Hacker News new | past | comments | ask | show | jobs | submit login
This Blog Is Now Running on Solar Power (louwrentius.com)
227 points by louwrentius on July 10, 2020 | hide | past | favorite | 120 comments

Imagine a Raspberry Pi secured onto a tree somewhere being powered by a solar panel and serving a website over 5G.

Imagine that Pi is running a SaaS product that generates revenue and pays for its own 5G subscription.

Now imagine that Pi is earning Bitcoin, pays for its 5G in Bitcoin, and puts up job postings online to get people (paid in Bitcoin) to provision and set up more replica Pis on other random trees in town.

I'm sad to see that this crypto-comment is now the top comment on my build. The one topic I really don't care about.

I just hope people will may be incentivised to build fun projects like these, copy my example and improve on it or create a variant.

Stats for mobile users: https://louwrentius.com/solar/index.html

Regarding the crypto stuff, please read this: https://louwrentius.com/cryptocurrencies-are-detrimental-to-...

I’m sorry, but you don’t seem to understand bitcoins intrinsic value proposition.

Loved the solar powered blog though ! Im looking at setting up my own now, thanks to your post.

Ethereum's upcoming 2.0 release kinda solves the issues you raised there.

1. Energy waste: solved by staking instead of mining. See proof of stake [1].

2. Centralized mining: staking in Ethereum 2.0 should be possible with a Raspberry Pi 4, therefore the viability of joining the network isn't closely tied to your geographic location as it is with mining (mining is only profitable with cheap electricity & hardware, which is why China is dominating Bitcoin mining right now).

3. Facilitates crime: well, that's still a problem, but it's also a problem with other things that give anonymity. Is Tor detrimental to society? Is cash detrimental to society? Should we force everyone to stop transacting with cash since most crimes are done with it, and like China to force them to transact digitally where it can all be tracked by the government, banks and corporations?

4. No intrinsic value: you should check out what's going on with DeFi (decentralized finance) right now in the Ethereum world. Over $2B is locked into smart contracts, people aren't just HODLing anymore, they're playing. Sure, it's not used to save the planet just yet, but I definitely believe it will develop to be enormously beneficial to society. Would love to elaborate if you want.

5. Stability: solved with stablecoins. Basically, you transact with a token that keeps it's value stable against a real world currency. DAI is a good example of a USD stablecoin. This defeats the "ponzi scheme" argument as well, since you can interact with many smart contracts with a stablecoin instead of a cryptocurrency. Your choice. If you acquired a cryptocurrency (which is somewhat like a stock) instead of a stablecoin and lost money, it's your fault.

I really like your article, well said. The issues you've raised are still real and need to be discussed. Nevertheless, I argue that those issues can, and are, being solved...

At least 4 of these issues had no solution 10 years ago, but just like innovations (proof of stake, stablecoins) fix them, further innovations will slowly fix other unsolved issues as well.

[1] https://eth.wiki/en/concepts/proof-of-stake-faqs

Reality check time!

- Raspberry Pi is way underpowered to run any SaaS on.

- There are no 5g providers that would accept bitcoin as payment. Highly unlikely there ever will be any.

- Bitcoin's network is too congested for it to be used for trivial payments such as subscriptions for SaaS.

- A bitcoin node takes hundreds of GB of storage itself. Bandwidth costs will probably be significant too. This means the SaaS would probably have a hard time getting to break-even.

- It may be more profitable to climb up the tree and steal the pi and then take its bitcoin private key (assuming there is a hot wallet). Competing SaaS may also look for rivals and take them down.

> - Raspberry Pi is way underpowered to run any SaaS on.

Probably sufficient if it's written in Go or another compiled language. Obviously depends on how much revenue you're making per hit, but certainly within the realm of possibility.

> - Bitcoin's network is too congested for it to be used for trivial payments such as subscriptions for SaaS.

There are alternative coins that would work quite well, if fees/txn is the concern. Current fees seem to be about $1.15/txn, which is quite high if you're only charging $5-10.

> - A bitcoin node takes hundreds of GB of storage itself. Bandwidth costs will probably be significant too. This means the SaaS would probably have a hard time getting to break-even.

Few people run a full bitcoin node, so I think it would be sufficient for this use-case to send txns to the network via a third-party service.

