Which Lemmy instances don’t rely on cloudflare?
from hddsx@lemmy.ca to nostupidquestions@lemmy.ca on 18 Nov 15:09
https://lemmy.ca/post/55386786

…or GCP? …or AWS?

#nostupidquestions

threaded - newest

tobogganablaze@lemmus.org on 18 Nov 15:12 next collapse

Had no problems here on lemmus.org.

hddsx@lemmy.ca on 18 Nov 15:23 collapse

403 error

tux0r@feddit.org on 18 Nov 15:12 next collapse

The one you host yourself.

hddsx@lemmy.ca on 18 Nov 15:19 collapse

I considered it, but then there was the big child porn incident. I don’t want to be liable for hosting child porn if someone posts it somewhere and it federates over. If there was a way to do text only I would consider it again.

I forgot what the solution was, but it’s not a risk I want to take - and I do host many services myself.

yardratianSoma@lemmy.ca on 18 Nov 16:04 next collapse

I mean, one solution would be to not allow photos on that instance. Just spitballing, because I can’t think of an efficient and ethical way to go about screening image uploads, that doesn’t involve some sacrificial moderator being waterboarded by human perversion.

Maybe, using AI? Can a locally hosted AI instance scan images? I might host a lemmy instance in the future, because I want to support the cause, but yeah, I ain’t going to jail for it.

bjoern_tantau@swg-empire.de on 18 Nov 16:30 collapse

One of the reasons I’m not proxying images.

The other being that it takes too much storage.

hddsx@lemmy.ca on 18 Nov 16:32 collapse

What does that mean? If you’re not proxying images, do you have access to images but you have to fetch them from the host instance or do you not have images at all?

bjoern_tantau@swg-empire.de on 18 Nov 16:34 collapse

They are fetched from the host instance.

c0dezer0@programming.dev on 18 Nov 15:12 next collapse

Noone is answering because they are still struggling 😅

Shadow@lemmy.ca on 18 Nov 15:15 next collapse

I don’t think any run on gcp or aws, but most use cloudflare

straycatstrut@discuss.tchncs.de on 18 Nov 15:24 next collapse

While the downtime was most active, most of the top instances were down. Lemmy.ml, feddit.org, discuss.tchncs.de and behaw.org were all up. You can use lemmyverse.net to browse things and the ones offline all show a “content error” in lemmyverse.

hddsx@lemmy.ca on 18 Nov 15:29 next collapse

So lemmyverse does not run on cloudflare? Then that’s a good tool next time. Thank you

straycatstrut@discuss.tchncs.de on 18 Nov 15:47 next collapse

I cannot confirm that (I have nothing to do with lemmyverse), but it was/is up and functioning so my instinct is “yes, it does not use cloudflare”.

I grabbed a quick screenshot hours ago showing the trauma; based on my recollection that I typed into the registration box of tchncs, these of the top 20 instances were all down: lemmy.world, sh.itjust.works, lemmy.dbzero.com, lemmy.zip, lemmy.ca, programming.dev, lemmy.blahaj.zone, infosec.pub, aussie.zone, readthat.com, lemmy.today. Sister sites on the piefed side (e.g. piefed.social) were also down because they’re the same admins using the same tech stacks.

A lot of lemmy instances put all their eggs in one basket and found out.

<img alt="" src="https://discuss.tchncs.de/pictrs/image/b020808a-a2b8-4fb9-bb0d-a49d44963e23.png">

tux0r@feddit.org on 18 Nov 15:48 collapse

Lemmyverse is Fediverse, more or less. (I posted on Lemmy with Honk before.)

FosterMolasses@leminal.space on 18 Nov 16:12 collapse

The fact that all the “normie” instances were down, but lemmy.ml was up was the funniest shit to me. I’m honestly surprised they didn’t milk it more for all the shit they usually get lol

<img alt="" src="https://leminal.space/pictrs/image/a93a82e2-f6e8-483e-8cdb-00c6d87885e0.gif">

tal@lemmy.today on 18 Nov 16:27 next collapse

To repeat my comment here:

lemmy.today/post/41970730/20432766

I mean, it’s easy to check whether a given instance is using CloudFlare.

