be on the same card but alpine wont finish loading unless it is being loaded from the USB port.

I’m using the R4S, it’s a different model with a different SoC. I know there used to be a weird bug on this board regarding SD cards that was supposedly fixed in 2022 (https://kohlschuetter.github.io/blog/posts/2022/10/28/linux-nanopi-r4s/)

That’s why I’m running alpine. It runs in ram and only writes to the SD card when I run “lbu commit.”

As an experiment, I wrote u-boot to a different blank SD card and put the supposedly bad SD card with alpine on it into an USB card reader and connected that to the nanopi. Sure enough, the nanopi loaded u-boot on the sd card, then loaded alpine just fine off the USB card reader.

The card is fine. It just wont run alpine linux off the built in card reader.

I’m of the opinion that a full rewrite in rust will eventually happen, but they need to be cautious and not risk alienating developers ala windows mobile so right now it’s still done in pieces. I’m also aware that many of the devs who sharpened their teeth on the kernel C code like it as it is, resist all change, and this causes lots of arguments.

Looking at that link, I’m not liking the MPL.

Rewrite the entire kernel exclusively in rust!

-hehehe-