A lot of you are new to Lemmy and the fediverse. You might have some questions about it, about me, or about discuss.online and it’s sister site utter.online.

Ask me anything. I’ll be watching this thread all day today: June 30th, 2023.

  • jgrim of SublinksOPMA
    link
    fedilink
    English
    arrow-up
    2
    ·
    edit-2
    1 year ago

    Right now, I’m restricted due to funds. It’s already pretty expensive to operate. My current goal is no data loss. Outages are okay until I can afford not to allow it at all. Lemmy is also still at version 0.18. Not even at full release. Much is yet to be done to make Lemmy fully stable and allow for scaling. Currently, a single backend instance can only use a single database endpoint. It will grow and I will grow with demands and funds.

    I’ve considered making a replacement for Lemmy that acts the same if it grows enough. I’d love to be able to participate in the conversations but without some of the limits of a somewhat monolithic system. That would allow an instance to run one East Coast and one West Coast independently and keep in sync.


    My system is like this:

    1. Load balancer
    2. Two web front-ends (can scale more with growth)
    3. 3 Databases (one is for Lemmy, one for Failover, one for reads for moderation)
    4. 1 Backend server (doesn’t scale horizontally)
    5. 1 Image server (doesn’t scale horizontally)
    6. Object storage with DigitalOcean for images.
    7. Bastion server for connecting to resources
    8. Grafana server for running some custom moderation dashboards
    9. Loki server to store logs from all the other servers and report to Grafana
    10. Another web server running the wiki
    11. Another database running the wiki

    For utter.online you add:

    1. One server for the services
    2. One database with failover
    3. One redis server
    4. ObjectStorage

    Finally, I’m also running a Matrix server. This is currently hosting the admin channels for Lemmy’s admin community:

    1. One server
    2. One database with failover
    3. ObjectStorage

    A lot is running and to keep an eye on. It’s also not cheap. So in short… I’m gonna wait a bit to grow the infrastructure more.