Page not found – Just Keep Noobing https://noobient.com Just Keep Noobing Wed, 15 Nov 2023 19:41:24 +0000 en-US hourly 1 https://wordpress.org/?v=6.4.2 Fixing Ubuntu containers failing to start with systemd https://noobient.com/2023/11/15/fixing-ubuntu-containers-failing-to-start-with-systemd/ Wed, 15 Nov 2023 19:41:24 +0000 https://noobient.com/?p=2505 Read more »]]> If you’re using either Docker or Podman, the situation is as follows:

Error: OCI runtime error: chmod `run/shm`: Operation not supported

Or if it’s a 18.04 or even older container:

Error: OCI runtime error: chmod `run/initctl`: Operation not supported

Apparently it only affects Ubuntu, the eternal problem child, but not Fedora or AlmaLinux. Those containers still work fine with systemd. Apparently it’s not even a Docker or Podman issue either, but a crun one, which is apparently the new default container runtime of Podman (but you can also use it Docker, if you want). They already fixed the issue in crun, so all you need to do is update. Right?

This is where things get tricky, because crun is installed as a package. So what if you try to uninstall the old version, can you guess?

The following packages will be REMOVED:
  crun podman

Oopsie. So yeah, Podman clearly depends on crun. The version we need is 1.9.1 or higher, that’s where they added the fix for this. Ubuntu 22.04 has version 0.17 (oof), 23.10 has 1.8.5, and 24.04 is something like pre-alpha at this point, so we cannot expect a working package anytime soon.

The alternative is to obtain a local copy from GitHub. I’m using 1.11.2 in this example, that’s the latest at the time of writing, but please update it accordingly:

CRUN_VER='1.11.2'

Now obtain the binary:

mkdir -p "${HOME}/.local/bin"
curl -L "https://github.com/containers/crun/releases/download/${CRUN_VER}/crun-${CRUN_VER}-linux-amd64" -o "${HOME}/.local/bin/crun"
chmod +x "${HOME}/.local/bin/crun"

Verify that you’re indeed on the latest release now (you might need to restart your shell):

$ crun --version
crun version 1.11.2
commit: ab0edeef1c331840b025e8f1d38090cfb8a0509d
rundir: /run/user/1000/crun
spec: 1.0.0
+SYSTEMD +SELINUX +APPARMOR +CAP +SECCOMP +EBPF +CRIU +YAJL

Now we still need to tell Podman to use the new binary:

mkdir -p "${HOME}/.config/containers"
cat << EOF > "${HOME}/.config/containers/containers.conf"
[engine.runtimes]
crun = [
  "${HOME}/.local/bin/crun",
  "/usr/bin/crun"
]
EOF

Or if you’re having this issue with Docker:

mkdir -p "${HOME}/.config/docker"
cat << EOF > "${HOME}/.config/docker/daemon.json"
{
  "default-runtime": "crun",
  "runtimes": {
    "crun": {
      "path": "${HOME}/.local/bin/crun"
    }
  }
}
EOF

Aaand ta-da, it’s working again. Don’t forget to update crun manually every once in a while. This manual workaround should no longer be needed once 24.04 is out. Until then, happy containering!

]]>
Fixing VS Code in WSL with systemd https://noobient.com/2023/09/05/fixing-vs-code-in-wsl-with-systemd/ Tue, 05 Sep 2023 17:14:53 +0000 https://noobient.com/?p=2478 Read more »]]> You might be as naive as I was and try to install VS Code natively in WSL using the official docs, but then you’ll get the following:

$ code
To use Visual Studio Code with the Windows Subsystem for Linux, please install Visual Studio Code in Windows and uninstall the Linux version in WSL. You can then use the `code` command in a WSL terminal just as you would in a normal command prompt.
Do you want to continue anyway? [y/N] y
To no longer see this prompt, start Visual Studio Code with the environment variable DONT_PROMPT_WSL_INSTALL defined.

