https://discord.cloudflare.com logo
Join Discord
Powered by
# r2
  • v

    Vitali

    04/26/2022, 10:54 PM
    I punted on the team putting in effort into listUploads because the upload ID & part Etag are encrypted values that maintain state. However, synthesizing the upload ID/part ETag generates a random number every time so you'll get unique (validly accepted) values every single time you do the listing (I figured better for us to stabilize that piece before implementing those features)
  • j

    john.spurlock

    04/26/2022, 10:55 PM
    i'm going to try and stitch together a larger object since
    PutObject
    craps out at around 500mb (see my exact numbers from yesterday)
  • v

    Vitali

    04/26/2022, 10:55 PM
    Yeah I got that 🙂
  • v

    Vitali

    04/26/2022, 10:56 PM
    I think the production zone is probably misconfigured in the networking layer somewhere
  • v

    Vitali

    04/26/2022, 10:56 PM
    That doesn't explain the multipart problem though
  • j

    john.spurlock

    04/26/2022, 10:57 PM
    send back stack traces for all 500 internal errors!
  • v

    Vitali

    04/26/2022, 10:57 PM
    That's a non-starter
  • v

    Vitali

    04/26/2022, 10:57 PM
    The stack traces are minified so they're useless
  • v

    Vitali

    04/26/2022, 10:58 PM
    I don't think we have any tooling to unminify them outside of sentry
  • j

    john.spurlock

    04/26/2022, 10:59 PM
    i see, well hopefully you have logging or something when those happen - i'll try to get a reproducible multipart 500 scenario, but they seem to be flaky - work some times but not others
  • v

    Vitali

    04/26/2022, 10:59 PM
    we should but I'm not seeing it
  • v

    Vitali

    04/26/2022, 10:59 PM
    At a minimum it should be in our logging system but I don't know how to query that. Let me double-check with someone who does
  • v

    Vitali

    04/26/2022, 11:11 PM
    Oh - what are the sizes of the individual parts you're uploading?
  • v

    Vitali

    04/26/2022, 11:11 PM
    We have an undocumented difference from S3 here - every part except the last must be equal size
  • v

    Vitali

    04/26/2022, 11:11 PM
    Need to update the docs on that piece
  • j

    john.spurlock

    04/26/2022, 11:11 PM
    oohhh
  • j

    john.spurlock

    04/26/2022, 11:12 PM
    definately not always the case
  • j

    john.spurlock

    04/26/2022, 11:12 PM
    that might have been the thing
  • v

    Vitali

    04/26/2022, 11:12 PM
    You're uploading variable size parts?
  • j

    john.spurlock

    04/26/2022, 11:12 PM
    do you support reuploading the same part with a different size?
  • j

    john.spurlock

    04/26/2022, 11:12 PM
    yes for testing i did
  • v

    Vitali

    04/26/2022, 11:12 PM
    currently no but I've been meaning to lift that
  • v

    Vitali

    04/26/2022, 11:13 PM
    Right now the validation happens in UploadPart but I need to fix that to happen in the publish step
  • j

    john.spurlock

    04/26/2022, 11:13 PM
    it seems to succeed btw, at least reuploading part 1
  • v

    Vitali

    04/26/2022, 11:13 PM
    The last part is fine to change the size of
  • j

    john.spurlock

    04/26/2022, 11:13 PM
    but i think maybe that was one of the times it seemed wedged afterwards
  • j

    john.spurlock

    04/26/2022, 11:13 PM
    ok def good to know!
  • v

    Vitali

    04/26/2022, 11:13 PM
    The implications are weird and unintuitive
  • j

    john.spurlock

    04/26/2022, 11:14 PM
    same limitation on uploadpartcopy? i'm looking at that now
  • j

    john.spurlock

    04/26/2022, 11:14 PM
    ie same size parts
1...616263...1050Latest