On Tue, Oct 24, 2017 at 6:06 AM, Nick Couchman <[email protected]> wrote:
>
> >
> > Great - sounds like we're in agreement. ;)
> >
> > I'll retag things as described and create the
> > "staging/0.9.14-incubating" branches [1]. Once the staging branches
> > exist, depending on the issue, some PRs may need to be updated to
> > point to the new branch as their merge base.
> >
> > - Mike
> >
> > [1] http://guacamole.incubator.apache.org/release-procedures-
> > part1/#release-branch
> >
>
> Probably should add GUACAMOLE-384 (server PR 118) to this list, too - I
> think it's close to being done, and seems like a good bug fix to have
> rolled into this release.
>

As the fix within #118 for GUACAMOLE-384 (a low-impact segfault during
disconnect) involves changing the order resources are freed prior to
joining on the SSH client thread, there is the potential here for a
regression resulting in deadlock, which would affect the stability of
the Guacamole server overall as stale connection processes pile up. As
much as it hurts the eyes to see that segfault, it doesn't affect the
stability of connections since it occurs only during disconnect.

Given the severity of such a regression, I would be against expanding
established 0.9.14-incubating scope to cover GUACAMOLE-384. The fix
needs more testing, particularly on other platforms, to ensure that
things do unblock as expected, that something else wasn't overlooked
during review that could lead to deadlock, etc.

- Mike

Reply via email to