If you then proceed, you’ll get a very odd-looking VS Code indeed. Scaling’s off, window decorations are off, etc.

Then you might as well go ahead and revert the damage:

sudo apt purge --autoremove code

Now start the Windows version:

$ code
/mnt/c/Users/bviktor/AppData/Local/Programs/Microsoft VS Code/bin/code: 61: /mnt/c/Users/bviktor/AppData/Local/Programs/Microsoft VS Code/Code.exe: Exec format error

What the heck? So after some investigation, it turns out to be a systemd-specific issue. The workaround is pretty simple:

echo ':WSLInterop:M::MZ::/init:PF' | sudo tee /usr/lib/binfmt.d/WSLInterop.conf
sudo systemctl restart systemd-binfmt
code

And bam! VS Code is up and running in WSL even on systemd.

Thanks a lot for the hint, cerebrate!

]]>
Electric Mobility Parity Index – 2023/07 Europe https://noobient.com/2023/08/08/electric-mobility-parity-index-2023-07-europe/ Tue, 08 Aug 2023 16:46:22 +0000 https://noobient.com/?p=2469 ]]> Electric Mobility Parity Index – 2023/06 Europe https://noobient.com/2023/07/03/electric-mobility-parity-index-2023-06-europe/ Mon, 03 Jul 2023 10:29:45 +0000 https://noobient.com/?p=2461 ]]> No, EV range is not fine https://noobient.com/2023/06/18/no-ev-range-is-not-fine/ Sun, 18 Jun 2023 18:43:51 +0000 https://noobient.com/?p=2434 Read more »]]> Preamble

I’d like to start by stating that I’m an EV fanatic, and have been ever since I read What it’s like to own a Tesla Model S by The Oatmeal. I’m not sure when that happened, but at least 6 or 7 years ago. I’m also an environmentalist since elementary. Yet, I see the world through facts and numbers. And when I bring those up to refute unjustified feelings, impressions, or suspicions, I’m always called a fossil fuel sellout, a petrolhead, destroyer of worlds, or simply just an inconsiderate asshole, amongst others.

There’s this ongoing trend of articles on EV enthusiast sites about how great EV range already is. How we reached range parity, and that EVs are a proper and full replacement to the internal combustion engine. How further improvements to range are completely unnecessary and uncalled for. Then you get a good dose of Copium with various narratives thrown around. The latest one is from CleanTechnica, titled Going To The Gas Station Must Really Suck. This one pulls the “home charging is so superior” card. But there’s an endless amount of such articles and posts, it’s pointless to try and list them all.

I get it. EVs need to be promoted. We need people to believe in the tech and give it a chance. But if you mislead people, they’ll only regret their choice and will hate EVs even more. And convincing an already disappointed customer is orders of magnitude harder. So why are EV fans trying so hard to alienate all these people? We really need to dismantle the issue, because to me it seems it’s just not something that’s being talked about enough. At all.

Range Anxiety

So what’s the fuss about to begin with? It’s rather obvious an EV can handle city travel. Even a PHEV with 40 km range can do that. The city is the easiest to conquer, and it was done the moment EVs were made. It’s pointless to even bring it up. The tech to do city travel has existed since the 1800s. No, I’m not kidding.

So the issue is obviously the motorway. This is problematic for at least 2 reasons:

  • distances are much higher
  • speeds are much higher

For this reason, travel becomes a whole lot different, because you need to refill your vehicle during the trip, not between trips. And when that happens, a a bunch of aspects become way more critical, namely:

  • refilling time
  • amount of refilling stations
  • refilling frequency, a.k.a. range obtained per refill

And guess what: EVs suck at all three. But just how much do they suck?