> - It may be more profitable to climb up the tree and steal the pi and then take its bitcoin private key

Honestly the human factor involved in the whole scheme is probably the hardest thing to make happen. There's maintenance, h/w failure, the difficulty of finding trustworthy contractors who don't need human guidance, etc.

Good points. The only thing I'll disagree here is that using Go doesn't make it magically faster. The SD card is also a performance bottleneck for any database intensive work. Although, some services may still be viable such as a VPN service maybe?

> - Raspberry Pi is way underpowered to run any SaaS on

Our MVP SaaS product was done on a Pi (Node.js backend), and several of our launch customers were using it before we deployed them to production. Helped us simulate micro instances and write more efficient code as we started to scale.

Definitely doable, but not for something as crazy as running Asana or Zendesk over.

> Raspberry Pi is way underpowered to run any SaaS on.

Depends. Here's a random SaaS which I assume is making money:


A URL shortening service. Yeah, I'm thinking this could fit on a Pi. Database and everything on the SD card. Latency won't be great, but it should be possible, especially if you rewrote it in Rust.

It doesn't have to run a Bitcoin fullnode. It can run a SPV wallet which would take a few megabytes if that.

> take its bitcoin private key

Generate and store the key in a Hardware Security Module

The biggest killers are related to BTC. It would be difficult to pay for 5G with BTC (although there probably could be workarounds like how Purse.io has humans with credit cards buy your Amazon shopping carts in exchange for your BTC) and also difficult to get subscription revenue in BTC. Something like Libra could solve all these problems.

> Raspberry Pi is way underpowered to run any SaaS on.

Now there's a sentence that triggered a lot of anecdotes to prove the contrary lol!

I gather that while it is indeed possible, I have my doubts as to how practical or economic it might be. Anyone willing to accept the challenge? ;)

Who would pay for a URL shortening service, especially one that can lose your URLs?

Maybe? But only if it ran on a tree and was powered by the sun

Dream killer. Username doesn't check out. =]

I'm sorry, I was nodding along when I stumbled over the 'bitcoin' word. That's a big no from me and the reason is probably not hard to guess in light of this project.

"Can you please show me the benefit to society of cryptocurrencies?" Sci-Hub? Wikileaks? Medical Marijuana?

If you're concerned about energy usage, there are much bigger fish to fry than Bitcoin mining. But if you want, s/Bitcoin/Libra. A computer can't get a bank account, but it can get a crypto wallet.

I'm opposed to the whole concept of cryptocurrencies.

Google 'cryptocurrencies louwrentius' to understand my position.

From your article:

> Cryptocurrencies have no intrinsic value. They are only worth what people are willing to pay for them.

How is this any different from US dollars, gold, or anything besides food, medicine, and shelter?

> Cryptocurrencies facilitate crime

Once again, the same argument could be applied to US dollars

Precious metals have been sought by humanity for ages because of their strong resistance to oxidization and fiat money is backed by governments. There is also a strong oversight on both of them by various organizations. These are the reasons why people are usually willing to pay for them.

At the moment, I do not think major cryptocurrencies have any of those qualities. Though it could change if all major governments have a favorable view towards it.


US dollars are the only accepted means to pay taxes. This is the big thing that makes national currencies work. Having a government behind a currency may not be a sufficient guarantee of value but it is, at least so far, a necessary one.

For bitcoin at least, it's an intrinsically deflationary currency. There's no incentive to spend a bitcoin today if it will be worth more tomorrow, so it ends up being primarily a currency of choice for transactions which are unfeasible in any other way, which ends up being primarily illegal transactions.

>US dollars are the only accepted means to pay taxes.

I agree with this, of course. But there is also a method that allows me to convert my AUS dollar in to US dollar. I just paid my taxes with $AUS?

Does the logic follow: $BTC->$USD->USTAXES ?

Your ability to convert your Aussie dollar holdings into US dollars with ease is contingent on there being a relatively stable demand for Aussie dollars, which are the only legal means of paying taxes in Australia, and settling debts denominated in Aussie dollars. Cryptocurrencies are not a legal means of payment of taxes and are not created as debt so lack that guarantee that people will want to exchange significant quantities of USD for that coin [I guess in theory some cryptocurrency could have the property of being emitted as debt, but I'm not sure decentralisation and credit creation go well together].

