• NeshuraMA
    link
    fedilink
    English
    arrow-up
    2
    ·
    1 year ago

    Ok so for starters I added a whitelist for lemmy.zip because the entire .zip tld is blocked in my network. Lemmy is now updated, I’ll check back alter so see if any error messages pop up.

    • DeemoOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 year ago

      Kinda a follow up your theory about a rate limit seems to be spot on was checking my subscriptions and noticed the upvote counts were stuck at 0 or -1 while cross comparing to the larger remote instance they are (50-100).

      After the update however vote counts seem to be working for newer posts recieved after the update

      Probably fix is due to new changes in 0.18.2 which dynamically increase and decrease workers on demand

      https://github.com/LemmyNet/lemmy/issues/3121

      https://github.com/LemmyNet/activitypub-federation-rust/pull/42

    • DeemoOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      Quick question regarding the rate limit.

      Is the rate limit fine tuned to initial history sync from new federated communities (ie if you join a new community only 10-20 days of posts history will sync after which point only new posts will received and no further history will be synced).

      Or is the rate limit more like the instance overall can perform a set number of federation actions in a given period, and the volume of new federated content from larger instances prevents new history from being synced.

      I noticed when subscribing to new communities new posts, comments, upvotes sync pretty much instantly history seams to be the part the gets bogged down.

      • NeshuraMA
        link
        fedilink
        English
        arrow-up
        2
        ·
        1 year ago

        The second option is the more accurate description of what’s going on, 0.18.3 should alleviate that somewhat as inactive instances will start being phased out now. Previously even dead instances took up time during the update cycle having all activity pushed to them even if they were dead.