So I bought a Native Planet Tellurian. Obviously in black (Henry Ford would be proud). Would have felt bad if I got a different color seeing as how I convinced them to add black to the otherwise vibrant color lineup; that said, the yellow/red/blue colorways are all quite dope as well.
Up front, I will admit that I am a total shill for the Native Planet guys since I think they are working towards the one true way to run your Urbit (for the sovereign individual: on hardware you own and control), and that ~dalhec-banler surprised me with a little bit of extra RAM in my box (are we still doing phrasing?) that I didn't actually order. But, I don't mean shill in the sense that they are paying me anything (they aren't, but maybe they should?) or that I have any equity in their business (missed the boat on that one..). Just in the sense that I am super stoked on what they are doing and will do my part to pump their bags (and therefore the prospects for individual sovereignty in the world's digital future).
That said, this writeup isn't for telling you how awesome they are. Rather, it is my experience in unwrapping, booting, and otherwise running a ship using their hardware (The Tellurian) and their software (ColonyOS, GroundSeg, and StarTram). I'm writing this in mid-March, so depending on when you are reading this it is likely that some of my gripes are already fixed or otherwise will be getting solved shortly (probably because I am raising a stink about them here). Anyway, I spend $550 bucks and in like 3 days I had a fancy little package on my doorstep just waiting for me to play with it.
The packaging for the Tellurian was straightforward and no-nonsense. Solid quality external cardboard box, akin to what you might expect from a Framework laptop or fancy mechanical keyboard. Inside the box was the Tellurian wrapped in bubblewrap and light corrugated cardboard. Nothing fancy but once you pick up the Tellurian itself and realize it is sturdy enough to use as a bludgeon against your greatest enemies, the packaging is self-evidently sufficient.
Also in the box are the power brick and cord so you can plug it into the wall. To be expected, but certainly you would be sad if you got this beauty and needed to wander into a hellscape like OfficeDepot to get a power source. On the top flap of the box is a QR code and url for how to get started. No pamphelts, no 'WARNING: Your offspring can get electricuted if they piss on this hardware'. Very Apple-eqsue.
Less Apple-esque is the hardware itself. Your motherboard hasn't been superglued to your battery and boobytrapped with warranty voiding stickers; everything is exposed and available for you to tinker with. As I've alluded to, the Tellurian is robust. I think the body is aluminium and it makes no attempt to pretend that it is lightweight, but its not tremendously dense. Wholly appropriate for a device intended to sit on your desktop or to be mounted on a wall. Sure, also acceptable for a device you will hide away in a close (or bolt the back wall of a gun safe?), but visually the Tellurian is a thing of beauty so you really needn't hide it away somewhere.
There are a variety of USB and DisplayPort connections to the motherboard, but you really don't need them. Yet, at least. Ethernet is also available if you prefer that over wifi. Which as we get into setup, you'll learn is basically unnecessary even for headless setup.
I'm cool so I got serial #0002; you probably will get something like unit one-trillion or something so don't try to compare yourself
Step 1, plug in the power (step 0 is to attach the antennas; the instructions mention to not attach them while the device is powered on because... reasons). What you don't have to do is plug in ethernet. I'm sure you could and it would Just Work, but I wanted to make this as hard on myself as possible so I left all my ethernet cables at the office and aimed to set up the Tellurian at home. I went to the site from the QR code in the box (nativeplanet.io/startup) and loosely followed their instructions (read: kinda skimmed the bold parts and then went back and read more closely when things didn't work first try).
That meant using my laptop (although you can also do this step with a phone apparently) to connect to the default wifi network broadcast by the Tellurian and then connecting directly to it. They had a nice interface to select my home wifi network which worked smoothly but when it disconnected from the tellurian generated network that action shut down the site immediately. This was a little disconcerting because I didn't know if/when the connection would come back. I'm decently patient and from a technical perspective had more or less expected this to happen, but it is always a little jarring when your experience goes 'kaput' and your next steps are unknown. I reconnected to my home network from my laptop and waited for something to happen.
Eventually I was able to navigate to http://nativeplanet.local, but when it loaded the UI complained that groundseg was not running on the device. After a little whinging, I had the genius idea to refresh the page. Upon refresh it prompted me to generate a new password, which I did easily (like the adult that I am, I used my password manager to generate a secure password). After entering the password the first time, it added another form field to confirm the password. It probably would be preferrable to me for that field to already be there so that I know I'll need to confirm the proposed password. This is a typical pattern so it shouldn't be intimidating to a new user, whereas fields appearing while filling a form can give a send of dread, aka "How many more fields until I am done with this shit".