I think you missed the point I was trying to make. I haven’t had my coffee so that’s my bad.

As long as there is a means of conversion you can start with anything and end up being able to pay your taxes. You can’t directly pay taxes with your time and skills, but you can trade your time and skills for USD, and then pay your taxes. You can’t directly pay your US taxes with the AUS dollar, but there is a market where you can exchange that for USD, then pay your taxes. The logic is inclusive of BTC, which makes the ideology that BTC is bad because you can’t pay your taxes with it, at very least, murky.

Sure, you can convert anything to anything if you can find someone else that wants to complete the trade at a suitable rate.

Taxes provide [part of] the reason to believe a market to exchange Aussie dollars will exist in two years time: tax and Aussie-dollar denominated debt mean a lot of people need Aussie dollars. That's the only reason the Aussie dollar is worth more than the plastic its printed on. BTC doesn't have that reason [to any meaningful extent; credit is a tiny part of the economy and nobody needs BTC to pay taxes] and unlike the fruits of your labour [which I'd hope are more useful or entertaining than an alphanumeric string even if they can't be exchanged in future], it doesn't have any non-exchange-related reason to be worth anything either. The some-people-need-it-to-pay taxes argument is why Aussie or US dollars aren't bad despite being as intrinsically useless as BTC.

So the argument is that BTC is bad because both intrinsically useless and almost entirely lacking a mechanism [similar to taxes] to create a stable flow of future demand to acquire it

>US dollars

You can pay taxes with dollars and buy oil anywhere in the world. And the US government will send the strongest army in history against anyone who will try to decline US dollars for their oil.


Gold is naturally rare and has a much wider network of places that will accept it as payment.

>food, medicine, and shelter

Because you can use those for actual real-world things? How is that even comparable?

Food, medicine, etc have intrinsic value. That's what the OP is talking about. The US gov't says that the US dollar has value and is willing to back it up. That is extrinsic value.

I agree with that but these are specifically RPIs running on solar power (which, btw, will be useless for mining btc), so they will be completly wasteless (the energy goes to waste otherwise). The only exception would be if you somehow rose the cost of mining for others, which would not happen since you're using pis to mine btc. Except I guess the environmental cost of producing these pis, but unless you buy tens of thousands your purchase would be a rounding error

locally no impact, globally supporting a system that burns enormous amounts of power (of which much is co2 producing).

would be a cool little project, and it falls to you to decide which side of the ethics of it you subscribe to.

I would love to have a little cryptocurrency miner, completely self sufficient, just doing its own thing until it dies. but I would also love a cryptocurrency that didn't have the same kind of environmental impact that bitcoin has. maybe another application? selling computing power?

Every project is a global rounding error environmentally, that doesn’t mean they’re harmless. The real yardstick is simply a cost vs benefit equation as we can’t mitigate everything. On that scale a project trike this is extremely harmful as the benefit is nonexistent and the harm is real.

Are you also opposed to automobiles?

I am, or at least I think they should be used sparingly.

They destroy the planet. And on a personal scale, over-reliance upon them encourages lack of exercise and a needlessly stressful life spent in traffic.

I reckon we could make a big dent in pollution and obesity rates by having most people reduce their car usage.

You inclined at all to explain where you see these two being comparable? I’m lost as to what it is you’re trying to say with this statement.


Sorry of my brevity caused you any frustration but I find it’s usually easier to ask questions and seek mutual understanding and clarity rather than make assumptions and accusations about intentions and torpedo the conversation before it’s even got going.

Check out LibraryBox/PirateBox. While not typically featuring a cellular connection they're at least an independent host for files/wiki/whatever and meant to be self-sufficient.


> pays for its 5G in Bitcoin, and puts up job postings online to get people (paid in Bitcoin)

'Cryptocurrencies' would have been a better mention that I would agree with rather than just only to use Bitcoin.

In this day and age, It's a good time to be in the business of building and selling these shovels i.e Solar Panels to people who aren't able to do this themselves.

Yup that is what 'Joe Armstrong' @joeerl once suggested' https://twitter.com/joeerl/status/1055808039369039872%3Fs=20...

