Yeah, vim buffers are much better my dude, get that full vim IDE going.
2 Likes
Just gotta get the muscle memory sorted out for splits and zoom
1 Like
I know Iāve mentioned it before, but itās crazy to me that my Intel Macbook Air is driving 2 4k displays, plus its own display, running 8 vms and I guess roughly 200 tmux panels plus a few dozen browser tabs. I can feel the load, but itās still very usable.
4 Likes
Feel the power of old hardware!
1 Like
Blasphemy
3 Likes
lol, Iāll get there, but right now I hate it a lot
2 Likes
A bit late: Proxmox also uses OVMF / KVM and MacOS can be installed. But it needs certain Arguments / arg values, a modified OpenCore boot loader and certain HW settings for the VM. The second hurdle is vfio especially passing through HW USB controller and the type of GPU.
This blog was a great help getting a Mac OS VM up and running:
https://www.nicksherlock.com/2021/10/installing-macos-12-monterey-on-proxmox-7/
1 Like
Guess you found out why Emacs users do what Emacs users do 
Havenāt seen something this bat shit crazy for a while. Reminds me of all the utopian transhuman hype around 2010.
I guess I should cross post this in the newsā¦
2 Likes
I read layered city, I see class segregation.
You are too poor to access the top layers to breath fresh air, go back down to the working class layer. The fact that it will be bound by Islamic law makes it even less interesting.
After reading this article I want to return to a small village from the city outskirts Iām currently residing in.
Maybe, but maybe, we could learn something from them trying. I need to see the promised design and technology first.
4 Likes
I have a Supermicro X11 board that has an NVME M.2 slot so I figured Iād use a 32GB Optane, but I cannot figure out how to convince the bios to boot from it. The system can see it, but itās never a boot option. I tried dual and UEFI only boot options and enabled EFI on the M.2 slot in the PCIe config, but no dice.
1 Like
Some old systems donāt have support for booting from m.2 slots. Actually, I had a few HP Proliant micro-servers that would only boot from an internal SD card slot. The easiest thing to do in those cases is to put the /boot partition on a SD or USB and put the root on the m.2.
The RPi 4 has added support for booting from USB at some point, but I had been having my rootfs on a usb m.2 before they added support, so I have my /boot on a microSD and once it boots, it loads the rootfs on the SSD. I learned this trick from the experience with the proliants.
The random MAC address feature in OpenBSD does some wacky things. I have a USB ethernet adapter and it simply breaks all network connectivity in that case. On at least one realtek NIC, the actual NIC adopts the random MAC (it appears in dmesg on reboots). This is only remedied after physically removing power from the machine.
On Intel NICs, it behaves as Iād expect. dmesg will always show the hardware address and then it will adopt a random one once the OS starts.
Something to look out for I guess. I havenāt done much testing with the Linux equivalent features in networkd and NetworkManager outside of a VM yet so weāll see what happens thereā¦
3 Likes
I attempted to replace a fraying cable on the 11-year old Apple Thunderbolt Display that Mrs. O uses for wfh. I successfully replaced the cable but the display wouldnāt power on. When I used the suction cup from my iFixit kit to pull the glass off again, it broke in half. RIP. Threw the whole thing out as it was no longer worth dealing with.
So I sprung for an Apple Studio Display. I know itās ludicrously over-priced, but the nano-texture glass is kind of amazing. There were serious glare issues before. Now, no issue at all. Mrs. O is in video calls roughly 50% of the day on average, so the fancy web cam thatās basically an iPhone plus the anti-glare were the selling points. It also looks nice. Iām just hoping it lasts as long as its predecessor.
3 Likes
I think one of the worst parts of Linux is network configuration. There is a lot of unnecessary abstraction. Iāve used various Linux distros for years and only relatively recently untangled the various network stacks.
