sorry for the folks using #FakeRelay... last night I was playing SRE and left things in an inconsistent state (sending 50% of the traffic to an instance that wasn't working properly). It's now corrected and I'm thinking about adding more things to the health endpoint to catch this.

You will see less statuses than you should have in the last ~10 hours, but that's fixed now.

Fortunately, we're still well within our SLAs 😌

If you're running #FakeRelay and it hogs CPU, give 1.0.43 a try... they said recreating HttpClients on C# is wrong, they were right

(I kept creating it to send the Date header, but ChatGPT gave me a nice workaround)

PSA for people using #FakeRelay: I've just changed servers. My rudimentary testing says it's still working, but if you see anything please ping me :)

It's time for my #introduction post on the new server :)

I'm Gervasio, I'm a runner, an aspiring magician, a father and a software engineer.

I live in #Uruguay and I'm working at a small startup (you can see a beautiful "coming soon" page at sumble.com) and before this I worked at #StackOverflow twice.

I enjoy #csharp, but right now I'm working on #python. I'm after interesting problems more than interesting technologies :)

I came to mastodon multiple times, but this became my home in the end of 2022. I tried to build a community (discovered I don't have the skills to do so) and then settled on a single user instance.

One thing I noticed when I was spinning up my server was that I couldn't discover relevant content, so I built #FakeRelay (an API to import content into a Mastodon instance) and then #GetMoarFediverse (as an example to use FakeRelay).

If you're curious about those, I recommend you first check out #fedibuzz, as it's simpler to run. If you still want to use FakeRelay, hit me up and I can get you started.

@g3rv4 Hi threre, checking out the #FakeRelay + #GetMoarFediverse solution, and wondering if you’re still offering single user instance admins like myself access to your fakerelay.gervas.io service? My instance domain is cdrum.social. Thanks! Really appreciate it! ;-)

@yeri I found this blog post that I’m exploring. Two options - #fedibuzz and #FakeRelay / #GetMoarFediverse - along with #fedifetcher.. Will report back!

I just found out about #FakeRelay and #GetMoarFediverse

https://github.com/g3rv4/FakeRelay
https://github.com/g3rv4/GetMoarFediverse

by @g3rv4 and these could have saved quite some time if I had known about them earlier...

@jan On mountains.social I use #FakeRelay (https://github.com/g3rv4/FakeRelay) and #GetMoarFediverse (https://github.com/g3rv4/GetMoarFediverse) from @g3rv4. You can specify which hashtags and from which instances you want to inject into your federated timeline. That really works out. My current config can be see here: https://github.com/spiezmaestro/mountains.social_GMFAction/blob/main/config.json

Most up-to-date configuration for the #FakeRelay. Please let me know when you want more hashtags gathered on mountains.social.

When I can't figure out what is wrong with my fakefirehose setup

#fakefirehose #fakerelay

Pain... Time to dig more... Later. Will get everything back to a functional state hopefully soon

Have a couple ideas-

but my blog will be entirely restarted... Don't think If I want to continue using Hugo... Any recommendations?

Finally got around to configuring #FakeRelay and #GetMoarFediverse on this instance, the way I had it on my previous instance. Now I automatically get content with hashtags that interest me in my feed again. It is such a lifesaver for small #selfhosted servers. #mastoadmin

Thanks to @g3rv4 and his excellent work on #FakeRelay (https://github.com/g3rv4/FakeRelay) and #GetMoarFediverse (https://github.com/g3rv4/GetMoarFediverse) there is now a way to more specifically pull relevant posts into the federated timeline. The instance admin can define which hashtags from which instances are pulled, thus providing content more relevant to the people on the instance.

3/4

#mastoadmin

Been writing up this post mostly as a note to self on how I set up my #mastodon #instance, and what I learned.

• Server setup and system requirements
• Switch to glitch-soc and customisations
• Relay / #FakeRelay / #Fedibuzz
• Translation with #libretranslate
• Puma optimisation
• Sidekiq optimisation

Let me know your thoughts

https://blog.thms.uk/2023/01/setting-up-mastodon

#mastoadmin #glitchsoc

@ryanhe 还有人搞了个 #fakerelay ，能配置拉某些特定实例的特定tag，不过用起来还是有点不太方便

I'm connected to a bunch of #Mastodon relays now, though I suspect I'll be able to remove a some in time to quell the firehose, about 90% of message traffic comes from relays.

Visibility to incoming ActivityPub traffic was very opaque, so I built a custom InboxProxy app that would receive the message, queue it, parses/pull stats, and forward it. This provides a pile of stats, allows me to instantly configure drop rules to drop some or all traffic from various sources, or from/about certain instances. It also buffers messages, so anytime that I am restarting, tweaking, or otherwise breaking Mastodon, it will queue them and once its back online, will deliver them. Apache cuts POST traffic to /inbox to InboxProxy.

I also run #Fakerelay, though I haven't liked the implementations of GetMoarFediverse or similar that I've seen, because they are too aggressive and don't really act neighbourly.

So I wrote a custom "TagFetcher" app that will pull a list of followed tags from the Mastodon database, peek at what tags are popular across a couple big instances. Then dynamically fetch messages on those tags from various large instances that I can't get via relay, starting with, and frequently pinging my own instance, so it has a fresh list of known messages, and adjusting fetch frequency of other instances up or down (within limits) depending on how many unseen messages are found, and sending new messages to FakeRelay. Delivers about 100 messages/hr avg.

🧵 3/3 🔚

Next up in my series on how and what I #SelfHost: #Social stuff. All of this runs in #Docker #Swarm on a single node.

#Matrix: Running Synapse (tried Dendrite first, but run into problems). I'm using samuelph/synapse-compress-state, but it's still heavier on disk usage than I'd like, given the number of rooms and time.

#Mastodon: I use a custom docker image, based on the official tootsuite docker image, except I bump MAX_ITEMS in feed history to 10,000 (my client no longer gets bumped to latest when I fall behind a bit), REBLOG_FALLOFF to 400 (fewer repeated boosts), upped some authenticated ratelimits (private server), and bumped max toot size, and added a cleanup script.

🧵 1/3

🏷️ #KeyDB #PGBouncer #ActivityPub #FakeRelay #SelfHosting

@golfhaus You could try #FakeRelay for #discovery (https://github.com/g3rv4/FakeRelay) but it's a problem I'm also still trying to solve.

@cassidyjames you can also add relays or even a #FakeRelay to get more content in your federated timeline and search results

I'm happy to see that @astro built #FediBuzz, it's even simpler to use than #FakeRelay and #GetMoarFediverse

You add the relays that follow the #hashtags you want, and that's it.

The only thing I haven't seen is it's source code, but it looks pretty pretty pretty good.

@coffe from my experience, my usage is around 20gb. Using #fakerelay and #getmoarfediverse to get content in. Seems to do a clean out every so often, hasn't gone above about 25gb...