Rubenerd

Abby Bloom’s article for the Guardian Australia on Monday put the difference more starkly than any I’ve read for a while:

Flying back to Sydney on US election day was momentous. It also felt like whiplash. The Australian governments swung into action even before we stepped off the plane in Sydney. Once in the terminal it was full-on with precise coordination across jurisdictions and levels of government – immigration, biosecurity, state health authorities, police, army, air force. It embarrassed me that an air force officer was pushing my baggage trolley as he escorted me to my [mandatory quarantine hotel] room. Of course this wasn’t a courtesy: he was there to make sure that I was securely locked in my room without a key, open window or balcony for escape.

Both the US and Australia are responding to the same pandemic but you would hardly know it. In the US magical thinking and the elevation of individual freedom above the public good has squandered precious time. The number of deaths each day in the US quadrupled in just the four weeks after I landed in New York. Today it is up 30% in the past 14 days. Hospitals are reaching capacity and beyond. In a little more than two months my mother will have completed an entire year in self-quarantine, isolated from loved ones except for outdoor visits while the weather permitted. She’ll probably turn 107 before both of us are vaccinated and can once again embrace. She has never met her first and only great-grandchild, born during the pandemic, and probably never will.

A colleague of mine recently had to spend Thanksgiving with his American wife in Australia, because they couldn’t go back home to see family. It’s the same story in so many places around the world, but the psychological impact of seeing the world’s most powerful country reduced to this is hard to overstate.

At least vaccine deniers complicit in the further spread of this disease, and who conflate freedom with freedom from responsibility, have gone from saying it’s mind control, to claiming the elites are hoarding it. Why would they if it doesn’t work? Whoever planted that brilliant seed might save more lives than anyone else.

Antranig Vartanian wrote about his jump from macOS to a FreeBSD ThinkPad at the end of November. It comes after a year or so of tweets and comments to my blog here discussing his increasing frustration at where the Mac platform is headed, and I find it hard to fault him.

His main reasons for the switch came down to macOS becoming less Unix-y, with services he doesn’t need and an outdated userland; Big Sur’s frustrating new interface; Linux and Windows having their own problems; and his working knowledge of FreeBSD on the server making the switch easier. I’ve noted the same trends, and have similar views.

Mac users like us have felt the ebbs and flows of the platform over the years, and the announcement of the M1 chip has reaffirmed the company’s interest in the Mac after a lost decade chasing iOS instead. My current Intel 16-inch MacBook Pro is a great machine, and all the benchmarks suggest the M1 based Macs absolutely smoke any PC in their class. Almost all PC laptops have crappy screens with bad colour, poor viewing angles, and low resolutions; and their trackpads don’t come close to matching the sensitivity and precision of what Apple had a decade ago.

(Font rendering is another sticking point. Ironically, *nix desktops still have better looking fonts than Windows. It’s another thing you become accustomed to on macOS, and the standardisation of their Retina screens only makes the difference more stark when you go back to PCs).

But software is another story, and I’m tending to lean in the direction of Antranig here from a usability perspective. Every Big Sur screenshot I’ve seen has reminded me of how I blanched seeing Windows Aero in Vista/7 and Metro in 8/10. We’re by no means in the minority here; prominent Mac developers have voiced their own frustrations at the seeming iOS 7-ificiation of the platform and the throwing away of their own user interface guidelines.

Apple have their reasons for doing this, but it shows their priorities for the platform are drifting further away from what users like me buy their kit for: a portable Unix workstation where I spend my time working and playing a few games rather than constantly tinkering with window managers and drivers. This wasn’t a coincidence or convenient alignment of our priorities, Apple used to heavily advertise macOS’s certified Unix pedigree. They’re the only ones left doing this after SGI and Sun.

I’ve got it to the point where I can configure a new Mac, save for updates, within an hour. Homebrew Cask lets me install graphical packages, and I lean heavily on NetBSD’s pkgsrc tooling for everything else. I’ve somehow avoided the headaches others in the community have had with VPN configs, firewall rules, OpenSSH, and others. Maybe I’ve just internalised so much of it from two decades of use.

That said, I’ve been careful to only use cross-platform software where I can, save for a few work-specific tools. Emacs, of all things, removed some of the last Mac-specific tools I used for years for note taking, reading RSS, using IRC, and expanding snippets. My Venn diagram of what my on-call Panasonic Let’s Note laptop with FreeBSD can do, and what my Mac can, overlap more than ever before.