After creating a password I was prompted for my startram code. I searched for Native Planet in my email but nothing came up so I searched for startram and it pulled up the registration code. Probably would be nice to have matching keywords around all NP email material so that generalized search terms would pull up everything at once without needing additional specificity. Back in the GroundSeg GUI, I hit the 'register' button and got a nice little Registering your key.. loading indicator with a soft blinking.
This was very calming, but after ~30 seconds I started to be concerned if it was working well. Since refreshing the page had worked on the initial rundown/restart for setting up the wifi connection, I was very tempted to do the same here but was unsure how that might effect the process. Probably would be nice to have some sort of indication of how long this process is supposed to take or what to expect to happen next.
I eventually got impatient and, given my success with refreshes in this setup, I hit refresh and it asked me to set up a new password again. This had me lose a little faith that it stored my state/responses as I continued to proceed through the process. So I called for help.
I sent a message in the Native Planet urbit group and got a response from ~sitful-hatred within like 3-5 mins: "hang tight, i think the controller container got stuck so i nudged it"
I also got a response from ~nallux-dozryl about submitting a bug report. I am not afraid of such activities (particularly with companies that I know will actually do something about logged issues), but it took me a moment to find the bug reporting interface. It's triggered by clicking an icon that looks like a bug. Sensible if you know where it is, but I might suggest making that bug reporting button a bit bigger and more explicit. And/or having it pop up after X amount of time on a given screen.
After skipping the initial setup option and going to the StarTram setup menu from the home screen, I successfully got my StarTram code registered . This was also at the same time ~sitful nudged the control container so I'm not certain where it was broken, and I don't know what the proper setup experience is for those that go directly to remote access.
From my laptop that I was using to access GroundSeg, it was super easy to use %houston from Assembly Capital to spawn the moon ~nattev-palnet-sarlev-sarsen, download the keyfile, and then drag n' drop the keyfile into the groundseg UI to boot a ship.
Not sure if this would be similarly straightforward on mobile, though. While I was waiting for the ship to do it's initial boot, I figured I would wander around the UI for a bit. There is a random unlabeled toggle in the top righthand corner of the control panel, and being the curious guy that I am, I obviously hit it.
It shut down my ship. Given urbit's nature, not a huge issue but it was a little disruptive.
While my ship was booting, I poked around looking at the logs (which seems to just be a dojo output?) and the settings. I didn't adjust my Pack and Meld since I know what it does it the weekly occurange would likely work, but I imagine someone less deep down the urbit rabbit hole might not have any clue what that is; could benefit from a tooltip.
Also, the last meld shows as occuring in 1969. Instead it should be something along the lines of "never" for a ship that was just booted.
The settings area does seem quite useful. For example the opportunity for exporting logs seems very nice and potentially useful for more than just NativePlanet use cases. Other elements might need a little more explaining, though. The MinIO section was empty and didn't seem to have a way to interact while my ship was on 'local' mode. Later on once I got fully set up, this area became much more populated. But I wasn't there yet.
Instead, it turned out that my input for startram registration didn't quite work. It recognized my registration code at let me know that it is good until 2/28/24, but it hadn't actually fully made the connection. ~nallux-dozryl and ~mopfel-winrux hopped into the Urbit Hacker House and troubleshot with me. It turned out the issue was mostly just the fact that I thought it was working and it wasn't. I think ~sitful nudged some things on his end as well. I just tried entering my startram code again and it worked like a charm. If you are having issues, I don't know what to tell you because it works on my machine. But also, just hop into /1/group/~raldeg/nativeplanet and you will get excellent customer service. As an extra note on startram subscription duration, it seems like it triggered based on purchase date, but my expectation as a user is that it would start based on initial registration date.
While talking with ~nallux and ~mopfel, I set up Minio since it was activated once I had StarTram properly running. I likely would have preferred to see the options for these settings in some subduded/unactionable state in settings with some tooltips on what the functionality does and how to get access to them (i.e. 'you need remote access for this feature, get it with StarTram: Buy Link'). Relatedly, I tried to get into the MinIO Local Storage Console, but it has a login window for which I have no idea what the credentials are (maybe the password is the one I set, but WTF is the username?) That said, it seemed like the s3 integration with urbit was working smoothly so I'm not sure that it matters? I even just quickly installed %silo and everything worked swimmingly!
A few days after I got fully set up, I saw a random chat from ~nallux in the Native Planet Group: "Username is your @p and password is the password you used when setting up minio"; Of course, glad I (and incidentally, you) know this now, but would be good to have this called out somewhere in the docs or GroundSeg GUI.