$ host lemmy.world|head -n1
lemmy.world has address 104.26.9.209
$ whois 104.26.9.209|grep ^NetName
NetName:        CLOUDFLARENET
$

You can browse anonymously on any instance that permits doing so, so if you just want to browse during an outage, you can do that anywhere.

IMHO, having an account on a second Threadiverse instance isn’t necessarily a terrible idea, not just because of CloudFlare outages, but because instances do have outages for various reasons. I have an account on olio.cafe (PieFed, not on CloudFlare) and on lemmy.today (Lemmy, not on CloudFlare) because I wanted to try out PieFed, and I have fallen back to that to post before if lemmy.today has issues.

That being said, I didn’t intentionally try to avoid CloudFlare. I mean, they’re used by a lot of major sites, and I don’t expect them to have a lot of downtime. I mean, every Threadiverse instance has had downtime for some reason or another. I’ve had Internet outages, as well as electricity outages. Not all that common or usually an extended thing, but they happen.

c0dezer0@programming.dev on 18 Nov 17:50 collapse

It seems you know your stuff.

Is there also a neat trick to find out if an instance rely on aws or gcp?

tal@lemmy.today on 18 Nov 19:13 collapse

CloudFlare is going to have someone talking directly to a CloudFlare IP address, so it’s going to be visible.

AWS or GCP provide servers which might be behind something like CloudFlare. If they were deployed like that, I don’t believe that there’d be a straightforward way to determine that that’s where the server is hosted.

If it’s directly-accessible, and not using a CDN like CloudFlare, then it’d work the same way as if you were checking whether they’re using CloudFlare, just do a whois query on its IP address. I don’t know a real instance offhand directly-accessible on AWS, but to grab a random AWS hostname and Google Cloud Platform hostname:

$ host ec2-23-20-1-1.compute-1.amazonaws.com.
ec2-23-20-1-1.compute-1.amazonaws.com has address 23.20.1.1
$ whois 23.20.0.0|grep ^NetName
NetName:        AMAZON-EC2-USEAST-10
NetName:        AMAZON-IAD
$ host 3.192.170.108.bc.googleusercontent.com
3.192.170.108.bc.googleusercontent.com has address 108.170.192.3
$ whois 108.170.192.3|grep ^NetName
NetName:        GOOGLE
$

For a real host, we can just ad-hoc scrape lemmy.world’s instance list:

$ curl -s https://lemmy.world/instances |tr '}' '\n'|grep -o 'domain":".[^"]*'|sed 's/domain":"//' >threadiverse-hosts.txt
$ xargs <threadiverse-hosts.txt -n1 host -- >threadiverse-hosts-resolved.txt
$ grep "has address" threadiverse-hosts-resolved.txt |cut -d" " -f4|xargs -n1 host -- >threadiverse-hosts-reverse-resolved.txt
$ grep amazonaws.com threadiverse-hosts-reverse-resolved.txt|head -n1
75.184.193.54.in-addr.arpa domain name pointer ec2-54-193-184-75.us-west-1.compute.amazonaws.com.
$ grep 54.193.184.75 threadiverse-hosts-resolved.txt|head -n1
c63b-77-100-144-83.ngrok-free.app has address 54.193.184.75
$

So there’s the hostname of a real instance using AWS directly, c63b-77-100-144-83.ngrok-free.app.

$ host c63b-77-100-144-83.ngrok-free.app|head -n1
c63b-77-100-144-83.ngrok-free.app has address 184.72.44.51
$ whois 184.72.44.51|grep ^NetName
NetName:        AMAZON-EC2-7
NetName:        AMAZON-SFO
$
HootinNHollerin@lemmy.dbzer0.com on 18 Nov 16:42 collapse

db0 (lemmy.dbzer0.com)