[mythtv-users] ivtv0 warning: ENC Stream 0 OVERFLOW
Al McIntosh
junkmail at allanmcintosh.com
Mon Jan 9 02:54:43 UTC 2006
Endaf Jones wrote:
>On 1/8/06, Al McIntosh <junkmail at allanmcintosh.com> wrote:
>
>
>>chris at cpr.homelinux.net wrote:
>>
>>
>>
>>>On Sat, Jan 07, 2006 at 09:34:41AM -0500, Al McIntosh wrote:
>>>
>>>
>>>
>>>
>>>>>>Dec 15 07:28:27 mythbox kernel: ivtv0 warning: ENC Stream 0 OVERFLOW
>>>>>>#2:
>>>>>>Stealing a Buf
>>>>>>fer, 512 currently allocated
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>More memory does not appear to help. Did anyone manage a fix for this?
>>>>
>>>>
>>>>
>>>>
>>>In my case, it turned out that my network card was on the same IRQ as
>>>all three IDE channels (software RAID-5), and while the box would
>>>usually record OK, if I was doing something network-intensive (like
>>>watching TV on the remote Myth frontend) the network card would
>>>generate so many IRQs it was delaying the drives to the point where
>>>the front-end would stutter, ivtv would overflow, and eventually the
>>>RAID driver would fail a drive. That would have been fine, except
>>>that the ethernet watchdog (tulip driver) would panic the kernel. I
>>>tried forcing the cards to specific IRQs but the tulip driver
>>>wouldn't behave. I switched to a different ethernet port and ALL of
>>>the problems went away.
>>>
>>>
>>>
>>Interesting, I have a rather fast mythbox, AMD 2800+ with 1G ram and a
>>single pvr250. The buffer overflow often occurs in the middle of the
>>night when the only load on the machine is mythbackend recording.
>>
>>
>
>Perhaps when mythfilldatabase is running?
>
>
>
The last few occurances have been in the evening or night.
mythfilldatabase has been running around 11am.
I do have mythcommflag set to run when the recording starts.
More information about the mythtv-users
mailing list