Bug 3694 - Control Connection Congestion
: Control Connection Congestion
Status: RESOLVED WORKSFORME
Product: SB 2/3
Classification: Unclassified
Component: Misc
: 55
: PC Windows XP
: P2 normal (vote)
: ---
Assigned To: Chris Owens
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2006-07-04 07:18 UTC by Marc Auslander
Modified: 2006-09-10 15:54 UTC (History)
1 user (show)

See Also:
Category: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Marc Auslander 2006-07-04 07:18:06 UTC
I turned my SB3 on this morning.  It had been used the night before, and the server was up continuously all night.

I selected a (long) mp3 track via the Browse Music Folder path and started it.

It started playing, but the play kept stopping and starting.   Display scrolling stopped and started.  Trying to use the remote, reponse was very bad and at one point the display went black for several seconds.

Wireless strength was good.

I unplugged and replugged the SB3.  Reconnected to the server.  I was surprised to find that the clip started playing again, with a proper display and good response to the remote.

Looking at server health, which had been enabled, I was control connection congestion reported.

I'm wondering if there is some problem in the firmware which gets into a state where it can't clear its queues.
Comment 1 Marc Auslander 2006-07-04 15:34:53 UTC
I'm seeing continued bad control behavior - mostly jumpy scrolling of the display or delated remote response.  server health shows no problems at all.

Control Connection     : OK
Streaming Connection   : OK
Player Signal Strength : OK
Buffer Fullness        : Low
Server Response Time   : OK
Comment 2 KDF 2006-07-04 16:13:18 UTC
Triode, any idea what the network test could be indicating here?

I'm not sure that this is a bug, rather than a support problem. (yet)
Comment 3 Marc Auslander 2006-07-04 16:39:22 UTC
This behavior started with Firmware 55.  I'm thinking of going back to 54 to see if it goes away.
Comment 4 Adrian Smith 2006-07-05 11:13:52 UTC
Control congestion means that writes to the slimproto socket block and the queue in select is building up.

This indicates a that tcp is not ready to receive more data, so either the network conditions are worse or the player is not taking stuff out of the receive buffer due to some internal issue.  I would suspect the network.

Marc - What server version are you running - if 6.3 could you download the NetTest plugin and measure what the highest bandwidth you achieve is and see how it varies.  Plugin is available here: http://wiki.slimdevices.com/index.cgi?PluginDiagnostics
Comment 5 Marc Auslander 2006-07-05 11:47:06 UTC
I am running the product 6.3.0.

I tried a complete reset - still see the problem.

I have now returned to firmware 54, and so far I see no problems, but it will be tomorrow before that is anything like definitive.  I'll update this tomorrow.

Since the problem is intermittent, its a little hard to pin down.  And it appears, strangely, that the congestion builds up when I'm stopped and then causes trouble the next time I try to run.

With 54, nettest is showing 100% av at 500kbps and 37% av at 6000kbps.  I'm running a B/G network with a B laptop which comes on and off the net.

nettest shows better performance with less variability running B only.  So maybe I was seeing short drops.  I will try 55 again B only and try 54 B/G to try to pin this down.

Is it possible that there were changes between 54 and 55 related to clearing congested queues that don't work well in my situation?
Comment 6 Blackketter Dean 2006-07-07 17:57:41 UTC
chris to follow up
Comment 7 Marc Auslander 2006-07-10 06:22:14 UTC
An update.

I think this is related to network drop outs as has been suggested. My other SB3 has better wireless and I haven't seen it do this.

 It occurs in inactive state when the TOD screen saver is active.  But once it happens, it won't clear unless I power cycle the SB3.  I did not see it with 54 (I went back for a while) but its intermettent so I can't guarantee that its a 54-55 change.

In this state, the control connection switches from congested to fail in a several minute cycle.

Is there any server debug info that would help?

FamilyRoom

Please queue up several tracks to play on this player and start them playing. Then press the Reset Counters link above to clear the statistics and update this display.
Summary

Control Connection     : Fail
Streaming Connection   : Inactive
Player Signal Strength : Poor
Server Response Time   : OK

Warnings


Player Signal Strength
This graph shows the strength of the wireless signal received by your player. Higher signal strength is better. The player reports signal strength while it is playing.

    < 10 :        0 :  0% 
    < 20 :       97 :  1% 
    < 30 :      927 : 13% ######
    < 40 :      100 :  1% 
    < 50 :     2083 : 29% ##############
    < 60 :     1984 : 27% #############
    < 70 :     1334 : 18% #########
    < 80 :      696 : 10% ####
    < 90 :        0 :  0% 
   < 100 :        0 :  0% 
   >=100 :        0 :  0% 
    max  : 73.000000
    min  : 16.000000
    avg  : 50.922725

Buffer Fullness

    < 10 :     5946 :100% ##################################################
    < 20 :        0 :  0% 
    < 30 :        0 :  0% 
    < 40 :        0 :  0% 
    < 50 :        0 :  0% 
    < 60 :        0 :  0% 
    < 70 :        0 :  0% 
    < 80 :        0 :  0% 
    < 90 :        0 :  0% 
   < 100 :        0 :  0% 
   >=100 :        0 :  0% 
    max  : 3.838603
    min  : 0.000000
    avg  : 0.173686

Control Connection

     < 1 :   208070 : 94% ###############################################
     < 2 :      561 :  0% 
     < 5 :     1057 :  0% 
    < 10 :      711 :  0% 
    < 20 :      806 :  0% 
    >=20 :     9918 :  4% ##
    max  : 657.000000
    min  : -1.000000
    avg  : 8.416433
Comment 8 Chris Owens 2006-09-08 14:41:46 UTC
Hi Marc, are you still on fw54?

I tried some experiments back when we thought this was being caused by the reduction in signal strength some users were reporting at the time, but didn't ever see this.  And those posts seem to have died down.

Nonetheless, I am still concerned about this bug.
Comment 9 Marc Auslander 2006-09-10 15:41:19 UTC
I have not seen this nearly as often.  And now with Firmware 62, my SB3's seem to use the good access point consistently.    
So i may not be a good test case anymore.
Comment 10 Chris Owens 2006-09-10 15:54:36 UTC
Well, I'm just going to mark it as 'WORKSFORME' for now.  As always (speaking to all bug db users) if you see this bug, feel free to reopen it and we will have another look.