Clicky mechanical keyboard with blank keys is a genuine godsend. It keeps people from messing with my desk :)

If you *can't* touch type, then you're using a few fingers rather than all ten and that *must* slow you up. If you. if you *know* you can touch type, then you can phrase your thoughts without having to worry about anything further.

All this said, I've know a bunch of Linux sysadmins who were excellent sysadmins but functionally very dyslexic when typing English on the same keyboard.
There is an element of difference between typing ls -al ./* and a full sentence.

As per the article referenced, Luke identifies as they/them. Their blog rather than his blog preferred.

Shouldn't the headline at least read deny-list rather than blacklist?

AWS2023 is specifically Fedora 34, 35 and 36. It's incompatible with EPEL. I don't understand how they will keep it supported for five years given that it is already unsupported upstream ...

Potentially, Ubuntu supports many more packages than Red Hat. Red Hat itself is small in terms of packages and bundled applications - and Red Hat don't necessarily support EPEL packages coming ultimately from Fedora.

If you absolutely positively have to have something supported for up to 10 years, and incorporate FIPS / PCI credit card processing / needs to work for US Government or similar or (insert $$$ generating vital commercial thing that requires heavyweight regulation here), then Red Hat and IBM might be your answer.

If not, there's Ubuntu Pro (or Ubuntu with third party support).

If neither, then there's OpenSUSE / Debian - and you may get just as good support from something completely non-commercial like Debian as you actually do from one of the paid for support plans from others. You will have to convince beancounters and bosses of this, of course.

It exists, it works, they chose Debian because it worked - see also https://meetings-archive.debia... where much is explained.

Is it too early for "Oh, the huge manatee!!" ?

Don't bother - it doesn't build cleanly - "Have you hydrated the filesystem" - error is in their issue tracker. Could do with better quality control

At this point, it looks slightly messy. Building your own distro - and RPM-based at that - is not a great idea. I hope they've _really_ good dependency management and a top notch security team. If this is their distro for Azure - here's hoping they're prepared to sit on top of every CVE. There's Go in the mix - I hope they've got
enough expertise to keep it going well.

As well as this, let's also hope that they speed up their own internal processes - like getting shims signed for secure boot / accepting other distros as guests on WSL2. Scrapping their own distro and rolling the hooks for everything to allow every other distro to peer on WSL2 would have been a much nicer idea.
They've been working with Canonical but have obviously missed a few tricks about Linux distribution management along the way :(

As ever, /me would much rather have seen something sane and Debian based, but, hey, this is the company that brought us Windows ME.

This is not just killing the goose that lays the golden egg but taking the last example of the species and vapourising it. Hudson did this and Jenkins arrived immediately. Oracle did this with Solaris, MySQL and the JRE.

Freenode is now wholly worthless and the person most obviously behind it is persona non grata, effective immediately. Donors of servers / bandwidth / hosting to Freenode - pull it because you won't gain glory by association. Andrew Lee hasn't covered himself with glory here and has terminated any reputation he retained with extreme prejudice, whatever you might choose to say about the libera.chat admins.

Meh, who cares? Some of us still use IRC daily - to be honest, I'm surprised more projects like Ubuntu, Alma Linux and co didn't just move to OFTC.

It _might_ be immaterial now: a few years ago, I had a 30TB RAID array made up of 30 1TB physical drives. One 10TB shelf failed as a disk lost parity During the rebuild, another disk failed ... the likelihood of this sort of thing goes up quite quickly on large arrays if something causes a few drives to fail over a period of time..

Building large arrays and maintaining them is non-trivial and that's where hardware RAID was traditionally good for not taking up the bandwidth of the CPU while rebuilding. If you've got 4 x 4TB in a four bay microserver or whatever and one dies, expect it to be less usable when the first disk dies.

I've also had battery backed hardware RAID fail because no one had checked/changed the batteries, they'd swollen and cracked the RAID card ...

RAID is not a backup : it won't preserve your irreplaceable data indefinitely

It's here - it works. It's available. Red Hat Enterprise Linux has just released 8.4 beta - what will be most interesting is how long it takes any of the clones other than Oracle to release an 8.4 when it finally comes out.

Oh - and I forgot - no security fixes until they've been fixed in RHEL first :( == DO NOT USE IN ANY CIRCUMSTANCES

Been here, seen it - Red Hat 9 -> Red Hat Enterprise Linux 2.1 with Fedora as an afterthought. Lots of discussion on LWN (lwn.net).

And even for the computers that are synchronising time - think of the number of updates to tzdata on every Linux system necessary ... INSANITY!