Weeknotes 202

Vibe: Despondent.

Someone fly-tipped their Christmas Tree (a fucking CHRISTMAS TREE in May!) in our front garden during the night of Saturday to Sunday. It’s bad enough getting rid of your own, nevermind someone elses. Fuckers.

I briefly tried out Tidewave this week on an Elixir Phoenix codebase. It was easy to get going with Claude Desktop but works with various editors. I was impressed until it stopped responding leaving me in limbo, but it was magical for a while.

Remember my Game Boy parts were stolen from my doorstep?

Well, after some brooding, I decided to re-order the same items again. And so AliExpress got even more of my money. To avoid the parcel being stolen again, I decided to ask Evri to deliver the order to a local Post Office for me to collect instead. I eventually received a notification that it was ready, so along I pop only to discover that the Post Office is now apparently “closed until further notice”. It appears they may have ceased trading. And they have my parcel. Inside.

What happens now is anyone’s guess, but this is the second parcel with the same contents that has failed to reach me. Is this a sign?

The idea of not using Google for web search was once inconceivable. Google completely dominated. I realised that I’ve been using Kagi full-time since July 2023 and I don’t miss Google at all. Kagi is not as good as the glory days of Google, but with A.I. arriving the glory days feel long gone.

My advice to you: keep your computers up-to-date. Of course, the problem with keeping computers current is just how easily software breaks. I suspect I’ve swapped a lot of fixed bugs for a lot of new bugs. So it goes. A double-edged sword.

I used Ubuntu 18.04 (Bionic Beaver) when I set up my home server back in 2018 not for any particular penchant for Ubuntu, but because it seemed easiest and well-supported as a major player distro, and now it needs to be ugraded because it is no longer supported. And hasn’t been for a while; don’t judge.

In order to get to where I wanted to be, the latest LTS release of Ubuntu, I had to upgrade several times from version to version.

There was a slightly scary moment when going from 18.04 to 20.04 where my ZFS pool briefly disappeared. The data stored on the pool never seemed at risk - I have a lot of faith in ZFS to do the right thing - however, I don’t have a lot of faith in myself to not take an interesting situation and turn it into a disaster. I took my time.

zpool status showed the pool was missing:

$ sudo zpool status
no pools available

But zpool import thought the pool was online:

$ sudo zpool import
   pool: storage
     id: 13125165091315077677
  state: ONLINE
 status: Some supported features are not enabled on the pool.
 action: The pool can be imported using its name or numeric identifier, though
        some features will not be available without an explicit 'zpool upgrade'.
 config:

        storage     ONLINE
          mirror-0  ONLINE
            sda     ONLINE
            sdd     ONLINE
          mirror-1  ONLINE
            sde     ONLINE
            sdf     ONLINE
          mirror-2  ONLINE
            sdc     ONLINE
            sdb     ONLINE

After some research (typing “send help” into Claude is research), a simple sudo zpool import storage -f fixed my issues, and the pool was back. Subsequent upgrades avoided breaking ZFS further thankfully.

Fair play to Canonical though, baring that ZFS hiccup the upgrades went fairly well with no major breakages as I’m currently aware. There were a few things to tidy-up but almost everything is working as before.

Anyway, I’m now running a distro with support until at least 2029.

During my one of my Ubuntu upgrade cycles I was amused by this message:

You have to download a total of 797 M. This download will take about 1 hour 41 minutes with a 1Mbit DSL connection and about 1 day 6 hours with a 56k modem.

56k!

Finished The Eternaut. It got a bit weird, and a bit less good, but I did enjoy it overall and will watch Season 2 when it is released.

In other ZFS news, a new zfs rewrite command was implemented.

For years users were asking for an ability to re-balance pool after vdev addition, de-fragment randomly written files, change some properties for already written files, etc. The closest option would be to either copy and rename a file or send/receive/rename the dataset. Unfortunately all of those options have some downsides.

As I understand it, this should help in particular with RAIDZ vdevs as you can now just add another disk, and rebalance your data across them, which was a big limitation of RAIDZ setups, and the reason I went for mirrored vdevs.

Sickness this week with a 3 day cold. May I be so bold as to say Jakeman’s Peppermint are the best.

I had the misfortune to need to create an Amazon AWS IAM user this week. You know those “data breaches” we hear about fairly often? I guarantee most of those are directly related to poor Amazon S3 bucket policies, and Amazon must share the blame for creating an absolute mess of a system.