[mythtv-users] Disk-drive hiccups

Bob Sully rcs at malibyte.net
Fri Jun 19 05:19:23 UTC 2009


James Crow wrote:
> Bob Sully wrote:
>> Hi all -
>>
>> Over the past several weeks, I have been having occasional instances of
>> partitions becoming inaccessible suddenly.  This has only happened with
>> the two partitions I use for Myth recordings and for video files.  Both
>> of
>> these are large JFS filesystems, have been for over a year with no prior
>> problems.  Both were initially 700GB partitions; I recently upgraded
>> both
>> drives to 1.5TB units.  I used the -c switch when doing the format (took
>> several hours on the 1.5T drives).  Even so, this is still happening;
>> sometimes it happens after an hour or two of playing media, sometimes
>> when
>> the system is idle (which is a real problem if the drive isn't available
>> when a recording starts (this happened once)).
>>
>> It's not a drive problem (hardware) because it's happened with the 750s
>> as
>> well as the 1.5T drives.  I'm wondering if it's a problem with JFS...?
>> I
>> don't want to go to ext3 because it takes a week (not literally, but it
>> seems like it) to fsck a big ext3 partition.  Would anyone recommend
>> that
>> I change over to XFS?
>>
>> Another cause COULD be software.  I had been running Ubuntu 8.10-x64,
>> then
>> upgraded to 9.04.  Soon afterward I started noticing this issue.  Has
>> anyone else seen this since a Jaunty upgrade?
>>
>> I'm using Jean-Yves' upgraded 0.21 binaries with VDPAU, which I'm quite
>> happy with.  This is a NOT a Myth issue, because I can't seem to access
>> the involved drive outside of Myth, either, after this happens; the only
>> fix appears to be a reboot.   I don't see anything in the system logs
>> which would clue me in as to WHY this happens, only that the OS can't
>> access the involved partition afterward.
>>
>> I did notice that when this happened yesterday, the involved drive
>> designator changed over from /dev/sdb to /dev/sdd....!  Curiouser,
>> curiouser!  I've not seen that happen before.
>>
>> Any ideas would be appereciated.  Thanks.
>> Bob
>>
> Bob,
>
>   I can't comment on JFS, but I am running Mythbuntu amd64 with 500GB
> (Seagate) and 1TB (WD) drives for file storage. I run XFS on my TV and
> video drives and have had no issues. Maybe you could check dmesg and
> other system logs to see if there are any errors related to the
> filesystem. You said that it happened with both the 750s and the 1.5s,
> maybe the new kernel has a bug/problem with your sata chipset.
>
>   I also had terrible problems with NFS. I am using CIFS for sharing my
> video directory between systems now. It seems that you are trying local
> access, but if not NFS may be a problem.
>
> Cheers,
> James
>

James:

Thanks for the reply.  No NFS for these drives.  I use it to mount a RAID
array on another box as a backup/NAS setup and another partition for a
secondary, "pooled" /home directory for my network.  But I've had enough
flakiness happen with NFS that I would NOT rely on it as storage for a
multimedia server, with the kind of prolonged relatively high bitrate
transfers necessary with HD-quality videos.

Anyway, I reformatted my *local* (eSATA, sitting on top of the Mythbox)
backup drive as XFS and we'll see what happens with it.  In the process of
doing the fresh backup to the new drive, the CPU load went up to 60(!),
and even after I killed the process, there was continuing heavy disk I/O
going on, and the load only dropped to 45.  This had happened before when
the machine was trying to access the drive that had "vanished".  This
time, I found about 30 minutes of unusual (and completely nondiagnostic)
garbage in the system log:

Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader last message repeated 2 times
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader last message repeated 2 times
Jun 18 19:50:39 vader kernel: r
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader kernel: r
Jun 18 19:50:39 vader kernel: <rr
Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader kernel: r
Jun 18 19:50:39 vader kernel: <3r
Jun 18 19:50:39 vader last message repeated 3 times
Jun 18 19:50:39 vader kernel: r
Jun 18 19:50:39 vader kernel: <r
Jun 18 19:50:39 vader kernel: <3r

I've been using Linux for over 10 years and have never seen kernel
messages like that before.  Anyone else know what it's trying to tell me??

Thanks...Bob
-- 
________________________________________
Bob Sully - Simi Valley, California, USA
http://www.malibyte.net
http://www.malibyte.com



More information about the mythtv-users mailing list