[mythtv-users] mythbackend sometimes goes to 100%
Michael T. Dean
mtdean at thirdcontact.com
Fri Jun 19 21:00:22 UTC 2009
On 06/19/2009 04:46 PM, Bob Cottingham wrote:
> On Fri, Jun 19, 2009 at 9:49 AM, Michael T. Dean wrote:
>> On 06/19/2009 10:25 AM, Adam Felson wrote:
>>> About 18-24 months ago, mythweb's recorded programs page would cause
>>> mythbackend to go to 100%.
>>> This has mostly been fixed, but I still occasionally find mythbackend at
>>> 100% and have to kill it. It's mostly just an annoyance, but it does cook
>>> the cpu.
>>> I'm using the release-0-21-fixes branch of svn.
>>> Has this been fixed in 22 development?
>> There was never any change to either 0.21-fixes (including all of the
>> pre-0.21 development, where the issue was noticed) or trunk (since 0.21 was
>> released) that was known to have fixed the issue. It just appeared, and a
>> lot of people started looking at it, then it disappeared. (And, as you can
>> imagine, fixing a bug that you can't find is /very/ difficult, so there
>> hasn't really been any progress since it disappeared.)
>> The issue has been reported by some people since then (generally as an
>> uncommon/unreproducible occurrence), but no one has been able to track it
>> down, let alone fix it. Therefore, I'd recommend sticking with 0.21-fixes
>> rather than upgrading.
> I had my backend going to 100% every day or two with .21-fixes without
> any apparent pattern until I turned off the upnp server when starting
> the backend
Have you tried recently with the UPnP server enabled? You /definitely/
shouldn't see problems that often with current 0.21-fixes as it has
virtually disappeared and seems very infrequent on most systems.
> and I've only had it happen once since then over a couple
> of months. I thought the upnp issue was still a known problem?
It's an issue that doesn't seem to affect any of the devs (and no one
who is affected has tracked it down/fixed it). But it should be very
rare for everyone, now.
More information about the mythtv-users