2 Replies Latest reply on Jun 19, 2020 9:02 AM by brianp9

    Compiled binaries broken on import

    brianp9 Lurker

      Hey!

       

      I've got an OVA file that once imported is on Debian 10 and has got two ruby applications running on it. This OVA works as expected in VirtualBox, Fusion, and ESXi 6, 6.5, and 7 on my test machine.

       

      Here's the issue. We've got a group of people who are all downloading and importing this exact OVA into the same versions of ESXi and for whatever reason the ruby apps fail to start. Upon investigation the rubygems that have native binaries are all incompatible with the system. They all require being recompiled again before the app will work.

       

      I thought maybe if they were running on 32-bit systems that could be a problem but I've confirmed they are all running 64-bit as are my test machines.

       

      I'm at a loss at this point. I can't think of any reason why some ESXi users would end up with incompatible binaries while everyone else can import and use the OVA just fine.

       

      Any ideas?

      Cheers.

        • 1. Re: Compiled binaries broken on import
          daphnissov Guru
          Community WarriorsvExpert

          First thought is download corruption. Publish the checksum for your OVA that you know was tested successfully, and have users verify that hash against their copy.

          • 2. Re: Compiled binaries broken on import
            brianp9 Lurker

            I really do need to make checksum checking a support habit. I don't often ask about it.

             

            I did not have all the users verify that it downloaded without corruption. Although the checksum is published so they may be in the practice of checking it.

            I know one of the users did confirm when they initially reached out that they had gone back to validate the checksum after seeing the problem.

             

            The other issue with the idea of download corruption is it's to consistent. Every user who's seen the problem has seen it display, and had it resolved, in exactly the same way. I think the issue is oddly to consistent for it to be download corruption.