• snowe@programming.devOPM
      link
      fedilink
      arrow-up
      3
      ·
      11 months ago

      yeah in addition to that I’ll do a full writeup, but had several problems

      1. lemmy-ansible had pict-rs set to 0.4.0, which is quite an old version, meaning migration couldn’t occur concurrently (it was copying files to object storage sequentially)
      2. upon stopping the migration at one point and attempting to restart (on a newer pict-rs version) the server had completely filled up, so the migration could no longer continue on either version
      3. therefore I had to mount more block storage, copy the entire drive over to the block storage, in order to perform the migration.

      there were lots more issues, but copying hundreds of gigs many times over is a cause to lot of the downtime.