[mythtv-users] recent "TFW -- took a long time" warnings

[Warpme]

On 5/31/12 9:50 PM, Warpme wrote:
> Hi*
>
> Recently our BE starts to have regularly report about delays in TFW:
>
> 2012-05-31 19:00:12.405548 W 
> TFW(/myth/tv/23809_20120531190000.mpg:50): write(57528) cnt 55 total 
> 3175696 -- took a long time, 2507 ms
> 2012-05-31 19:00:12.450454 W TFW(/myth/tv/9071_20120531185000.mpg:76): 
> write(57716) cnt 7 total 389912 -- took a long time, 1400 ms
> 2012-05-31 19:15:20.906030 W 
> TFW(/myth/tv/21109_20120531191500.mpg:58): write(57528) cnt 12 total 
> 653300 -- took a long time, 1113 ms
> 2012-05-31 20:05:10.326524 W 
> TFW(/myth/tv/23809_20120531200500.mpg:51): write(57716) cnt 50 total 
> 2861548 -- took a long time, 2885 ms
> 2012-05-31 21:05:11.311221 W TFW(/myth/tv/8105_20120531210500.mpg:51): 
> write(57716) cnt 19 total 1047912 -- took a long time, 1221 ms
>
> It started week or two ago - but I can't correlate it with anything 
> special regarding my HW or SW.
> HW was untouched and SW in as well beside routinely upgraded mythtv 
> (build every week or two).
> Initially I suspect failing HDD so I replaced HDD with /myth volume to 
> new one.
> OS+DB HDD remains untouched.
> What surprise when after installing new shiny 3TB WD green I still see 
> those errors.
> So my question is: where issue might be ?
> Do anybody have idea how to nail down root cause of those errors.
> What interesting - system works OK (I don't observe any anomalies).
>
>
> some details:
> v0.25-128-ga9edae2-dirty-20120529, archlinux, kernel 3.3.7, 1st HDD WD 
> Green 2TB with OS/DB, 2nd WD Green 3TB with myth vol.
>
> -br
>
>
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://www.mythtv.org/mailman/listinfo/mythtv-users
Well,
If anybody experiencing issue like described in my original post:
it is worth to check kernel VM settings regarding dirty pages swapping.
Pls do test with following:

echo 300 > /proc/sys/vm/dirty_writeback_centisecs
echo 102400 > /proc/sys/vm/dirty_background_bytes
echo 864000 > /proc/sys/vm/dirty_expire_centisecs
echo 10 > /proc/sys/vm/dirty_ratio

kick:
watch -n1 grep -A 1 dirty /proc/vmstat

& look how nr_dirty related to nr_dirty_threshold (shouldn't cross it).

If all is OK You can put in /etc/sysctl.conf

vm.dirty_writeback_centisecs = 300
vm.dirty_background_bytes = 102400
vm.dirty_expire_centisecs = 864000
vm.dirty_ratio = 10

In my case since 10 days no singe TFW error (and sys is quite busy...)

-br
-------------- next part --------------
A non-text attachment was scrubbed...
Name: warpme.vcf
Type: text/x-vcard
Size: 83 bytes
Desc: not available
URL: <http://www.mythtv.org/pipermail/mythtv-users/attachments/20120622/0a0e7ca0/attachment.vcf>