gingerlime
gingerlime
We're not using `fork_worker` (rather using `prune_bundler`), but we noticed recently that it takes longer time to notice changes after deploy, much slower than it was with puma 3.x (we...
sounds like exactly the same thing to me, but happy to open a new issue 👍
Hey @nateberkopec 👋 . I work with @joker-777. We completely understand the need to reproduce it. We can reproduce it some times on our staging environment, but we so far...
We're still bumping into this, yet it's hard to reproduce reliably, even on our staging environment. One thing that definitely jumps out however is that the `raise GemNotFound, "Could not...
side-note: would you be happy with a PR to better log when worker fails? at the moment it only logs the first item from the backtrace, and not even the...
@nateberkopec we found a workaround, and also managed to reproduce the issue hopefully. The workaround for us was to install `io-wait:0.2.1` globally using `gem install io-wait:0.2.1`. We're still not sure...
I'm not sure I fully understand, but obviously I'm not familiar with Bundler or Puma at an intimate level. In particular why the problem only surfaces with phased-restart, after clearing...
> Note that one can delete old releases, just not the original one that started Puma. I.think. We definitely delete old releases, including the one that started puma and it...
I just noticed that when we run `bundle update rails` it actually removes the `io-wait` dependency in Gemfile.lock (in rails 7.0.3, we were testing rails 7.0.2.3, [maybe here?](https://github.com/rails/rails/commit/dae605c764f9fe205753498a790bb49134123fed)). So I...