I’m glad to see gentleman like Antranig are able to productively move onto a FreeBSD desktop that works better for them. I’m sticking with the Mac for now, but it sounds like all I need is some more compelling hardware. I sure wouldn’t mind more upgradeable ones!

Algorithms determine more of our lives than I think people realise. It’s hard to think of anything that hasn’t had machine learning or even a cold decision tree affect it somewhere along the way, and the trend is only accelerating with ubiquitous compute power, storage, and capture devices.

Gabby Bush, Henrietta Lyons, and Tim Miller wrote in the University of Melbourne’s Pursuit magazine about the limitations of these systems for predicting grades in the UK, but it’s questions on ethics and accuracy are broadly applicable. This paragraph in section two so elegantly summarised the promise:

Algorithms, in their most utopian sense, could ensure that decision-making is more efficient, fair and consistent. Algorithms could, if designed and regulated with ethical considerations in mind, make decisions more objective and less biased.

With ethical considerations in mind being the operative phrase. They expanded a few paragraphs down:

[..] predictive algorithms don’t account for a multitude of factors that might determine a person’s success.

Nor do I think they ever could. Algorithms inherently carry the acknowledged or unknown biases of their developers and the datasets they’re trained against. No developer can be perfecly neutral, and data cannot have perfect coverage. We don’t even fully understand all the factors in a decision or event; algorithms can help us surface them, but then we’re potentially falling into the same trap. There are potential legal implications, and we may not approve of algorithmic outcomes on the human values of equality and fairness.

Algorithms, when applied appropriately, can save time and even lives. But neither negate the need for human oversight and careful application. You’d think this would all be obvious in 2020, and yet here we are:

Ultimately, the UK government backed down and students’ results will now be based on the grades that were predicted by teachers based on their students’ progress. However, this is the latest disturbing precedent of how governments’ careless use of algorithmic decision-making can have devastating consequences for citizens and particular social groups.

It’s our responsibility in this industry to keep sight of this, and inform decision makers. They can appeal to ignorance; we can’t.

I think we have an evergreen.

People like reading about new server hardware! I’ve had more than a dozen emails, some from the usual suspects among you, and a few from people who’ve never posted. There are common themes in all of them, so I’m aggregating them into the following:

Why not Ryzen/Epyc?

I’m over the moon that AMD are smashing it again; I used their kit almost exclusively in machines I used to build myself. I went with a Xeon because I got an unreasonably good deal on a motherboard, and because they still have a slight—albeit shrinking—edge in hypervisor compatibility.

Why not a second-hald Dell rackmount unit, etc?

I could have got one, and then have had iDRAC etc. But I live in a studio apartment, and a desktop tower is already pushing space and accoustic limits. I’d have absolutely considered a LACK rack with one or more of these if I had a loungeroom or basement.

Why FreeBSD?

New readers may be unfamiliar, but FreeBSD has been my preferred server OS for a long time. I detail why on my software page. It’s OpenZFS integration is the most pertinent for this build; it’s the only file system I trust with my data. I also run Debian VMs to keep a foot in the Linux world, and because we run it at work.

Is ECC memory worth spending that much extra?

Yes, especially when refurbished ECC-capable hardware is less than new memory without error correction. Allan Jude has quipped that ZFS without ECC is still more trustworthy than regular file systems with ECC. But why not go all out.

Integrated graphics are terrible, use a GPU!

It’s mostly a moot point for a server that’ll only ever been SSD’d, Plex’d, or Minecraft’d into. I have no idea how offloading works, but potentially I’d be up for getting a GPU if I could do transcoding on it, or if its feasible.

Noctua are fine, but they’re overpriced. Get XYZ instead

First, too late! And second, as I mentioned, this was the one component I splurged on because I love their engineering. Noctua’s used to be aspirational devices I never thought I’d justify or own, but with all the money I saved elsewhere, and given how shambolic this year has been, I thought why not.

You could get an Adaptec SCSI PCI card for those Iomega drives

People cottoned onto why I wanted this specific board :). I’m not sure if that’d be remotely feasible, but it’d be fun. My machines have always a bit of nostalgic pointlessness in them.

Will it run Emacs?

Okay Dave, I took the bait! No, Emacs is for local hacking. Remote config files on this machine will be edited via Ansible, or the built-in nvi editor. I like keeping my servers as sparse as possible, and use tools provided by the system.

Claire is back, with her own channel, and she made my favourite bread of all time! Clara and I are so happy.

I love my little HP Microservers cubes, but they’re starting to show their age. They’ve performed admirably over the years and taught me everything from Xen and InfiniBand, to Minecraft and Plex. But I’ve extended them well beyond what they were designed for, and they’re loud when under the concurrent loads Clara and I subject them to now. I also want some better performance for media transcoding, and CPU features like AES-NI for encrypted storage. In the words of Gough Whitlam, it’s time.

I enjoyed building my game machine more than the games I ended up playing on it, funnily enough, so I decided to channel my interest into building a new homelab server again. The first spark came from seeing a few generations old workstation Supermicro board on Gumtree for peanuts, and rediscovering the majesty of the Antec 300 case. Suddenly I was down the rabbit hole again researching the best parts for the dollar for this new machine.

This is what I’ve cobbled together over the past few months for far less money than I would have expected! All are refurbished or second hand except the CPU cooler and power supply.

• Supermicro X11SAE-M. Micro-ATX LGA1151 board with 8 SATA ports and USB 3 headers, and C236 for ECC memory and VT-d support. The case could easily fit a larger board, but this does all I need.

• Xeon E3 1275 v6. Best cheap chip you can get for this board before the price and TDP shoot up. 4C/8T, 3.8 GHz with 4.2 GHz burst, and integrated graphics so I don’t need a separate GPU card to view POST before jumping to SSH. Insert benchmark disclaimer here, but its Geekbench score is more than 5x the Microservers’.

• 2x 16 GiB Kingston PC4-19200 DDR4-2400 ECC, unbuffered, unregistered DIMMs. Was aiming for Crucial/Micron, but got these tested second-hand for almost nothing.

• Corsair HX-750 Platinum. Modular, quiet power supply from a brand that’s earned my trust over the years, and supports 9x SATA devices. Spent bit extra for Platinum over Gold because it’ll be running 24/7.

• Noctua NH-U12A. CPU radiator and cooler with the best thermal performance and noise in its class, as expected. Only thing on this list I splurged on, because it’s such a beautiful device! I love precision engineering like this.

• Internal Zip, Jaz, and Hex Speed Creative CD-ROM from my first computers in the 1990s, or bought for nostalgia. Will the ever be connected internally? Probably not! But it tickles me having them in a modern(ish) tower.

• Lenovo Ultrabay 9.5mm Multi Burner in an Icy Dock Flex Fit Duo, alongside the aforementioned Jaz drive. It’s all this massive tower has space for after I’ve added all that other pointless hardware above! Will be mostly used for reading CDs and DVDs to rip, etc.

And on the software side:

• FreeBSD 12.x or 13-CURRENT with Xen, running FreeBSD and Debian guests VMs. The exact configuration is TBD, including what gets a guest or a jail.

• She’s inheriting the hostname aino.lan, from Sailor V and Sailor Moon. My first Pentium 1 tower got the name Mizuno Ami, so she’s playing the supporting role! Thinking of getting this RedBubble magnet to make it official, she’s adorable.

Next steps:

• Getting all these parts delivered at some point! They’re slowly trickling in, but might be a while still.

• Getting some matching coloured SATA cable pairs for my mirrored ZFS pools. PC Case Gear has some nice coloured-sleeve ones.

Stay tuned for more updates :).

I emphasise with Manton Reece’s tyranny of the blog title; sometimes nothing really fits.

Save for a blip in Victoria which has since been flattened as well, we’re extraordinarily lucky in Australia and New Zealand with our Covid cases. Contact tracing and home isolation has kept our new cases at zero or in the single digits for a while. I’m not sure whether our proximity to Asia made us more aware and serious earlier, or the fact we’re massive islands, or strict quarantines, or that our state premiers filled the competency vacuum left by Australia’s PM. I imagine we’ll be analysing this for years.

I still worry that complacency will breed another spike. People here are largely living life as they did before now, albeit with government stimulus spending, check-ins everywhere, and the bulk of workers staying at home. I take the train into work a couple of days a week now, and it’s still a far cry from regular morning peak hour. But shops and restaurants are full again.

It’s those retail QR check-ins that led me to think something was serious in my email this morning:

Rapid COVID-19 test

… uh oh. I went through the motions: where had I been, who had I come into contact with, where did I slip up, did my hand washing and mask reduce the chances, who do I need to call right now?!

Back in July I got news that a potential Covid case had been linked to a coffee shop I’d been to the same day. I went into isolation and ordered a mail in testing kit. It turned out to be a false alarm; the person didn’t test positive, and I left the coffee shop before they arrived.

This email was also a false alarm:

As this unforgettable year draws to a close, I’d like to take a moment to thank all of our volunteers who so generously contributed their time to the University [of Technology, Sydney]. Whilst 2020 has been full of challenges, our remarkable community has continually come together to innovate, research, mentor and support UTS philanthropic causes.

It was an alumni newsletter, with that as the subject line. How did the person in charge of sending that blast-out think it’d be received? Can you imagine getting that subject line flash on your phone during a meeting, and pretending to be composed while you wrack your brains with the above questions?

Me thinks that some of that innovative research should be spent on reading the room.

This would have been a great Music Monday had it been written on Monday and consisted primarily of music discussion. Scratch that, it would have made a lousy Music Monday now that I coach it in those terms.

I was working on the balcony this evening as I’ve been during since Covid started. It was another scorcher today but there was a pleasant breeze for much of the afternoon. Out of the grey I could smell that telltale ozone, then Paul Simon chirped about the rain on the interstate. Then it started bucketing down.

Frank Nora of The Overnightscape would refer to this as a synchronicity, though he has more interesting and crazy ones that I seem to! But that doesn’t stop me appreciating those fun little coincidences where the world lined up in a specific way that, had I not been paying attention, would have completely slipped by.

These brushes with reality make me think how many other things I’m missing out on; not by deliberate choice, but because I’m blissfully unaware. Not just the bigger picture ideas that are consequential and important to the IT industry, or the environment, or health, or anything else I care about. But the small things.

I sit out here doing work, glued to the company chat, my email client, conference calls, and my text editors. Then I pop in headphones to listen to a podcast. Heck, I was listening to this specific song to keep my energy levels up while emailing a client about the finer points of specific software licences. But then I had a microbreak of watching the sky and listening that ended up lasting more than a minute, and I got something great out of it.

I’d been feeling a bit melancholic all day, but feeling the rain lash the sleeves of my shirt as I sit out here under a dark sky, cowering at my laptop and finishing the last of my cold coffee, I couldn’t help but smile. That silly little alignment of circumstances absolutely made my day, where otherwide I’d have been irritated to be wet, or at having another video conference call be interrupted. What felt like a cage out here is now something I’m enjoying. I might stay out here a bit longer.

The old weary traveller waits by the side of the road. Where’s he going?

The title was supposed to be Unlearning unsaved buffers, but I couldn’t resist adding the same prefix to the last word, even though it makes no sense. Or especially because it makes no sense. Or cents. Or pounds. Pounding this joke, or flogging it, despite it not being funny. Or especially because it isn’t funny. Oh no, I’ve internalised that ABC skit!

I’ve made the switch to org-mode in Emacs from nvALT to capture, store, and search for notes. I plan to write a proper series of posts about it; no one plugin or program has mapped with how my mind works so well in years. It’s also one of the last Mac-specific tools I’ve been able to replace with a cross-platform one.

But in doing this I discovered another weird habit that, with hindsight, made absolutely no sense. For certain types of notes I was opening MacVim instead, and immediately typing in a new buffer. Then I’d inevitably need to context switch a few times, which necessitated opening even more buffers for even more disconnected thoughts. Suddenly I’d be staring down the end of the day, with all these unsaved buffers hovering around my mind and computer screen, as if mocking me for being so scatterbrain.

Which leads me to how Emacs has helped me in another way I didn’t expect. You’re presented with a splash screen and scratch space when you open it by default, but you either need to provide it a filename with C-x C-f to start writing, or invoke Emacs with a file path. This flew in the face of the above habit which initially frustrated me, but now I start each note and file with a deliberate subject and folder! Also telling, I rarely need to revise the name either, showing that the filename itself is useful for framing what I’m about to write, and setting the scope. Now I reach the end of the day, and I have more org-mode nodes, and well structured text files.

Tools can be used to enforce bad behaviors or learn new ones. The trick is figuring out which is which.