It the Pi is providing a useful service that people are paying dollars for, that it then uses to pay for its 5G subscription, why introduce intermediate steps of converting its money to bitcoin, gold boullion, or pirate treasure, in order to pay someone?

I'm imagining that the Pi is an independent agent, not owned by a human, and therefore can't get a bank account. The Pi is paid in Bitcoin to begin with, so no conversion is necessary. The assumptions are that there is a 5G service that can be bought with Bitcoin and that the Pi can supply a lucrative enough service to get paid enough to reproduce itself.

Wanted to put solar powered sparse mesh networks in forest as emergency trackers..

Lorawan. You want lorawan.

LoRa as a Data Link and thr RadioHead library on top of it is probably more flexible

Not to nitpick, but wouldn’t Etherium Be the right cryptocurrency? I try to ignore it, but I remember that that’s the “smart contracts” one, so it could make the job contracts in such a way that if they are completed, payment is automatic or something?

How would the contract know that the job is completed? Ethereum contracts can only access events on the chain itself. There is no way to go out to the network and, say, ping the new Pi, because there is no way to cryptographically verify that this happened. So some off-chain server somewhere has to issue a command (with some private key) to the contract to say, "Yes, I checked that the work was done, you can disburse the funds". In which case you never needed the contract anyway and just could have sent the money directly.

You know there is this device thats much faster, powerefficient and has 5G - a phone or a tablet! You could write write an app for Surface X or similar.

Anyway, i am just fighting the Pi-obsession, there are much more capable devices out there.

Imagine a similar device, embedded into tombstones and monuments, forever serving hotspots with content to remember to the dead, maybe even running their archived blog and content.

Sounds like Universal Paperclips!

Also fix bugs, review commits, etc :)

WARNING: before anyone gets overenthusiastic, please observe that adopting this setup probably caused more harm to the environment than it prevented. Despite their romantic reputation, photovoltaics aren't green - they cost a lot of (usually dirty) energy and precious metals to manufacture, and are not renewable devices, so you want to make them count, i.e utilize them to a significant fraction of their potential, especially so on a grid like the Netherlands. This is not what this project is doing.

I'm absolutely sure the author means well and it sure is an interesting exercise, but this is not what transitioning towards green IT looks like. Sorry if this disappoints people's expectations, but these issues are more and more overlooked.

(Can't give you numbers rn as I'm on my phone, but you can work them out from e.g solar electricity emission factors)

> not renewable devices

If this means recyclable, that's slowly changing.

I like this project a lot and am inspired to do something similar. I'm a liveaboard boater in the UK, I have a 55 foot narrowboat with 350w of solar on the roof and a 550ah battery bank. I could probably run my own little server similar to yours pretty much 24/7 as when I'm not using solar to charge the battery bank, I'm using the boat's engine to recharge.

So not quite as green but during the months from April to September, the solar panels are enough to charge the batteries every day.

My modem/router uses 4g, it's a Teltonika rut950 that is using OpenWRT and has GPS built in which I've been keen to make use of.

I'm definitely inspired to do something like you've built!

Thank you, I would recommend to go for it. Would be nice to tie the GPS into some blog that would track where you are (aproximation for privacy reasons maybe).

A floating solar-powered server, should be nice!

Floating solar powered server buoys, anyone?

Is it always on? This is really cool. I like the idea of self-sustaining electronics(robotics in particular).

I wonder if there's any gain using a UBEC or something vs. the car cigarette lighter usb thing you're using to jump from 12V to 5V.

Edit: yeah I was going to say, I think I saw someone do a toolbox build before where it was solar powered/had a GSM module in it for communication... but I don't think it was a website, still neat.

Thank you, yes it is always on.

I already have replaced the car cigarette lighter usb thing with a proper buck converter but it reduced power usage with maybe 5% or less.

I need to update my article on that.

Other thing I was wondering about, I saw you have a percentage on battery what about "hours left". Not sure if that's easy to accurately estimate.

Even the battery % is a guesstimate, but I think it's possible. It's a nice idea. Thank you.

For self-sustaining robotics, do they oil themselves?

You probably don't need any lubricant for an exceptionally long time.

