This is very helpful advice.  Yes, I have a feeling this issue I'm
seeing is related to a specific machine; which happens to be my
squeezebox server as well.  I did install a TCP/IP hack to increase the
limit of TCPIIP connections since I use this machine for several
different TCP/IP related services.  However,that should help things, but
make things worse.

Ever since I added secondary accounts to slacker and sirius radio on
mysqueezebox.com, I noticed the problem happens much less frequeently
(it used to be sometimes every 5 minutes.. now its every 45 minutes - 1
hour)...even longer if I dont do anything TCP/IP related.  I guess I
could try to increase the buffer to a very high value; including the
timeout to see what kind of affect it has.

The condition I'm pointing out is definitely not that bad; and in fact,
works great.   However, I have so many internet services running on this
machine, you probably wouldnt believe it... I have 12 TV tuners on this
machine serving through the internet, ASP.net based apps, FTP, and many
other TCP/IP based services.

The good news is I inreased the reliabilty considerably by adding 2
accounts (one for each client).  I dont know why that helped, but
there's not question that it did.  I will also take a closer look at
tcp/ip based activity using my favorite utility TCPView; which works
okay.  

Thank you for your support for this product with minor enhancements.  I
dont expect any major changes.  As long as logitech doesnt abandon their
customers who bought the classic squeezebox hardware devices, I think us
softsqueeze users will be fine.

bpa;515339 Wrote: 
> As I said before I cannot test these services as they are not available
> to me.
> 
> Softsqueeze is not my code - I just do some basic maintenance on it to
> keep it ticking over until SqueezePlay is ready (which is taking longer
> than expected).
> 
> Changing Softsqueeze to add some additional status would probably
> require changes in SBS which I cannot do so it is unlikely. Spoftsqueeze
> is a strange beast, within SBS it is subclassed of a Transporter model
> as it was  used to test new products graphics but its audio/network is
> not quite an SB2 so it has a minimal implementation of the slimproto
> protocol especially since changes in 7.3.
> 
> Your best bet would be to isolate the problem. I think you need to
> record sessions (probably on both sides of local SBS) using something
> like wireshark (was Ethereal). Note time when recording started and the
> time when the problem occur - examine the recording forensically at the
> problem timepoint and go backwards. I've done this a number of times
> tracking rare protocol events which required recordings of many hours -
> it is tedious but that is what it will require. 
> 
> From your description, I think the problem could be associated with a
> TCP reset which is used to clear half-open TCP connection but often
> still ends with a half open connection but stuck.   Applications need to
> look for the ECONNRESET error but often with "don't block on read" on a
> TCP socket - the Reset error code is overlooked as code assumes there is
> no data to read.  I think there could be issues handling ECONNRESET in a
> proxy configuration (e.g. should a RESET be propagated or absorbed or
> link closed by the proxy).
> 
> To get started, I suggest looking at the Reset count in the TCP
> connection statistics (e.g. netstat -s) on both sides of the SBS and
> from both ends (i.e. monitor 4 sets of counts).  The Reset count usually
> only increment at beginning and end of connections. See if Reset count
> is increment around the time you have a "hiccough" in your audio.


-- 
mkanet
------------------------------------------------------------------------
mkanet's Profile: http://forums.slimdevices.com/member.php?userid=7965
View this thread: http://forums.slimdevices.com/showthread.php?t=62060

_______________________________________________
plugins mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/plugins

Reply via email to