Tiekoetter.com Forum

A Looking Glass for AS208694 is now available.

You can check it out here: https://www.lg.as.tiekoetter.net

Invidious works again!

My Invidious instance is currently dead since Google/YouTube blocked all my proxy IPs and I can't get a valid token anymore. There is a possibility I can use my own AS for IPs soon but that is not 100% sure yet.

Hey thanks for the feedback!

No the search feature is not intentionally limited. I have encountered it myself a few times. As far as I have seen it depends on the time of day due to how many people are using it.
Why it is limited way more than the actual content and media I don't know.
I probably have to add a few more twitter accounts to my instance to avoid this.

This will take a bit of time because these accounts have to wait a but before they can be used for the instance.

What you are describing is the current problem with the bing and yahoo engine. They don't give quality results for an unknown reason for some percentage of results. The reason why bing is the default on my instance is because it gives near limitless results as it doesn't rate limit. Before I had google enabled as default but it broke in early october (refer to this GitHub issue if you want to read up on it: https://github.com/searxng/searxng/issues/5286)

Either way since duckduckgo and brave also increased their bot detection the only "quality" general search engine left is presearch.

You can switch the engines when you go to preferences -> engines. There you can enable/disable engines however you want.

Hope this helps and thanks for your feedback!

Sorry for the late reply.

Invidious is a front-end / proxy for YouTube. You can check the source code from the Invidious project at https://github.com/iv-org/invidious

If you have any specific questions regarding my instance of invidious feel free to ask.

Yeah the status page is just checking if the index responds with 200 or at all. I could check specific paths but I cache most of it.

The reason why it behaves like this is because my instances gets hammered with 5 times the usual traffic for almost the last 72 hours.
Looks like legitimate traffic.

I would have to add more twitter accounts as the current number always sits limited at 95-98% which gives you the "rate limited" error message. If you refresh a few times you may get through to a valid session.

I will have to monitor this for the next few days. Adding more twitter accounts is very tedious.

Thanks for voicing your concerns!

You can change this on under your profile -> settings. There you can change each different notification type from "notification only" to "notification & email". I have this set to notification only so it doesn't spam inboxes by default.

Hope this helps and thanks for your feedback!

That is a good question and I was looking at Discourse initially but couldn't get it to work how I would have wanted. It also throws countless errors and seems more hacked together than a thought through software. The docker image was not really working, this may also be an error on my part but they officially don't support docker as an install method which I thought was a red flag.

I came across NodeBB in my research for a modern forum software. Looked promising. They provided a clean docker install method. I also noticed that Netgate (the company behind pfSense) uses NodeBB for their forums.

It also looked easy to maintain, which turned out true, so I settled on it.

Yeah that was the intention; to keep everything self-hosted. Although cloudflare is great for some tasks one would have to take their word that they are not reading user data when they act as the TLS termination.

I have extended the IPv6 access via AS208694 to redlib.tiekoetter.com and nitter.tiekoetter.com

Just checked it. It was actually Twitter that was down and not the Nitter instances. It is still a bit flaky but so is Twitter itself right now.

It looks like Nitter got nuked.
https://status.d420.de/
My instance only has cached results available. I will have to take a look later today.

Hello,
please post your feedback / issues / comments regarding IPv6 access via AS208694 (www.as.tiekoetter.net)

I would appreciate your input to evaluate the functionality of AS208694 for future adjustments.

I have implemented IPv6 access via AS208694 for searx.tiekoetter.com, invidious.tiekoetter.com, wikiless.tiekoetter.com and this Forum.

AS208694 is operated by me and serves as another layer of control. Currently only IPv6 is possible as IPv4 is far too expensive for an individual to get. The IPv6 access is terminated at the same server, it is just routed differently. It works in conjunction with the current IPv6. I may expand this to the other services in the future.

More information about AS208694 can be found here: www.as.tiekoetter.net

For the missing features please refer to the GitHub repository of the developer of Nitter: https://github.com/zedeus/nitter

The "following users" suggestion is on the roadmap (account system) I don't know how far this feature is in development. You can open an issue there regarding the NSFW options.

Thanks for the feedback! I hope my hosted services continue to be useful to you and others!

The Nitter instance is now fully restored again.

After a bit more testing and one full day of evaluation Wikiless is now restored.

The aggressive caching didn't really help. I added Anubis on top of Wikiless and after a few minutes it completely removed the rate limit on the back-end.
Meaning JavaScript is now required for Wikiless but the barrage of bot requests gets filtered out. In my opinion a fair trade off.

Shoutout to Anubis. This software is incredible at stopping bots/scrapers https://github.com/TecharoHQ/anubis

I have implemented very long caching rules. This should mitigate the hammering of the wikipedia.org servers. From my usage when encountering an "INVALID_HTML" page (429 from upstream) refreshing a few times will get the page eventually. These cached pages are considered fresh for 6 hours and stale for up to 1 day, afterwards they are discarded if no other user requests the page in the meantime.

This is not optimal but a good enough intermediate solution.