Another day, another update.

More troubleshooting was done today. What did we do:

  • Yesterday evening @phiresky@[email protected] did some SQL troubleshooting with some of the lemmy.world admins. After that, phiresky submitted some PRs to github.
  • @[email protected] created a docker image containing 3PR’s: Disable retry queue, Get follower Inbox Fix, Admin Index Fix
  • We started using this image, and saw a big drop in CPU usage and disk load.
  • We saw thousands of errors per minute in the nginx log for old clients trying to access the websockets (which were removed in 0.18), so we added a return 404 in nginx conf for /api/v3/ws.
  • We updated lemmy-ui from RC7 to RC10 which fixed a lot, among which the issue with replying to DMs
  • We found that the many 502-errors were caused by an issue in Lemmy/markdown-it.actix or whatever, causing nginx to temporarily mark an upstream to be dead. As a workaround we can either 1.) Only use 1 container or 2.) set proxy_next_upstream timeout; max_fails=5 in nginx.

Currently we’re running with 1 lemmy container, so the 502-errors are completely gone so far, and because of the fixes in the Lemmy code everything seems to be running smooth. If needed we could spin up a second lemmy container using the proxy_next_upstream timeout; max_fails=5 workaround but for now it seems to hold with 1.

Thanks to @[email protected] , @[email protected] , @[email protected], @[email protected] , @[email protected] , @[email protected] for their help!

And not to forget, thanks to @[email protected] and @[email protected] for their continuing hard work on Lemmy!

And thank you all for your patience, we’ll keep working on it!

Oh, and as bonus, an image (thanks Phiresky!) of the change in bandwidth after implementing the new Lemmy docker image with the PRs.

Edit So as soon as the US folks wake up (hi!) we seem to need the second Lemmy container for performance. So that’s now started, and I noticed the proxy_next_upstream timeout setting didn’t work (or I didn’t set it properly) so I used max_fails=5 for each upstream, that does actually work.

  • phiresky@lemmy.world
    711·
    1 year ago

    server load is too low, everyone upvote more stuff so i can optimize more

    edit: guess there is some more work to be done 😁

  • xavier666@lemm.ee
    33·
    1 year ago

    Even though i’m not from this instance, this is such a nice way of keeping the users posted about changes. I wish more companies (I know this is not a company) went straight to the point, instead of using vague terms like “improved stability, fixed few issues with an update” when things are changed. I hope all instance owners follow this trend.

    • Ruud@lemmy.worldOPM
      11·
      1 year ago

      The owner of your instance has been a big help. You’ve also chosen a good instance!

      • xavier666@lemm.ee
        9·
        1 year ago

        @[email protected] is awesome. He keep us aware of what’s happening, planned maintenance hours, etc. His commits on making lemmy scale horizontally is what kept lemm.ee snappy even when we had a huge influx of users. I hope Lemmy as a whole continues this ethos of collaboration.

    • ramblechat@lemmy.world
      7·
      1 year ago

      As an IT support person, I’ve learnt to dumb things down for management. They don’t want to hear stuff like “Increased the SGA, changed the buffer size and added a function based index…etc”. Sometimes I’ll do a short and a long version something like “the issue was around memory settings which have been increased”, plus the detailed info.

  • kionay@lemmy.world
    21·
    1 year ago

    You guys had better quit it with all this amazing transparency or it’s going to completely ruin every other service for me. Seriously though amazing work and amazing communication.

  • CIA_chatbot@lemmy.world
    171·
    1 year ago

    It blows my mind with the amount of traffic you guys must be getting that you are only running one container and not running in a k8s cluster with multiple pods (or similar container orchestration system)

    Edit: misread that a second was coming up, but still crazy that this doesn’t take some multi node cluster with multiple pods. Fucking awesome

    • guyman@lemmy.world
      31·
      1 year ago

      A lot of containerization is unnecessary and perpetuated by people who don’t know what they’re doing.

      • CIA_chatbot@lemmy.world
        2·
        1 year ago

        Kinda shitty dig there but whatever, I’m a certified k8s admin, I would say I’m pretty damn good with containers.

        With the amount of traffic I’m assuming lemmy.world gets I was just imagining it taking advantage of something more robust than just running Docker. It was more a comment on being impressed that lemmy must be damn efficient.

    • 𝕽𝖔𝖔𝖙𝖎𝖊𝖘𝖙@lemmy.worldEnglish
      7·
      1 year ago

      I couldn’t get it to work.

      I was able to enable it and add the code to my authenticator but it would not accept a login with the 2FA code.

      Fortunately I was still logged in elsewhere to disable 2FA again or I may have gotten locked out

    • pitninja@lemmy.worldEnglish
      4·
      1 year ago

      It’s always been safe to use 2FA if your authenticator app supports SHA256. Unfortunately, it turns out that a lot don’t. The only solutions are going to be Lemmy switching to SHA1 or users switching to auth apps that support SHA256. I think the first is more likely to happen than the second.

        • pitninja@lemmy.worldEnglish
          1·
          1 year ago

          I believe that is correct, from what I’m reading. I think Lemmy is probably going to switch to SHA1 as the default. Research has shown that it’s basically as safe to use for 2FA as SHA256 and SHA512 and obviously it has universal compatibility per the spec. The spec only lists SHA256 & 512 as allowed alternatives, not required for full adherence to the spec. I imagine Lemmy will change it so that SHA1 is the default option with maybe an option to still do SHA256 with some well explained warnings.

          • pathief@lemmy.worldEnglish
            1·
            1 year ago

            From what I’ve researched you’re 100% correct.

            The main problem is that even though Authy doesn’t support SHA256, Lemmy still enables 2FA. Lemmy doesn’t ask for an auth token before enabling 2FA nor generates any backup codes. The user is prompted with success but will be locked out on of the account on sign out. Not good.

  • Spectator@lemmy.worldEnglish
    14·
    1 year ago

    I’m not sure wtf you just said, but lemmy.world feels very smooth today, so thank you for your continued hard work!

  • Kodiack@lemmy.world
    13·
    1 year ago

    Awesome work - things seem to be running much more smoothly today.

    Do you have anything behind CDN by chance? Looking at the lemmy.world IPs, the server appears to be hosted in Europe and web traffic goes directly there? IPv4 apparently seems to be resolving to a Finland-based address, and IPv6 apparently seems to be resolving to a Germany-based address.

    If you put the site behind a CDN, it should significantly reduce your bandwidth requirements and greatly drop the number of requests that need to hit the origin server. CDNs would also make content load faster for people in other parts of the world. I’m in New Zealand, for example, and I’m seeing 300-350 ms latency to lemmy.world currently. If static content such as images could be served via CDN, that would make for a much snappier browsing experience.

  • KSP Atlas@sopuli.xyz
    12·
    1 year ago

    Shouldn’t the correct HTTP status code for a removed API be 410? 404 indicates the domain wasn’t found or doesn’t exist, 410 indicates a resource being removed

  • MiddleWeigh@lemmy.world
    11·
    1 year ago

    I took a SM break for a few days, and it’s running noticeably better today…I think. (:

    Thanks a bunch for floating us degenerates.