[mythtv-users] Odd i/o issues

Kevin Kuphal kkuphal at gmail.com
Tue Jan 13 22:22:50 UTC 2009


On Tue, Jan 13, 2009 at 3:58 PM, Michael D. O'Brien <obrienmd at gmail.com>wrote:

>
>
> On Tue, Jan 13, 2009 at 3:13 PM, Michael D. O'Brien <obrienmd at gmail.com>wrote:
>>
>>> In the last week I've upgraded my back and frontends to Ubuntu 8.10 and
>>> the newest myth-fixes.  Last night, I noticed that the show I was viewing
>>> was pausing every few seconds, rendering it nearly unwatchable.  I knew the
>>> backend was recording a few other shows and the frontends hadn't shown this
>>> behavior while the backend was idle earlier that day, so I did some looking
>>> around:
>>>
>>> -top on the backend and frontend showed plently of available memory and
>>> idle cpu
>>> -it's not a network issue, the maximum usage in iftop on the frontend
>>> looked to be about 12mbps (out of 100) and on the backend around 30mbps (out
>>> of 1000), due to recording 2 shows off our hdhomerun while I was watching
>>> this other show.
>>> -on the backend, top shows iowait hovering around 20-40%, which isn't bad
>>> and was common while recording 3 HD shows and watching 1
>>> -on the backend, iostat -xk 1 shows something interesting: my raid5 array
>>> participants are all being used, but it seems that two of the drives hover
>>> around 10% usage, while one is constantly pegged at 100% usage.  Is this
>>> normal behavior in iostat -xk for a raid5 array?  it seems odd to me.
>>>
>>> My backend is a C2D e6320 with 3GB of ram running Ubuntu 8.10 and the
>>> newest myth-fixes.  Myth records to a raid5 array of 3x750gb drives with a
>>> 1.5tb xfs file system on top.  It's connected to the network via 1000bT
>>> ethernet.
>>>
>>> My frontends are AMD x2 4000+ with 1GB of ram running Ubuntu 8.10 and the
>>> newest myth-fixes.  They are connected to the network via 100bTX ethernet.
>>>
>>> Thoughts?  I'm intrigued by the what seems to be strange iostat results
>>> on the backend...
>>
>>
>> smartd?
>>
>
> SMART values look fine for all drives.  Also checked xfs file system, it's
> clean and not fragged at all.
>

Actually, I was referring to whether smartd may have been running a check on
the disk during that high I/O period.

Kevin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20090113/e9f8ffdd/attachment.htm 


More information about the mythtv-users mailing list