# HG changeset patch
# User Jun Wu
# Date 1481938472 0
# Sat Dec 17 01:34:32 2016 +
# Node ID 6c9ce8399350d8287599cd802b91adf73db08759
# Parent 69d25b06467d65bf6d1e85d34d8fc57ec321b51d
# Available At https://bitbucket.org/quark-zju/hg-draft
# hg pull https://bitbucket.org/qu
It's probably not worth worrying about just yet, but with Linux PID namespaces,
multiple processes may think they have the same PID, even if that PID maps to
different values inside the kernel.
Mozilla has encountered problems with multiple hg processes running in separate
containers and PID na
On Sat, 17 Dec 2016 16:40:12 -0800, Gregory Szorc wrote:
> It's probably not worth worrying about just yet, but with Linux PID
> namespaces, multiple processes may think they have the same PID, even if that
> PID maps to different values inside the kernel.
>
> Mozilla has encountered problems wi
Excerpts from Gregory Szorc's message of 2016-12-17 16:40:12 -0800:
> It's probably not worth worrying about just yet, but with Linux PID
> namespaces, multiple processes may think they have the same PID, even if that
> PID maps to different values inside the kernel.
>
> Mozilla has encountered
Jun Wu wrote:
> Pid namespace breaks mercurial lock, which is a symlink with the content
> "host:pid".
Would adjusting the format to:
host:pid:starttime:boottime fix this?
Sadly, it seems like process start time is actually stored as jiffies
since system boot on Linux, and NTP can make canonicali
Excerpts from timeless's message of 2016-12-18 23:52:05 -0500:
> Jun Wu wrote:
> > Pid namespace breaks mercurial lock, which is a symlink with the content
> > "host:pid".
>
> Would adjusting the format to:
> host:pid:starttime:boottime fix this?
No. A pid number in a pid namespace is highly like