[vdr] lagged responce on Nexus 2.2

Jon Burgess jburgess at uklinux.net
Tue Oct 4 23:22:25 CEST 2005


Vladimir Shved wrote:
> That does not seem to be an issue here, output should be attached with this
> message.
> 
> --Vlad
> when lagged, playing recording(nothing is recorded)
>  2  1      0 769560  15368  47308    0    0     0     0  869   603  1  1  0 98
>  2  1      0 769560  15368  47308    0    0     0     0 2934   559  2  7  2 89
> procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
>  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
>  3  2      0 769560  15380  47308    0    0     0    12 3690   541  0 13  0 87
>  2  2      0 769560  15380  47308    0    0     0     0 3689   549  3  9  0 88
>  2  1      0 769560  15380  47308    0    0     0     0 3703   560  1 11  1 87

I reckon the very high "wa" number may be a clue. This indicates that 
the system is spending almost all of its time waiting. Normally this 
indicates a disk IO bottleneck. What seems slightly unusual in your case 
is that you have next no IO recorded (si/so/bi/bo).

I see a blip of IO activity at startup due to VDR reading of the 
recording list and is associated with real disk IO. This stops after 
around 10 seconds:

[root at shark ~]# vmstat 1
procs -----------memory---------- ---swap-- -----io---- --system-- 
----cpu----
  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us 
sy id wa
  0  2  85192 308404   9336  96340    0    1    40    27   85    25  2 
1 96  1
  0  1  85192 301776   9692  96540    0    0   428   372  807  1682 13 
10  0 77
  0  1  85192 300040  10004  96704    0    0   444   128  690  1403 22 
6  0 72
  0  1  85192 298552  10292  96892    0    0   468    52  843  1425 27 
4  0 69
  2  1  85192 297064  10628  97144    0    0   588     0  747  1451 26 
7  0 67
  2  1  85192 295500  10968  97420    0    0   616     0  711  1417 25 
5  0 70
  0  2  85192 292392  11324  97828    0    0   764     0  724  1627 15 
9  0 76
  0  2  85192 290904  11752  98236    0    0   836     0  761  1814  4 
3  0 93
  0  2  85192 289912  12084  98456    0    0   552     0  671  1529  3 
3  0 94
  0  2  85192 288920  12440  98684    0    0   584     0  699  1562  3 
5  0 92
  0  2  85192 287432  12832  99100    0    0   808     0  737  1646  4 
6  0 90
  1  2  85192 286068  13284  99408    0    0   756    36  753  1674  4 
4  0 92
  0  0  85192 285588  13428  99524    0    0   260     0  601  1035  6 
3 47 44
  0  0  85192 285572  13428  99524    0    0     0     0  552   802  3 
0 97  0
  1  0  85192 285572  13428  99524    0    0     0     0  536   773  4 
0 96  0
  1  0  85192 285572  13428  99524    0    0     0     0  551   868  4 
0 96  0
  0  0  85192 285572  13428  99524    0    0     0     0  519   784  4 
0 96  0
  0  0  85192 285572  13428  99524    0    0     0     0  653   927  2 
1 97  0

When i'm playing a recording my system shows regular block-in but is 
 >95% idle:
[root at shark ~]# vmstat 1
procs -----------memory---------- ---swap-- -----io---- --system-- 
----cpu----
  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us 
sy id wa
  2  0  85164 277768  13596 105656    0    1    40    27   85    25  2 
1 96  1
  0  0  85164 277396  13596 106040    0    0   384     0  649  1044  3 
0 97  0
  0  0  85164 276652  13596 106552   60    0   572     0  551   827  3 
0 94  3
  0  0  85164 276032  13596 107064    0    0   512     0  679  1084  2 
0 98  0
  0  0  85164 275428  13596 107704    0    0   640     0  573   846  3 
0 97  0
  0  0  85164 275180  13604 107960    0    0   256    52  649   895  3 
1 94  2

I think it would be worth noting which processes are high in the 'top' 
list, is it VDR?

I've seen VDR occasionally switch the system time backwards and forwards 
repeatadly which can upset some of VDR's internal time() based loops 
causing spikes of CPU usage. Are there any messages like these in 
/var/log/messages?

Sep 25 15:09:14 vdr[4631]: System Time = Sun Sep 25 15:09:14 2005 
(1127657354)
Sep 25 15:09:14 vdr[4631]: Local Time  = Sun Sep 25 15:09:11 2005 
(1127657351)
Sep 25 15:09:15 vdr[4631]: System Time = Sun Sep 25 15:09:15 2005 
(1127657355)
Sep 25 15:09:15 vdr[4631]: Local Time  = Sun Sep 25 15:09:18 2005 
(1127657358)

It might be worth trying to turn off the time-update option to see if 
that has any effect.

	Jon



More information about the vdr mailing list