[mythtv] Ticket #10770: backend stops responding
David Engel
david at istwok.net
Fri Jul 6 18:51:48 UTC 2012
On Fri, Jul 06, 2012 at 01:54:35PM -0400, Steven Adeff wrote:
> On Fri, Jul 6, 2012 at 1:20 PM, David Engel <david at istwok.net> wrote:
> > On Fri, Jul 06, 2012 at 12:34:49PM -0400, Steven Adeff wrote:
> >> On Fri, Jul 6, 2012 at 12:11 PM, David Engel <david at istwok.net> wrote:
> >> > On Fri, Jul 06, 2012 at 11:24:01AM -0400, Steven Adeff wrote:
> >> >> On Fri, Jul 6, 2012 at 11:19 AM, <noreply at mythtv.org> wrote:
> >> >> > #10770: backend stops responding
> >> >> > --------------------------------------------+----------------------------
> >> >> > Reporter: Steven Adeff <adeffs.mythtv@…> | Owner: gigem
> >> >> > Type: Bug Report - General | Status: closed
> >> >> > Priority: minor | Milestone: 0.25.2
> >> >> > Component: MythTV - Scheduling | Version: 0.25-fixes
> >> >> > Severity: medium | Resolution: fixed
> >> >> > Keywords: | Ticket locked: 0
> >> >> > --------------------------------------------+----------------------------
> >> >> > Changes (by David Engel <dengel@…>):
> >> >> >
> >> >> > * status: infoneeded => closed
> >> >> > * resolution: => fixed
> >> >> >
> >> >> >
> >> >> > Comment:
> >> >> >
> >> >> > In [changeset:ca273324961b95e300890ed40e833249b5460bd8/mythtv]:
> >> >> > {{{
> >> >> > #!CommitTicketReference repository="mythtv"
> >> >> > revision="ca273324961b95e300890ed40e833249b5460bd8"
> >> >> > Hopefully remove more scheduler/recorder deadlocks.
> >> >> >
> >> >> > Change the scheduler to unlock schedLock when calling the recorder
> >> >> > synchronously. If the recorder happens to be calling the scheduler at
> >> >> > the same time, this allows the recorder's call to grab the lock as
> >> >> > needed.
> >> >> >
> >> >> > Fixes #10647, #10770 and #10771 .
> >> >> > }}}
> >> >>
> >> >> will this be backported to 0.25?
> >> >
> >> > Yes, but probably not for a few days to make sure it doesn't cause
> >> > unexpected problems in master. If you'd like to try it yourself on
> >> > 0.25, there is only one hunk that fails to apply and it's trivial to
> >> > fix by hand.
> >>
> >> I'm currently running the patch attached to this ticket, is that the
> >> same fix? it's almost run for an entire day no issues yet, but I would
> >> give it 72hrs before saying conclusively.
> >
> > The core changes are the same. The newer version adds some protection
> > against an obscure and unexpected race condition and won't miss
> > sending a SCHEDULE_CHANGE event.
>
> does that make my testing with it useful at all?
Yes, your testing is still very useful.
> (ie should I bother responding any further or just wait for this to
> propagate to fixes?)
Just keep running the patch you already have and let us know if you
see any more deadlocks.
David
--
David Engel
david at istwok.net
More information about the mythtv-dev
mailing list