1. 5

  2. 3

    I feel your pain. Pretty much every time I use Docker, now. So many questions and so much hair pulling…

    1. 3
      1. 1

        Awesome. I’ve updated my blog post. Glad I could help contribute. :)

      2. 2

        Thanks for painting the picture. I knew about the same bits and pieces of this problem but never really grasped it.

        This should be fixable, no? I might hack at it this weekend. Or this evening because it’s bugging me now.

        1. 1

          There are two bugs really.

          1. SpamPD shouldn’t lose its flags on HUP. So that should be as simple as finding that signal handler in the Perl code and making sure the new process has the same argv. SpamPD still seems maintained (they had commits 4 months ago) so I went ahead and creates an issue since I didn’t see it in the closed list:


          1. Why is OpenBSDs rc system sending SpamPD a hangup signal? Does it send every process a hangup signal? Why does that happen on boot and not when you call rcctl directly? There’s gotta be a reason for that behaviour. That’d probably involve hitting up their mailing lists or getting on IRC. I might jump on IRC later today.
          1. 2

            1, SpamPD is pretty small and doesn’t explicitly do anything with HUP. The processing, and the logging of the restart, is happening somewhere in Net::Server(::PreForkSimple). SpamPD might need to capture HUP to control the restart of the server.

            1. I don’t know what’s going on there. The rc scripts don’t send a HUP. Maybe something in the handoff from rc and the resulting login tty.