BTW (back to the topic of this thread):
Mastodon 4.1.1 has just been released. 😬

@girish said in How to install Elasticsearch on Mastodon?:

planned to add an elasticsearch addon

As an add-on? 🤔 Are you hinting at something new? Or did I miss something about Cloudron add-ons? 😉

@nebulon i've identified the issue and it was my S3 configuration.

The endpoint and hostname should never have the bucket name.

S3_ENDPOINT=https://s3.us-west-001.example.com
S3_HOSTNAME=s3.us-west-001.example.com

@girish great that works! remember to rebuild the feeds and redeploy elasticsearch if that's in use.

@matwinck you have to make someone an admin - https://docs.cloudron.io/apps/mastodon/#admin

@alwynispat sure of course:

export WEB_CONCURRENCY=4 # number of worker processes
export MAX_THREADS=5 # the number of threads per process

export STREAMING_CLUSTER_NUM=2 # number of worker processes

export SIDEKIQ_THREADS=4
export DB_POOL=25 # must be at least the same as the number of threads

@RazielKanos said in deleting old crap:

Heyas, I just ran a cleanup and was amazed that of my nearly 10GB it just deleted 60 MB(!) of cached files and 800 KB of preview cards. How can I get rid of all the crap lets say... order than 2 months? can't be all 5GB in the database, right?

tootctl media remove and tootctl statuses remove are run in the background with Mastodon v4 (settings can be found on the administration page).

What I am running every night (via cron job) is:
/app/code/bin/tootctl media remove-orphans
/app/code/bin/tootctl preview-cards remove --days=7
/app/code/bin/tootctl cache clear
/app/code/bin/tootctl accounts prune
and now with v4.1:
/app/code/bin/tootctl media remove --prune-profiles --days 7

My instance has around 40-50 users and is connected to 8 public relays.
It is about 50 GB in size now (it was 80-90 GB with v4.0).
Hope that helps.

I have it up and working with the new Cloudron LibreTranslate app. Running on the same Linode as my Mastodon.

https://forum.cloudron.io/topic/8527/libretranslate-with-cloudron-mastodon-server/1

@girish I've disabled backups for now as I try and work out why there is an issue with it. My Linode backs up daily at the moment so worst case I recover the whole system.

@dxciBel said in New external users sometimes don't receive confirmation mails:

Not that I know of. User management was left to the app during install.

Phew. I think you should be OK then! Ignore me.

https://www.reddit.com/r/Mastodon/comments/yzp1qz/mastodon_server_hardware_requirements/

This is a discussion on Reddit about what it takes to host a Mastodon instance (Mastodon uses the ActivityPub protocol). The requirements are quite high. One large instance with 29,000 users is hosted on this hardware:

2 AMD 5950 16c/32thread 128GB ram with nvme ssds running sidekiq jobs
1 AMD 5950 16c/32thread 128GB ram with nvme ssds running elasticsearch and redis
1 AMD 5950 16c/32thread 128GB ram with nvme ssds running Postgres DB
2 AMD 5950 16c/32thread 128GB ram with nvme ssds running PUMA (mastodon web app) and nginx (only one is currently powered on - other is being reworked
1 AMD Epyc 32c/64th 400GB ram with 10TB of nvme ssd acting as a s3 storage provider and another nginx/PUMA web front end

This is a total of 96 CPU cores! Another person with 500 users says,

"For those interested in a data point, I am running an instance with nearly 500 users. Currently running on an AWS Lightsail instance with 4GB Ram, 2 vCPU, for $20/month, and an S3 instance for media storage (1TB - nearly full) for $3/month. I am constantly monitoring resources - as the system is running well but nearly topped out.

I will likely be moving to 8GB RAM, 4 vCPU instance, and 2TB S3 in the coming weeks."

These requirements are quite high. I also recently learned that bandwidth is a continuous 1-2 Gbits/s. No wonder Mastodon is having trouble keeping up with the influx of Twitter users!

@alwynispat you've got DB_POOL specified twice 🙂
I'm doing:
be945764-0726-41d0-8886-d387d17c123a-image.png

@jdaviescoates indeed, and idiot guide for an already running cloudron/mastodon app would be nice.

Got deepl running with simpel guide, so hope that for adding s3 bucket this can be the same

@b2breporter Oh - not totally sure, but I think we figured it out. We were originally trying to post images that were much larger in file size.

@alwynispat do you mean nginx? How is NPM involved here?

edit: oh, maybe you mean your nginx proxy manager?

Unfortunately mastodon is not really designed much to change the domain and the user handles there. Some info on that is at https://docs.cloudron.io/apps/mastodon/#federation

okay, update looks easy. But reading this now
https://github.com/cybrespace/cybrespace-meta/blob/master/s3.md

Thanks for the hint, I have update the docs to indicate that registration is only disabled by default when Cloudron authentication is used.

@marcusquinn said in "Searching toots by their content is not enabled on this Mastodon server.":

Wordpress search using the ElasticPress plugin.

I guess to make searching on the woocommerce supplements shops you run work well/ fast?

@johannesjom generally the permissions are basically everyone can do everything on hetzner cifs storage, in this particular case the changing of permissions from the app startup script fails. So apps which do not require this, will likely work just fine.

@redegelde an idea is to do a "dry run" first. Just clone the app and play around with the commands to see how well it works.

@nebulon It was still a problem until the most recent update. Server reboot was super quick! Thanks!

https://github.com/mastodon/mastodon/discussions/21916 is the upstream discussion for future reference.

@ScottyMojo right, if you read the linked docs it says you need to start the app manually keeping recovery mode.

@doodlemania2 Seaweed FS may be what you're looking for to cache/gateway object storage.

https://forum.cloudron.io/post/56024

@shanelord01 you can ignore them. It's just logs from the box code trying to handle well known first before it hits mastodon.

@nebulon Yep - seems to be due to insufficient processes and threads in config.sh

I'm playing with:

at the moment and it seems to be catching up. 🤞

@pauljacobson ah yes, the fix for haraka is coming. I recently submitted a patch upstream for this which has now been merged - https://github.com/haraka/Haraka/pull/3104 .