< Notes...

Well, well, well. It is possible, after all, to create a /boot partition on software RAID10, and been able to boot from it just fine—something I tried many times, and couldn’t do. It turns out you need to flag each drive as a bootable disk, before carving out partitions, and creating the RAID10.

This section from the NetBSD man page for the sleep command cannot be funnier, yet completely valid. I mean, quite useful if you have the time, right?

BUGS

     This sleep command cannot handle requests for durations much longer than
     about 250 billion years.  Any such attempt will result in an error, and
     immediate termination.  It is suggested that when there is a need for
     sleeps exceeding this period, the sleep command be executed in a loop,
     with each individual sleep invocation limited to 200 billion years
     approximately.

Hashing known_hosts is a good idea, as it reduces the amount of information an attacker will collect, if/when (ha!) your machine gets compromised. Usually, a line on the known_hosts file looks like this:

less ~/.ssh/known_hosts
...
tilde.team ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAID1zw6+VOW8L4Rr3swbUVju3GGcknaV/fyhSJwH7NLfu
...

After running ssh-keygen -H it will look like this:

less ~/.ssh/known_hosts
...
|1|c6NiIH06AidrkPUman0oPEx6+6Y=|rPhYjXlkLgYaNR8jwaNMy7mur4I= ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIJsJP1XDyRhEPdtgBeXYm2hf4GKG9aLlqA1+ZPgBadbl
...

That’s it! Oh, and don’t forget to add to your ~/.ssh/config the following, so that future entries are hashed as well:

Host *
	HashKnownHosts yes

grep -i passwd *.log | awk {'print $7'}
...
/etc/passwd
/..%5c..%5c..%5c..%5c..%5cetc%5cpasswd
/....//....//....//etc/passwd
/..%2f..%2f..%2fetc%2fpasswd
/.htpasswd
/config/.htpasswd
/.env.passwd
/backup/.htpasswd
...

With almost 2,000 different tries—of which the above is a small sample—it seems someone is hell-bent on finding a passwd of some sorts. I don’t have one, OK?!

I miss terminal based applications. Old database driven applications that were usually accessed via VT100 terminals, and later on via IBM 3270, or emulated software, like TN3270, to access IBM mainframes. I wonder if they will ever make a return (no mainframes needed this time, of course).

Hmm, is this a sign of crossing—or having crossed—a certain age threshold?

It took me till now to install it, but worth it: bat is better than cat (and, arguably, less) in, for what I can see, almost every single way. The only thing that makes the last two “better” is that they are normally installed per default.

If you want to quickly test your disk speed on Linux (might work on macOS, but not 100% sure) simply do:

sudo dd if=/dev/zero of=test bs=1G count=7 status=progress ; rm -f test

Decently happy with my results: 7516192768 bytes (7.5 GB, 7.0 GiB) copied, 5.10763 s, 1.5 GB/s. I figure, though, that’s expected on an NVMe M.2 SSD.

I have been struggling for days to build an ISO image that will auto-install Ubuntu workstation. Canonical’s way is to install server, then add ubuntu-desktop (and other packages needed/wanted), then remove ubuntu-server, and ubuntu-minimal-server. That’s mind boggling to me. That’s nuts.

I haven’t been able to create an “autoinstall” ISO for Ubuntu 22.04 Desktop. I haven’t found a way to make the live ISO image not to start the guided wizard GUI. As a workaround, I am using Ubuntu 22.04 Server, and adding ubuntu-desktop and gdm3 afterwards to the server install.

It turns out Ubuntu font is quite pleasant, clean, and legible. I am such a sucker for fonts!

Firefox web browser font rendering on Linux is super nice. Chromium slight blur is no match. I am seriously considering using Firefox exclusively under Linux (Wayland). For me, at least, it seems “old” is “new” again. LOL.

Going back to this, work has now provided me with a Linux (Ubuntu) laptop, and a docking station. I am to relinquish my dear Mac, and stop using (for work) macOS. I would have never thought of this happening!

Moving from my beloved iMac 2015 (running macOS Monterey), to this little abomination of a Linux desktop (because, you know, 2023 is “the year of the Linux desktop” /s).

This blog post is everlasting. Since 2018, when it was written, things have gotten worse. I don’t see the light at the end of the tunnel anytime soon.

We put virtual machines inside Linux, and then we put Docker inside virtual machines, simply because nobody was able to clean up the mess that most programs, languages and their environment produce. We cover shit with blankets just not to deal with it. “Single binary” is still a HUGE selling point for Go, for example. No mess == success.

Our engagement with Canonical to deploy and manage Ubuntu with MAAS and Landscape is over, and now the “fun” begins. So much to customise, and tune! I have spent the entire day tunning and rebuilding images, back and forth. I am hoping to work on post-install scripting tomorrow.

New Red Hat owners, IBM, “f***** around, and they are finding out”. Their decision to stop publicly releasing RHEL source code is going to come back to bite them. New distros pledging to be RHEL compatible have spawned, and now SUSE and Oracle are, too, having their saying.

I don’t think this is a good news. It could mean LXD “might” get killed further down the line. That would make me—and others, I am sure—sad.

SUSE is positioning itself to take advantage of Red Hat’s changes (which has sparked criticisms). Excellent approach!

This is rubbish. Same as this. I can’t think of any other word to use for them. Canonical’s Ubuntu management and deployment tools are not even close to Red Hat’s.