Some of us old-timers where pretty grumpy when bicycles switched to sealed bearings. They aren't perfect (more rolling resistance), they don't last forever, but they do last (stay greased and unfouled) a hell of a long time and require less skill to replace.

Haha, yeah it was a bit vague.

I have fantasies about building small drone submarines and releasing them in the ocean. Those I don't know how they'd sustain themselves. A little ground rover that has panels on it sounds more realistic. And yeah the life span I don't know how long they would last for.

But on a farm or homestead I would like to have little solar-powered wheel-driven drones pretty much that run around and do their own thing.

It would be neat though, have a sat module on the drone submarine, and then the sub surfaces at night. I think at my local lake is more realistic. Of course in reality these things would operate in supervision/not break laws. But I saw an intriguing video [1] before and I don't know... it's personification I guess.

[1] https://youtu.be/0iDBF23gI6I?t=52

"little solar-powered wheel-driven drones"

You may be interested to learn that a company is already doing this, with the "Tertill", a small solar powered random-walk weed eater.


Yeah I saw that recently I think it's cool. The wheel design is interesting, I don't know how effective that little weed whacker it has is. I was also surprised how small the panels seem.

edit: I follow this one guy on YouTube he built this thing [1], that's the basic idea. It doesn't really have a purpose, but it's cool.

In the long run though, imagine you're gone(passed on) and you built some robot that's out there still doing its thing. Probably talking RTG/space more like, than something on Earth that erodes.

[1] https://youtu.be/nv2FbwjIZRE?t=8

All this is pretty hard. I recommend the rctestflight YouTube channel. Really useful if you want to build any robot for real.

As stated in the blog, my solar panel produces way more power during 'peak solar' moments than both the Pi and the battery can consume (charging).

So I'm trying to keep my iPad charged up with this setup too (kind of a hobby). So you may notice during the day that the solar production is 60+ watt and the load could be similar.

Nice! A few days ago I dusted off some cheap panels I had lying around and started using them to charge my phone and other devices. I figured since I was WFH might as well put those panels to use. The added bonus is that now that I am charging my phone during the day, I am less likely to use it.

I was also planning on doing something similar to your setup which is why I had the panels in the first place. I have a backyard with decent sun exposure, so compared to your setup, it would almost be cheating.

Maybe a blogroll of solar powered blogs is in order!

Thank you. Please pull through on your project, to start a movement :)

Just reuse an old car battery from craigslist or something to power the setup during the night.

It is a fun little hobby project. I enjoyed building it, although it's quite simple.

I must say that the load on the battery is not equal exactly to the actual load because the solar charge controller doesn't seem as precise in its measurements unfortunately.

It seems that HN is now hitting my Pi 'hard' but it doesn't even register on the CPU (5% across cores).

I do notice a small increase in power usage. I wish I had implemented a mechanism to share my Grafana dashboard tracking my solar metrics. (sorry)

It's a neat project, we added solar based watering to our greenhouse and having grafana as a dashboard was super handy.

Renogy's Rover MMPT chargers speak modbus over rs232 so you can interface it directly with a PLC or other industrial automation tools. The whole thing is self sufficient and was a real blast to put together.

That sounds awesome! Do you have a writeup about that somewhere? Would be nice to share! I love self-sustaining solutions.

Not yet although I've been considering it.

There's a lot of things that are really handy but not easily searchable(like din rails, they're the lego blocks of industrial automation) which I think would be neat to aggregate in one space.

Please come back in about 8 to 10 hours when the sun comes up to see what I did with the icon (currently a battery).

I would like to see Applied Science (Ben Krasnow) to power his website/blog with nuclear power.

I am serious.

Many blogs are probably powered by nuclear power :-)

That's also true of solar power :-)

Can you make the right hand div with the solar status a tad wider? The scroll bar cuts off the units of measure on Chrome + Windows.

I'll look into it, thanks.

Love both this and lowtechmag projects,

cant wait to get the time/excuse to do this for my own personal website which doesn't exist yet lol

Low-tech magazine is really worth a visit:


Aha, finally a GOOD reason for a website to only work 9-5 business hours!

I have to thank Gaston Planté for giving me a way to power my blog when the sun goes down.

Don't let the sun, go down on me...

Nice write-up of a nice project!