Debian
By default, Debian uses ifupdown which I believe is specific to Debian distros and is gradually being phased out. Network Manager has a plugin for it, so you can use both. In that case, Network Manager is an abstraction on top of ifupdown. You can use nmcli or edit ifupdown config directly.
On Debian, I now install Network Manager, remove the ifupdown plugin, configure network and then uninstall ifupdown.
Natively, Network Manager uses its own key-file plugin which is driven by file-based configuration in /etc/NetworkManager/system-connections.
Ubuntu
By default, Ubuntu uses netplan as an abstraction layer on top of systemd-networkd. Netplan can also configure Network manager. I havenāt tried it, but I think you could have a netplan > NetworkManager > ifupdown network stack on Ubuntu (which illustrates how ridiculous it all is).
On Ubuntu, I configure systemd-networkd directly in /etc/systemd/network.
RHEL
RHEL distros use Network Manager, but with their own ifcfg-rh plugin so the configuration is backed by files in /etc/sysconfig/network-scripts.
Again here, I disable the ifcfg-rh plugin and use the native key-file plugin.
ARCH
Of course on Arch, you choose what you want to use during install. In my case, I just go with vanilla systemd-networkd since itās going to be installed anyway.
At the end of the day, there are basically 3 options: ifupdown, systemd-networkd or Network Manager. ifupdown is old and not supported on all distros, so the other 2 are really the best options. You can go all in on one and configure all Linux distributions the same way. In my case I am using a little of both, but I definitely do recommend stripping out the unnecessary abstractions.
Note that I had originally planned on going all in on systemd-networkd, but there were 2 issues. The first is that the version that ships with Debian is missing some features. This is also probably the case for RHEL/CentOS 7. Second, I havenāt confirmed this, but Iām guessing that Network Manager is a dependency for a lot of RHEL-centric stuff like oVirt and FreeIPA.
4 Likes
I preferred Debian and Alpineās /etc/network/interfaces config file. I really liked netplan for the fact that you can test the config before applying it and if things go sideways, netplan saves your butt. I could technically do a network apply script and run it locally on the box, like through screen or abduco, but when I get into configuring it, I always forget to do a plan B to recover the network after n minutes.
I also like the ifcfg-eth things in RHEL based distros, although they arenāt as easy to work with as the interfaces file.
On Void, I use ip commands to configure static IPs, bridges, VLANs and routes. I just place it in /etc/rc.local and it does the job at startup. Bonus points that I can make a backup of the file, execute the rc.local file and sleep. And if I donāt ctrl+c before the sleep is over, I can make it delete all the config on the ethernet port and execute the backed up rc.local file.
ip commands are available in virtually all linux distros. It is the configuration tool.
3 Likes
It is the first way many of us probably learned to configure network on Linux (certainly was for me). I donāt know about Alpine, but I wouldnāt be surprised if Debian gets rid of it in the next release or two. Itās just old at this point.
That is a nice feature of netplan. If it were more widely supported and feature-complete, I would consider using it.
Itās not terrible to use, but I just donāt understand why it needs to exist. Thereās already a perfectly good syntax for configuring Network Manager that isnāt limited to RHEL distros.
Yeah if you donāt have systemd, then that make sense. Go back to the old BSD ways.
1 Like
The ifcfg stuff works outside of NetworkManager and it can be set to not be controlled by nm. Very important if NM config gets screwed up, probably used at least for ensuring out-of-band management interfaces donāt get affected by the NM configurations.
1 Like
I knew this was the case for ifupdown, but not for the RHEL ifcfg config. Do you know what mechanism it uses to apply configuration when not using nm? IIRC, all RHEL distros come with nm, so Iāve never seen ifcfg outside of that context.
Based on this, it appears that the RHELās ifcfg was originally their own backend for ifupdown.
The main reason for using ifcfg-rh plugin is the compatibility with legacy configurations for ifup and ifdown (initscripts).
https://developer-old.gnome.org/NetworkManager/stable/nm-settings-ifcfg-rh.html
1 Like