[07:51] There is a user in #ubuntu who appears to have a defunct process whose parent is init, am I correct in thinking that shouldn't be possible? Am I correct in having them submit a bug report against upstart? [07:52] Output of "ps aux" showing that the process is defunct: http://paste.ubuntu.com/1349958/ output of "pstree" showing that it's owned by init, http://paste.ubuntu.com/1349935/ . [07:58] Jordan_U, no, it is entirely possible to have a defunct process under init [08:00] Jordan_U, I imagine you assume that all processes under init are controlled by upstart, but that is not the case [08:00] I did assume that, apparently incorrectly :) [08:01] Is PID 1 not upstart? [08:03] Jordan_U, imagine, for example, you run a process from the shell that daemonizes into the background, thus detaching from the shell process. Because the parent process is gone, init will then "adopt" the process. [08:04] Jordan_U, that does not mean however that the same rules apply for those processes that would apply for processes run by upstart [08:05] chris|: Is PID 1 not upstart? [08:06] Jordan_U, pid 1 is upstart [08:10] chris|: Is it not the responsibility of PID 1 to reap its children? Are there times when a child cannot be reaped? Am I misunderstanding something fundamental? [08:28] wow [08:28] I googled for 'upstart reap defunct' and your bug came up Jordan_U [08:28] "33 minutes ago" [08:29] Jordan_U: its possible, perhaps even likely, that the defunct process is stuck in a syscall [08:38] Jordan_U: BTW I would expect upstart to reap zombies yes [08:38] Jordan_U: finding where it actually does so in the code, however, is eluding me [08:38] * SpamapS heads to bed [08:39] SpamapS: I though that would leave it in a "D" state rather than a "Z" state (as shown in ps). [15:46] Jordan_U: agreed, Z should just mean that it has exitted, and init should be calling waitpid on it to let it die