BUT: This is clearly missing a power saving mode which displays a placeholder page with a cache expiration time set based on weather forecasts. And an 8bit song "Ain't no content when it rains" to the melody of "Ain't No Sunshine" ;)

Thank you, I will put it on my list for improvements ;-)

Very cool project . Not to be pendatic , there is lot network equipment between request and the cluster which still very much run on non renewable power.

Once starlink is live , perhaps requests between two phased array clients can make networks truly solar .

That is true and addressed as an issue in the blog (to be transparent). It's a problem low-tech magazine also faces and wants to improve.

I'm in an apartment and my solar is in a terrible spot. The solar array is crazy over-sized to provide enough power to sustain the Pi and charge the battery.

I think it's easily doable for a person with good solar to put the entire chain Pi-switch-modem on solar.

I did not mean the blog owner’s equipment which the author covered in the post. I meant the ISPs and transit providers in between the server and any user wanting to see the blog. There is not much we can do to control the power sources in that .

While undersea cables and ISP network switches etc perhaps are efficient on a per packet level, they consume substantial power I would imagine .

Ok, sorry, I misunderstood.

Assuming the Starlink satellites were powered by batteries charged by solar and that they were delivered to orbit by solar.

The first is theoretically possible, the second is not true.

The battery sizes on most satellites are very small due to weight considerations and probably more so in starlink , and is designed to last few hours at best, so how it is charged first time is not likely significant .

I had mentioned elsewhere I am sidestepping the carbon footprint of undersea cable laying to building and launching satellites as they are fixed upfront and not per packet sent.

However I am pretty confident laying cables and digging all the other infra to your home and all the ISPs connects is definitely more carbon intensive than building and sending satellites on reusable rockets.

Have the power requirements of the ground units been published yet? That might be chewing through a lot more power than a terrestrial ISP.

Perhaps it will not be as efficient , however we cannot control the power consumption of the transit on wired or satellite, satellite by its design is going to be solar though. I don’t think ISPs are look at carbon footprint for their equipment at the moment

I am side stepping the entire argument on cable laying v sat launch for setting up the infra . Likely satellites are have lesser carbon footprint than cable laying, but there are lot of variables .

If we consider carbon CapEx - Falcon 9 is 500,000 kg of fuel. For that amount of fuel you could lay down a trans-atlantic cable. Then conside that sattelites still need to be serviced by large base-stations across the globe linked by optic fiver.

Then consider that a single cable lasts much longer than sattelites do in orbit and carries more data.

Then conder the fact that radio transmission across almost a thousand kiloneters is a lot more power intensive than optical transmission.

Satellite internet is cool, but energy efficiency is not one of it's features.

Most cable laying ships are in 10,000 tonne size , ships this size use typically 50 tones of fuel a day at cruse speeds , cable ships travel much slower than that. The fuels ship typically use is one of the dirtiest as compared to RP-1 kerosene a falcon 9 uses. I am not so sure rockets are not cleaner approach.

On the downside The life of a cable is 20 years or more , as compared to only 5 years of a starlink sat . Definitely lot more data can go through a cable , it is not apples to apples comparison .

I ran an esp8266 running a website on Python, powered by a 9V battery.

Awesome! You have a url or blogpost about it? Now put it on solar :-)

This is what i like to see - using actually energy efficient systems for energy efficiency!

If you don't mind my asking, what was the total cost for this project? I run a lot of website projects off of Pi-0Ws and have been poking around with doing a solar project myself.

Controller is about 70 euro's. Buck converter (not documented but added) 7 euro. Panel was 110 Euro. Extra Sensor was 10 euro. Car battery was free. Maybe + 20 for cables and connectors.

Frankly, I'm not sure how economical it is, but it is fun to me.

I wonder if it's possible to take it a step further and make the pieces that connect the server to the internet (modem, the cable infrastructure, etc...) to run off the grid?

I wish I could, but I can't. During rainy days, I can't even support the Pi.

This is mostly because my solar panel is in a terrible spot and I had to over-size it just to get enough power.

Is this so great? There was some wind power controversy with the Gibbs documentary Planet of the Humans. As I've understood, there are some concerns also with solar power, especially due to the rare earths needed for the construction of such panels. I mean, sure, I love the thought of being self-sufficient and get off the grid, but why are they then also taxing solar cell panels to death?

