[mythtv-users] Re: Bad night of recording - everything hosed

Mark Knecht markknecht at gmail.com
Wed Sep 21 13:50:46 UTC 2005


On 9/21/05, Mark Knecht <markknecht at gmail.com> wrote:
> Hi,
>    Starting last evening we had a really bad night with recording. It
> seems that the system was set up to record two programs in parallel
> pretty much all through the evening. Additionally there were one or
> two people viewing existing recordings while recording was going on.
> AFAICT it appears that every program was recorded with some sort of
> problem such that now it plays like you were doing a fast forward on
> your remote.
>
>    Uptime on the backend server is about 3 weeks.
>
>    Looking at the logs, again starting last evening, I see this:
>
> 2005-09-20 12:00:07.481 New DB connection, total: 1
> 2005-09-20 12:00:07.513 New DB connection, total: 2
> 2005-09-20 12:00:07.557 Using protocol version 15
> 2005-09-20 12:00:07.563 adding: dragonfly as a client (events: 0)
> 2005-09-20 12:08:50.639 adding: dragonfly as a client (events: 1)
> [mpeg2video @ 0xb792c810]ac-tex damaged at 16 7
> [mpeg2video @ 0xb792c810]Warning MVs not available
>
> 2005-09-20 13:30:13.189 New DB connection, total: 1
> 2005-09-20 13:30:13.212 New DB connection, total: 2
> 2005-09-20 13:30:13.236 Using protocol version 15
> 2005-09-20 13:30:13.242 adding: dragonfly as a client (events: 0)
> 2005-09-20 13:30:41.905 IOBOUND - blocking in ThreadedFileWriter::Write()
> 2005-09-20 13:30:52.318 IOBOUND - blocking in ThreadedFileWriter::Write()
> 2005-09-20 13:30:56.520 IOBOUND - blocking in ThreadedFileWriter::Write()
> 2005-09-20 13:31:01.730 IOBOUND - blocking in ThreadedFileWriter::Write()
> 2005-09-20 13:31:05.732 IOBOUND - blocking in ThreadedFileWriter::Write()
> 2005-09-20 13:31:09.933 IOBOUND - blocking in ThreadedFileWriter::Write()
>
> 12:00 noon would have been a recording of Charlie Rose. 1:30PM would
> have been a recording of Sensible Chic. Most likely the system was
> doing mythcommflag on CR when the 1:30 show started, but there's
> nothing very new about this. Further none of the programs recorded
> before 8PM (4 programs) seem to skip, while everything after 8PM (10
> programs) all skip.
>
>    After sending this email I'm be rebooting the server but I'm
> wondering if anyone else has seen anythign like this.
>
>    What a mess!
>
> - Mark
>
Answering myself, it appears that the NFS storage server, which also
happens to be a Myth frontend machine, had two spontaneous reboots
yesterday. From that machine's logs:

Sep 20 13:00:01 myth14 cron[11670]: (root) CMD (test -x
/usr/sbin/run-crons && /usr/sbin/run-crons )
Sep 20 13:00:01 myth14 cron[11671]: (root) CMD (rm -f
/var/spool/cron/lastrun/cron.hourly)
Sep 20 13:10:01 myth14 cron[11683]: (root) CMD (test -x
/usr/sbin/run-crons && /usr/sbin/run-crons )
Sep 20 13:20:01 myth14 cron[11695]: (root) CMD (test -x
/usr/sbin/run-crons && /usr/sbin/run-crons )
Sep 20 13:22:01 myth14 syslog-ng[5027]: syslog-ng version 1.6.8 starting
Sep 20 13:22:01 myth14 syslog-ng[5027]: Changing permissions on
special file /dev/tty12
Sep 20 13:22:01 myth14 Linux version 2.6.12-gentoo-r6 (root at myth14)
(gcc version 3.3.5-20050130 (Gentoo 3.3.5.20050130-r1,
ssp-3.3.5.20050130-1, pie-8.7.7.1)) #2 Tue Aug 2 16:31:31 PDT 2005
Sep 20 13:22:01 myth14 BIOS-provided physical RAM map:
Sep 20 13:22:01 myth14 BIOS-e820: 0000000000000000 - 000000000009fc00 (usable)
Sep 20 13:22:01 myth14 BIOS-e820: 000000000009fc00 - 00000000000a0000
(reserved)Sep 20 13:22:01 myth14 BIOS-e820: 00000000000e8000 -
0000000000100000 (reserved)Sep 20 13:22:01 myth14 BIOS-e820:
0000000000100000 - 000000000bf40000 (usable)
Sep 20 13:22:01 myth14 BIOS-e820: 000000000bf40000 - 000000000bf50000
(ACPI data)
Sep 20 13:22:01 myth14 BIOS-e820: 000000000bf50000 - 000000000c000000
(ACPI NVS)Sep 20 13:22:01 myth14 BIOS-e820: 00000000ff7c0000 -
0000000100000000 (reserved)Sep 20 13:22:01 myth14 191MB LOWMEM
available.


So I guess that following the reboot maybe the mythbackend machine
never got properly in sync with NFS somehow? It seems that the IOBOUND
lines in the backend server came after the NFS server had completely
rebooted so I don't understand that, other than to say that maybe the
backend machine was able to buffer video for 5 minutes before things
got crazy...

Bummer.

- Mark


More information about the mythtv-users mailing list