I run a Mastodon (mammut.gogreenit.net) and PeerTube (pt.gogreenit.net) instances for myself and friends.

I am interested in IT, Electronic Music, Winter Sports, Renewable Energy, Off-Grid living, Sustainability, The Right to Repair, Veganism and Animal Rights

  • 2 Posts
  • 16 Comments
Joined 1 year ago
cake
Cake day: October 15th, 2024

help-circle
  • He may have a point, but as someone else pointed out, a lot of these self-hosted services are running on out of date equipment that wouldn’t be used for anything else.

    I run all my Fediverse stuff on an old Dell R620 that a friend gave me. Mine is totally specialised for hosting… Yes, it is overkill for my ~10 users, but hopefully more friends will join. Also, it’s not the most efficient way of doing it - the device is probably 10 years old and uses ~130w 24/7. A newer NUC or equivalent would probably only use 40-50w. However, who else is going to use this machine? No company would touch it, “everything” is going cloudy, so it would either be stripped for parts and the rest dumped either in landfill or sent to some 3rd world country.

    You can claim it’s a waste of electricity to use it, but a lot of energy and materials were used to create the server in the first place, and most of that will be lost, even with recycling.

    People run Fediverse (and other services) on a Raspberry PI - fine for a couple of users, but too restrictive for my use. These things only use 5-20w, which is amazing.

    My electricity supply is from a “green” supplier, and I have a local SolarPV system that powers the system when there is enough sun. Last summer I managed to run it for over a month using my local system only. There’s no reason that we can’t build more renewable sources of electricity. Here in the UK there’s a proposed 140 DCs in in the planning phases, which is ridiculous, all for AI BS.

    Self-hosting isn’t for everyone of course - not every household should do this, but there’s no reason why groups of friends, families or “activity” groups couldn’t do this effectively.

    It’s absolutely optimized for the long term - how many Google services have been discontinued when there are still users, just not enough to be profitable? Self-hosted services can run as long people are interested. A mail server created 20 years ago is still compatible and useable today because it uses ratified, slow-moving standards.





  • Chewie@slrpnk.nettoSelf-hosting@slrpnk.netWhere to begin?
    link
    fedilink
    English
    arrow-up
    2
    ·
    7 months ago

    Also i know some basics on raid but I’ve only ever messed with raid0 with usb drives on a pi. I have 8 bays but 2 are currently vacant. What is the process of just adding an extra drive to a raid, or replacing one that already exists?

    It depends on your RAID controller (or software RAID). I use hardware RAID (on Dell and HP servers) as it’s easy and a known technology, although these days people seem to be anti-HW RAID a bit.

    When replacing a drive, you just eject the old drive, wait a few seconds put the new drive in, and most HW RAID controllers will start automatically rebuilding the array. Make sure your controller and drive bays support “hot swap” first! With HW RAID, replacing drives is great, because you can increase the capacity over time, because you can replace each drive with a bigger model, and once the last drive has been swapped over, you can expand the array and start using the extra capacity without having to move data around. With HW raid, most servers have an “Out-Of-Band” system (iLO, iDRAC, IPMI) which you can configure to alert you if a drive has died (or is about to die).

    I would recommend keeping at least 1 spare of the same model HD of whatever you use, just in case.

    I got burned by having a WD drive fail, and WD were being assholes about sending me a replacement (it was under warranty). Before I got the replacement, another drive started dying, and I couldn’t afford to buy another drive. In the end I lost 12TB of data 😭

    And re the above - “RAID is not a backup” :) plan accordingly…

    For software RAID, most Linux OSes support it automatically. I only use it as it’s easy to expand partitions (most of my Linux machines are VMs on a system with HW RAID).

    This might be a useful article https://www.howtogeek.com/40702/how-to-manage-and-use-lvm-logical-volume-management-in-ubuntu/ (with a link to a previous one which is an introduction), which explains a bit about SW RAID.


  • Chewie@slrpnk.nettoSelf-hosting@slrpnk.netWhere to begin?
    link
    fedilink
    English
    arrow-up
    1
    ·
    7 months ago

    I rate OPNsense. I’ve not tried pfsense, but I use Enterprise-level firewalls daily. When you’re used to Palo Alto, Cisco or CheckPoint firewalls, it is a lot harder to use, and the interface isn’t great, and had fewer features, but for free (and cheap support if you need it), it’s pretty amazing. Upgrading to new versions is seamless, and once when something happened and it broken, I reinstalled it from the .ISO, uploaded my backed up .xml config file, and it was back to normal. It’s more than adequate to use for my home internet connection and all the services I run in my DMZ etc.





  • Absolutely. In a previous company, we migrated from on-site MS Exchange to Google Mail (ugh). Apart from it being a crap experience (it was a new service), and feeling like we were beta testers as things kept changing daily, so writing training material was a PITA, once there was an outage, and even though we had ~10K users on it, they basically said “get in line” when we were chasing for updates etc even though we were a paying customer!

    Fuck them.


  • No licence is required for this model. I wouldn’t give away something that needed a shitty licence without mentioning it (but then I probably wouldn’t have taken it for myself in the first place).

    I made that mistake with an Infobox device at an auction a few years ago. You can use all functionality for 30 days, then after that you need a licence or NOTHING works and you have to wipe it all and start again. What a pile of shit, I only wanted to play with it and run a few DNS zones.

    It’s a shame, I was hoping to learn something from it as they cost a fortune, and usually overkill for what I need, but instead i’m never going to recommend their products. I could open it up and see if I can run Linux on it, but I don’t need yet another device that runs bog-standard Linux - any boring hardware could be used for that purpose. 😞





  • APC do a really crappy small one for telecoms cabinets, but none for servers
    

    I wonder if the lower discharge current capability of LFP batteries is why? That’s the one thing I’ve read fairly consistently about them is that they can’t supply the same high current as lead acids but are otherwise superior in every way. Now that you mention it, the only place I’ve ever really seen LFP UPSs for servers is in the big, central UPSs where they can run batteries in series for a much higher voltage.

    I don’t think so. Cheaper batteries have that problem, but a decent brand does not. Check out this one: https://www.powertechsystems.eu/home/products/48v-lithium-ion-battery-pack/48v-105ah-5-38kwh-lithium-ion-battery-pack-powerbrick/ I bought one for my house, and have a 5KW inverter connected to it. Its specifications say that can do 120A drain continuously. I have used it to boil my 3KW kettle a few times in one day (but not often - I usually use the power for other things), and it has been fine.

    e.g. most of the LFP UPSs I see max out at 1000 VA where 1500 is more typical for lead-acid UPSs.

    That’s just a limitation of the product, not the technology.