Refilling Time

This one is the simplest. Pretty much all current ultra-fast chargers charge from 20% to 80% somewhere around 30 minutes. Compared to gasoline, that’s pretty bad, because gas takes maybe a minute or two to pump into your tank, the rest is just overhead, i.e. stopping, paying, and speeding up again. So even if we say a generous 10 minutes with a cigarette and a restroom break, that’s still at least 3 times worse.

Refilling Stations

I already cover this in my EMPI updates, but long story short, as for ultra-fast charging stations (the only ones we may consider even remotely comparable to petrol stations) , we have around 100 times less than gas stations. Gee, we’re getting into some worrisome territory now, huh? But just wait until we get to…

Range

Oh boy, this is a juicy one. To sum things up, driving cycles suck ass, but that’s another article for another day. In any case, the Tesla Model 3 LR, which is quite high up there on the list of EVs with the most range, still only has 626 km with the Aero wheel caps, according to the WLTP cycle.

But wait, why only 626, that sounds pretty darn neat, ain’t it? Well, yeah, it sounds great. But the WLTP cycle has pretty much nothing to do with reality, or at least with the motorway part of it. You know, the part where it matters the most? There’s a youtuber called Bjørn Nyland who single-handedly left all these idiotic, useless “driving cycles” in the dust that the automotive industry managed to come up with in the past 20 years. He compiled this incredible sheet with a shitton of real world driving tests, often under various conditions. Now this is something we can work with.

Just do the fucking math!

So according to Bjørn’s sheet, a 2021 Model 3 LR, driving alone, in 25 °C, dry weather, i.e. the most ideal conditions possible for EV travel, can pull 422 km at the speed of 120 km/h. Wow, one third of our promised range – gone. And this is only 120 km/h, while the majority of European countries have a limit of 130 km/h or higher. In Poland it’s 140 km/h, and Germany has parts with no limits. Yes, you read that right.

Unfortunately, 130 km/h range tests are very hard to come by, the closest one I could find is a Model 3 Performance test from 2019. These numbers are pretty awful, so I’ll be generous and settle with a 200 Wh/km consumption for 130 km/h. The 2021 Model 3 LR appears to have a usable battery of around 76 kWh, which leads us to a full motorway range of 380 km.

But didn’t you forget something? Do you drive your car from 100% to 0%? Of course you don’t. You’ll stop at 20%, because what if the charging station is full, or out of order, or whatever? So you’ve just thrown another 20% of your precious range out the window. We’re down to 304 km.

But to quote Billy Mays, wait, there’s more! Because you don’t do ultra-fast charging to 100%, either! Why?

  1. Because that kills your battery. Unless it’s LFP, but those have shorter range due to lower energy densities, so even if you charge to 100%, it’s worse for motorway travel.
  2. Because the last 20% of charging is much, much slower, and when you’re on a several hundred km road trip, you don’t have time for that. Neither does the next person waiting for you in line. And that’s the lovely hypocrisy of EV makers: they advertise 100% of their range combined with their 10% to 80% charging times.

Oops, suddenly 40% of your range is gone for motorway travel, where it matters the most. We’re down to 228 km. That’s a pretty big far cry from the 626 km we started with, huh? And to be honest, according to Bjørn’s real world tests, even charging to 80% isn’t optimal, and you should probably be charging to 70% or even 60%.

So now we know that in the best scenario possible, we can travel 228 km at motorway speeds. That’s 36% of the original 626 km we got promised. Since our speed is 130 km/h, i.e. the European legal limit for the most part, that’s 1.75 hours, which is 105 minutes. The Model 3 LR charges from 20% to 80% in 31 minutes, and that’s only on the fastest chargers out there, so again, I’m being generous.

So for every 105 minutes of driving you’ll have to stop for 31 minutes. In other words, you’ll spend 23%, or almost 1/4th of your time charging. And that’s when you’re driving alone, in perfect weather, in one of the highest range cars, using the fastest chargers, and I’m even assuming these ultra-fast chargers are always in the right spot, right when you’re hitting the 20% mark, and there’s never a queue, and the charger can charge at full speed (it’s not always the case, because who knows why). Gee, that’s a heck of a lot “ifs”, don’t you think? If the weather gets colder, your range decreases, and charging times go up. Then there’s the rain, the heat (AC!), the wind, the weight of your passengers, and oh did I mention towing? Queues at the station? Should I go on?

EV range sucks. Period.

Stop making excuses

So when you point out all these tangible, hard facts, as a last resort, you’ll be shrugged off anyways. Why? Because you need to take breaks anyway! Have a coffee every 100 minutes. For 30 minutes. Or maybe the idea is to drink so many coffees that you develop a bad case of diarrhea and need a restroom break at every stop anyway? I really don’t know.

My other favorite excuse is, “My EV covers 99% of my driving”. Yeah, good for you. It’s great that you can afford a 2nd car just for that one trip in the year. Most of us can’t, or don’t want to. Rentals are also obscenely expensive in most parts of the World, North America being one of the few exceptions. Telling us to earn more money so that we can afford a 2nd, gas-powered car for that one darned trip won’t help EV adoption, believe me. I need my car 100% of the time, even for that one trip home during the holidays. Ironically, these are the same people who laugh at the concept of PHEVs – you know, the thing that solves this very problem. PHEVs are compliance cars and the root of all evil, yada yada yada.

Remember that article about home charging I mentioned in the beginning? That’s also a recurring theme. Yeah, EV range is lower, but, but, home charging. Now, please, explain to me, how on Earth does home charging, that takes several hours, help me, stuck on a highway, in the middle of an 800 km trip? I really don’t get it. How’s that even relevant? Yes, the ability to refill my car in my home is extremely convenient – that is assuming I own a house in the suburbs with an actual garage, which, again, does not apply to most city dwellers anyway. The point is, literally no one has ever refuted that home charging is awesome. But it is utterly irrelevant in terms of range. Range is only an issue if you wanna go to great distances – far, far away from your home, and time is of the essence. No one cares or worries about city driving. It’s not an issue. Let it go.

There’s also this argument that we should slow the fuck down. I’m sorry to break it to you, but no, we’re not going to drive slower, we’re not gonna change our driving habits just because some EV fanatics told us on the internet. Europeans drive fast, we like it that way, and it will stay that way. Get over it.

But the award of most idiotic excuse probably goes to the one where they say “just go by plane”. Because burning Kerosene is oh so much greener. And spending extra hours to get to the airport, check in, take off, then check out, wait for your luggage, and get to your destination is super convenient and flexible. Or something.

Stop pulling an Apple. Stop telling us we’re holding it wrong. It won’t help EV adoption. In fact, it’ll only achieve the exact opposite. It annoys and alienates potential future EV owners. Stop pushing the 9 millionth article about how awesome EV range is. It’s not, and you’re either in denial or don’t understand the needs of drivers, because you climbed too high up in your ivory tower.

The solution

The next steps are completely obvious. First, we need higher range. It’s being mocked constantly by EV lunatics, like “yeah right, you need 1000 km cars, but then you’ll want 1500 km cars”. But that’s wrong. No one needs 1000 km cars. Apparently ICE cars reached range equilibrium at around 460 miles, which equals to 740 km. The trick is that these cars can actually travel that far, in real life, and not just in some nonsensical synthetic benchmarks. That 626 km range the Model 3 promises would be fine. If it was actually true. But it’s not. We’re lucky if we can squeeze out maybe 1/3rd of that number on the motorway. It’s an inflated lie. It’s pretty much Wi-Fi speeds all over. Has anyone ever seen Wi-FI that can actually pull 24.4 Gigabits per second? Really? Yeah, me neither. I’ll have another article about driving cycles soon, but until then, let’s just say we need better range.

The other one is, of course, charging time. And I can’t stress enough the difference between charging speed and charging time. Charging speeds are increasing constantly, but so does battery capacity. Unfortunately, that means we’re still stuck with this awful 30 minute 20-80 charging. Yeah, sometimes it’s 10-80, but that won’t change things drastically. We need to get this down to maybe 10 minutes, and we’ll see much fewer naysayers, queues will become much, much less of an issue, we’ll have much less downtime at the stations, the list goes on.

Bottom line: get motorway range up to 500 km. Get charging time down to 10 minutes.

Until then, stop saying EV range is fine. It’s not.

]]>
Electric Mobility Parity Index – 2023/05 Europe https://noobient.com/2023/06/02/electric-mobility-parity-index-2023-05-europe/ Fri, 02 Jun 2023 18:58:16 +0000 https://noobient.com/?p=2430 ]]> ASRock outs N100DC-ITX fanless Mini-ITX motherboard after 5 years of standstill https://noobient.com/2023/05/11/asrock-outs-n100dc-itx-fanless-mini-itx-motherboard-after-5-years-of-standstill/ Thu, 11 May 2023 15:40:44 +0000 https://noobient.com/?p=2409 Read more »]]> It’s been quite a while since I last talked about new fanless, CPU onboard Mini-ITX systems. Not because I didn’t want to, but because there’s been literally nothing happening in this space. It’s ironic because my main point for these self-built Mini-ITX systems (over pre-built “mini PCs”) has been that you can just replace the mobo every few years, keep the case, the SSD, etc. and be done with it… well yeah, assuming there’s actually a mobo that you can upgrade to. Unfortunately this hasn’t been the case for the last 5 or 6 years. The last newsworthy system was the ASRock J5005-ITX, and that J5005 chip was released in 2017Q4. Yes, ASRock eventually had the courtesy to release another mobo with the J5040, but really, that’s much of the same Goldmont Plus / Gemini Lake, a.k.a. J5005 on steroids, nothing extraordinary. Definitely not something you’d want to spend a good sum on.

There was also the disappointing paper launch of the Gigabyte N5105I H (Tremont / Jasper Lake), which looks like a somewhat decent upgrade over the J5005, but with one big caveat: it’s nowhere to be found. It’s been like 10 months already, but nope, you still can’t have it. At least I could learn from this article that this glaring lack of new Mini-ITX systems happened due to the “chip shortages” since COVID et al. To be honest, even if they DID make this board available, the lack of M.2 Wi-Fi slot is a huge bummer for HTPC builders. I’ve been complaining about the lack of M.2 SSD support on these Mini-ITX systems, now you would get that with this Gigabyte one, but at the expense of Wi-Fi + BT. Gee, that’s not what I meant.

So after all this storytelling, ladies and gents, now we have a real successor, so please welcome the ASRock N100DC-ITX!

Let’s take a quick look, shall we?

  • It rocks the brand new Gracemont / Alder Lake Intel N100, with a hefty 3.4 GHz boost clock. Base clock speed is not yet public at the time of writing.
  • The processor has the option for both DDDR4 / DDR5 / LPDDR5, but ASRock went with good old DDR4, 1 slot at that. Maximum capacity is a big question mark, because the Intel specs say 16 GB while the ASRock specs say 32 GB. We’ll see about that.
  • TDP is quite hard to believe: just 6W, while the previous Mini-ITX systems usually came with 10W. It may be just the fact that they’ve changed methodology again – we’ll see the real numbers once I can get my hands on one.
  • PCI-E stays at 3.0, but I’d be hard pressed to imagine a scenario where this becomes a serious bottleneck.
  • HDMI is bumped from 2.0 to 2.1, nothing to write home about, same 2160p @ 60 Hz. I’m also a big advocate for DisplayPort (and against HDMI), but virtually all TVs in existence use HDMI instead of DP, so what could you do?
  • 1x M.2 Key E for Wi-Fi and BT. The specs say it only supports CNVio chips, but once again I’m not entirely sure that’s actually the case – there’s been misunderstandings about this in the past. It’s not a big deal though, if the Intel AX200 doesn’t work in this mobo, just try with an AX201 and be done with it. These things are cheap, usually in the $20 range.
  • And finally, at last, after all these years, drum roll… 2280 M.2 Key M slot for… storage! That’s right, finally we can use M.2 SSDs in these, even PCI-E (NVMe) ones at that, instead of clunky 2.5″ SATA SSDs with their stupid cables, eww!
  • If that’s not enough, this mobo even has a PCI-E x4 slot… and an open ended one, too! That means you can actually install an x16 GPU in this gem, like the fanless MSI GT 1030 with DisplayPort. Isn’t that amazing?

And how exactly does it fare against the aging J5005? Well it’s pretty darn impressive, to say the least…

What could I say? I’m impressed. I’m amazed. It looks like ASRock ticked all the boxes with this one. I really can’t think of anything else I could wish for. It also shows perfectly just how much we needed an upgrade over the ancient Gemini Lake.

Now there’s only one question left: WHERE CAN I BUY ONE?!

]]>
Electric Mobility Parity Index – 2023/04 Europe https://noobient.com/2023/04/23/electric-mobility-parity-index-2023-04-europe/ Sun, 23 Apr 2023 15:25:27 +0000 https://noobient.com/?p=2382 ]]> Canonical starts charging for Ubuntu LTS security updates https://noobient.com/2023/04/19/canonical-starts-charging-for-ubuntu-lts-security-updates/ Wed, 19 Apr 2023 10:55:00 +0000 https://noobient.com/?p=2325 Read more »]]> So Ubuntu has this model where they pretty much freeze package versions for an Ubuntu release after release, and then they only backport security updates from upstream. There’s nothing new here, most distros do it this way. The idea is that this way they can polish the gazillions of package versions and ensure they work well together, then never touch it again (too much).

Cool, but Ubuntu comes out with a release every 6 months, and it’s not very feasible to keep 10 different Ubuntu releases updated simultaneously, right? So then came the idea of Long Term Support! Let’s put out an LTS once every other year, only support these for 5 years, and “regular” releases are not really supported anymore, right?

Riiight, but then how will Canonical make money from it? Apparently now they came up with a solution:

Get more security updates through Ubuntu Pro with ‘esm-apps’ enabled
Learn more about Ubuntu Pro at https://ubuntu.com/pro

Well, that’s new, apt asking me to buy a subscription for security updates. On the very latest LTS release. Apparently Canonical learned a lesson or two from Red Hat, because this is the same exact stuff that happens on RHEL after installation – it starts nagging you to set up your subscription with subscription-manager.

But what the feck is ESM anyway? Never heard of it. So apparently this is what Expanded Security Maintenance is:

Security maintenance for the entire collection of software packages shipped with Ubuntu. ESM enables continuous vulnerability management for critical, high and medium CVEs.

You mean the same CVEs that Debian also takes care of, for free? You know, the patches that you usually just copy over into Ubuntu? So it seems that without a Pro subscription you’ll only get “best effort” updates to anything outside main. Well, if they made any efforts to keep me updated without a subscription, clearly, I wouldn’t buy a subscription, so I have a pretty good idea how much effort they’re gonna put into this. Or am I supposed to feel better to have only a few cherry-picked known CVEs on my system lurking around, because Canonical put the patches to those behind a paywall?

So what’s this Pro sub? Let’s check it out:

Geez, this is literally the Red Hat Developer Subscription all over again, except even RH is more generous with its 16 nodes instead of just 5. And I’d like to point out that I got this prompt in a WSL instance. Not a physical machine, not even a VM, just WSL.

How much would it set me back, should I run out of these 5 nodes? $25 / year for a single desktop machine. What if I want VMs? Well, you can’t buy just 1, because they figured no one needs just a couple Ubuntu VMs, so the only option is “unlimited” VMs on a single physical node, for a generous $500 / year. Suuuure. And what’s even gonna happen to Docker containers? Kubernetes? So many questions.

Charging for security updates on very old releases is fine. If you refuse to upgrade, pay the price. But:

  • 22.04 is an LTS release with 4 more years left of its support lifecycle.
  • There’s no newer version to upgrade to. It’s not my laziness, it’s your greediness.

So really, Ubuntu is setting a horrible precedent here. Charging money for security updates on the latest release, really? Red Hat Linux was turned into RHEL back in 2003, but even they had the courtesy to offer a free alternative, Fedora. Then came CentOS and rebuilt all the source code, and you got RHEL without the price of RHEL. Canonical, on the other hand says nope, pay up or else.

Yes, that “else” is exactly what’s going to happen if they really stand by this decision. After the Red Hat takeover, CentOS tried to pull the same crap with its CentOS Stream debacle, and what did they achieve with that? Bingo, everyone switched to Rocky, Alma, or Fedora. And in the case of Ubuntu, it’s not even necessarily the money it involves – it’s the maintenance burden, the deployment nightmare. It’s Windows activation and KMS in disguise – it’s a pain in the ass, there’s no other way to put it.

We love Linux deployments because they’re quick and simple to deploy, reset, and dispose of. Because I don’t have to worry about activation, licensing, expirations, and keep inventory of a truckload of keys and their corresponding instances and their numbers. This move eliminates all these advantages.

So Ubuntu, thanks for all the fish, you just made another step to become completely irrelevant. I can only guess how long before Ubuntu has its own CentOS moment.

]]>
Using Docker in WSL https://noobient.com/2023/04/10/using-docker-in-wsl/ Mon, 10 Apr 2023 18:37:56 +0000 https://noobient.com/?p=2312 Read more »]]>
  • Enabling systemd in WSL 2 on Windows 11 22H2 without Insider
  • Using Docker in WSL
  • Using systemd in Docker
  • The saga continues, we’re still trying to get to the perfect Ansible development environment on Windows using Windows Subsystem for Linux (WSL). Now WSL should have working systemd, it’s time to set up Docker.

    First of all, install Docker Desktop for Windows, obviously. Once done, go to Settings, and make sure you’re using the WSL 2 based engine:

    Then also ensure it’s offered inside your WSL distros. Go to Resources / WSL integration. Here you may enable it for the default distro, or individually for each distro installed:

    Then open a terminal for your WSL distro and check if the docker command is available:

    $ docker --version
    Docker version 20.10.24, build 297e128
    

    If you get -bash: /usr/bin/docker: No such file or directory, ensure that Docker is indeed running on Windows, because it “injects” these executables into your WSL distros:

    $ ls -al /bin/docker
    lrwxrwxrwx 1 root root 48 Apr 10 20:50 /bin/docker -> /mnt/wsl/docker-desktop/cli-tools/usr/bin/docker
    

    Therefore, if Docker Desktop isn’t running, you can’t even call these executables inside WSL.

    Now try to run a container. I can’t really remember if this was done automatically or not, so you may or may not get the following error first:

    $ docker run -it --rm ubuntu
    docker: Got permission denied while trying to connect to the Docker daemon socket at unix:///var/run/docker.sock: Post "http://%2Fvar%2Frun%2Fdocker.sock/v1.24/containers/create": dial unix /var/run/docker.sock: connect: permission denied.
    See 'docker run --help'.
    

    In this case add yourself to the docker Unix group first:

    sudo usermod --append --groups docker bviktor
    

    Then exit your existing WSL shells, reopen them, and try again. Now it should be good:

    $ docker run -it --rm ubuntu
    root@1e227a618e3f:/#
    

    Oh yesss. There you have it, Docker integrated with WSL. Stay tuned, we have a lot more to get to that Ansible Galaxy development environment.

    ]]>