Commits on Sep 23, 2024

1. Disable grab download resume …

   This is to mitigate cases like k0sproject#4296
   
   By default grab tries to resume the download if the file name determined from either the url or from content-type headers already exists. This makes things go side ways, if the existing file is smaller than the new one, the old content would still be there and only the "extra" new bytes would get written. I.e. the download would be "resumed". 🤦
   
   This is probably not a fix for the root cause in k0sproject#4296 as the only way I've been able to make grab fail with `bad content length` is by crafting a custom http server that maliciously borks `Content-Length` header.
   
   This is a minimal possible fix that we can easily backport. @twz123 is already working on bigger refactoring of autopilot download functionality that gets rid of grab. Grab seems to bring more (bad) surprises than real benefits. In the end, we just download files and we should pretty much always just replace them. No need for full library dependecy for that.
   
   Signed-off-by: Jussi Nummelin <jnummelin@mirantis.com>

   

   jnummelin committed Sep 23, 2024
   Configuration menu

   • View commit details
   • Copy full SHA for ff0da08
   • Browse repository at this point

   Copy the full SHA

   ff0da08 View commit details

   Browse the repository at this point in the history