As a side note, at this stage in the setup process, the number of tabs i have open has increased a decent amount, and the all black favicons don't show up too good on dark mode browsing (on Brave, at least). The white groundseg one shows up fine but giving the nativeplanet.io one a white background could be nice (keeping them different definitely recommended). Of course, this feedback stands for any website/application that is using black favicons. Looking at you, Groups and Talk.
All this took me about an hour from unboxing to "I've got a ship up and running that I can access remotely and use to poast dank memes with my friends". I was going to quit and go to bed, but then ~nallux-dozryl challenged me to set up a custom domain, so I continued marching on. To Native Planet's credit, they have included a link in the GroundSeg UI (right next to the custom domain element in the settings panel) to instructions on their site for setting up a custom domain, so it was really easy to figure out next steps.
I don't know what the overall plan for custom domains is, but it almost seems like it would be possible to connect with some registrar that does some affiliate/refferral revenue and make the instructions cater to the interface for that specific partner. That way people who haven't yet gone down the domain ownership rabbithole might generate some additional potential referral revenue via people who want custom domains, without actually incurring additional cost for Native Planet customers.
Personally, I already own way too many domains and so I just hopped into Namecheap, set the DNS records as indicated in the instructions, and boom! Easily able to access my ship at nattev-palnet-sarlev-sarsen.rooftopdao.io. And what is extra nice about this is I don't have to give up the full domain to GroundSeg, but rather can just carve off whatever subdomains I want to use while everything else can keep pointing at my server.
Since Urbit ship custom domain setup was crazy easy and no issues whatsoever, I expected the same experience from the S3 custom domain. But somehow S3 setup didn't go quite as smoothly. I floundered a bit, worrying that I'm an idiot. As a worthy note, there were not separate instructions for s3 setup and I made a few assumptions (i.e. I should enter s3.nattev-palnet-sarlev-sarsen.domain.tld). Those didn't work, nor did trying to reverse my assumptions. So I reached out to the Native Planet guys again.
~nallux quickly (like, under 3 mins?) confirmed that I did the data entry correctly, so it was an open question as to why this worked for the ship custom domain setup and not the S3 custom domain setup. After a little back and forth with them, we figured out that this part of the new custom domain feature just hadn't been pushed to master. ~sitful did exactly that and tried again after he made updates and it worked like a dream. Mind you, this all took less than 20 mins.
After making the S3 bucket connection, I was having issues with the upload working in the Talk UI while it worked fine in the Groups UI. Troubleshooting with ~finned-palmer and ~sitful and we eventually narrowed it down to cacheing catching me slipping. After a quick hard refresh of Talk, everything returned to working as expected!
From top to bottom, including support time--and getting hotfixes pushed--from the highly responsive Native Planet team and my own commiserating, setup took ~2 hours. Probably about half of that was due to my being one of the first people to put the whole flow through it's paces, and the issues that took longer to fix (StarTram registration, S3 custom domains) had actual bugs that have since been fixed so other people are unlikely to experience the same issues. There are probably a variety of GUI tweaks, tooltips, and walkthrough refinements that could be made to make the novice user feel more comfortable with the process, but overall I think if I did it again I would be able to do it in under 45 mins including ship boot time.
In the days following initial setup, I went to various different locations and tried to make heavy use of the ship. So far everything has been equally performant to my main planet which is running on a way-overpowered VPS. I have installed the heaviest apps I know of, i.e. Uqbar's ziggurat and ~midlev-mindyr's Astrolabe, as well as Holium's Realm apps and have yet to have any performance issues.
I am going to slowly start stacking additional ships into this box, and eventually my plan is to run ~sarlev-sarsen off it as well. Most of my hesitation about moving my pure and unbreached planet to the Tellurian is in potentially fucking up the pier transfer so I want to get some dry run's under my belt before transfering more precious cargo. Performance wise, though, I am confident in the ability of the box to far outpace what urbit/vere/arvo are capable of at present.
I imagine many of my above stumbling blocks will be fixed in short order, as they are minute and eminently solvable by the team at Native Planet, so I'm going to share a list of things that I would love to see in the future. These are only vaguely influences by informal conversations I've had with the various guys at Native Planet, but hopefully we can meme them into existence:
~sitful-hatred and ~mopfel-winrux were the guys that used umbrel to make it easy to sidecar a bitcoin node to an urbit ship; would love to see them pull that over to the Native Planet landscape, specifically on the Tellurian.%eth-provider and not have the whole network running off Tlon's infura endpoints.Regardless of which of these things come first, my initial experience with the Tellurian's setup and operation has been excellent and it demonstrates the Native Planet team's ability to focus on their customers' needs to deliver an awesome experience in the present for the urbit end user who wants to properly own their digital experiences without accidentally becoming a sysadmin.