No these are baseless claims it seems. Similar with the MIT study about carbon costs from making a BEV. A blog that was funded by some PR firm paid by oil companies took that story and ran with it. The points of that blog were refuted by the MIT team later on but no-one listened of course. It has caused significant damage as it managed to stick to the brains of a lot of people.

I wouldn't be surprised if that "documentary" was also paid for.

It seems that reputable Scientists refute the main points as stated by this 'documentary'.

IDK. A municipality of Kvaløya in Norway were warned against even touching their own tap water, after highly toxic machine oil used to lube windmills had leaked into their reservoire of drinking water. I know this, because I grew up there. So windmills aren't all fun and games, and IMHO their power efficiency is questionable compared to their many alternatives. Norway used to rely solely on hydropower, and were self-sufficient on this resource for many years, but in the latter years, the developement of windmills instead of hydropower seems to be more of a political move than a scientific, or even economic one. Or, well, it's certainly good for someone's economy... But for most people it's just far more expensive, and far less efficient. And, as I've already stated, it can even be dangerous. On top of this, the blades cut down all kinds of birds, and in particular eagles, which is kind of a big deal in Norway since many of these species are on the "Red List" (i.e. in danger of becoming extinct).

I'm currently working in my spare time one something similar but with a focus on a water tight box with a solar panel that I can place anywhere: https://jwillmer.de/blog/projects/independent-iot-system-par...

Please note that mobile users - due to my bad mobile layout - will find the solar stats at the bottom of the page (lot of scrolling).

The blog is still "wasting" or "leveraging" the same amount of power as it did before.

The money invested / and the solar cells could have been used to mitigate the carbon footprint for other power uses, but instead it's being used here for the blog. Shutting down the blog is how to get back ahead, if your goal is to reduce carbon footprint.

just pointing out that the carbon footprint of the blog and the carbon mitigation of the investment are independent events, they don't need to be tied together.

I think this is addressed in the blog, the current setup with the VPS is not gaining anything.

So, who is going to run a solar-powered k8s cluster based on a bunch of Raspberry Pi 4's and host something more significant?

A single pi4 has an impressive amount of processing power...

In my day job I'm a .net and react guy and I made a proof of concept for a next version of a possible product (tech preview, to show the people that their estimates were just insane - on the high side) on a pi 4, running in .net core on manjaro on my pi4.. this is a fairly intensive thing, using signalR with lots of messaging going on and cpu usage is ~10% at peak...

So.. my next task will be to run that on solar too :D

I hope you will do that. Please put it on solar :-) and blog.

We ran the web on pentiums, a Pi - especially the 4 - should be able to do this without problems.

I assume you're paying for a static IP from your ISP, or is there some other way you're routing traffic to the server?

Even without a static IP, it usually won't change that often. If it does change, then you can always update your DNS with the new IP automatically.

Quite often ISPs chuck you behind a NAT now which makes things harder.

I have never encountered this! I’m in the USA. Which ISPs?

Both is true.

My home connection has a static IP. But I'm using my old VPS with HAproxy (see my article) to either direct traffic to my solarPi or to my VPS if the Pi is down.

So in some way you could see that as 'cheating' but I discuss that in the blog.

Rather than vps backup, why not use your house mains supply as a backup?

Using your house's mains is a better idea, but if that's not possible for whatever reason, you might want to look at using a fail-over DNS record rather than a VPS running haproxy. It's not a standard feature of DNS, but some DNS providers including Amazon's Route 53 offer it. https://docs.aws.amazon.com/Route53/latest/DeveloperGuide/dn...

It will route to your solar setup while it's responding to it's healthchecks, otherwise to the failover.

Also, depending on how much traffic your blog gets, you might be able to do away with the failover and front it with a CDN. Cloudflare's free tier will probably be enough. If you get too little traffic, you wont maintain a hot cache and it just won't work.

That is a good point.

Originally, I was planning on doing so! I have all the ingredients (relais and charger) to do so.

Maybe I will implement it at some point.

How would you do it? Would it work to just connect the mains charger to the battery and have it kick in once the battery has reached 70% charge?

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact