Old 8th June 2011, 17:53   #1
schematicdnb
Junior Member
 
Join Date: Jun 2011
Posts: 6
SC_Trans causing frequent buffering client-side

I've searched endlessly for a couple weeks now and have not been able to find a solution for this, or really even anyone else having this problem so I'm finally posting about it myself.

I'm using SC_Trans 2 Beta Drop 40 on a VPS, and listeners are getting frequent buffering. Now there are a number of oddities that don't seem to make any sense together so I'm just going to put them in a list:

- First, I know it's an issue with SC_Trans because if I eliminate it from the equation (connect a source like EdCast directly to the server, instead of to the transcoder) the problem goes away.
- This is happening with both sc_serv 2 beta, and the stable DNAS 1.9.8
- The length of the buffer varies from listener to listener, as well as the frequency though the time between buffers seems to be random and sporadic. This is happening to ALL listeners, but some more than others.
- The VPS has a 100Mbit connection, which is barely being touched so there is no risk of it being an actual bandwidth issue. Listeners also typically have 5Mbit+ connections.
- Log files show absolutely nothing. No indication of any errors, buffering, anything of the sort. All that shows up in the logs is the usual source and client connections.

I'm completely stumped. I COULD just not use the transcoder since that eliminates the problem, but I've not found any other way to handle remote DJs/scheduling/Auto playlists as nicely as the SC_Trans does.

Halp? Thanks in advance!
schematicdnb is offline   Reply With Quote
Old 9th June 2011, 08:42   #2
MADxHAWK
Senior Member
 
Join Date: Jan 2010
Posts: 181
Well without further informations it would be hard to find an answer.

Are sc_trans and DNAS running on the same system? If so are you using the external ip to connect from sc_trans to DNAS or do you use localhost (127.0.0.1).

With VPS i think you mean a Virtual Privat Server. The problem with virtual servers are, they sharing memory and cpu-time with other virtual servers on the same system. So maybe another server causing a very high cpu load or request a lot of memory. Since sc_trans have to encode every song on the fly, a high cpu load on the server could cause a buffer on the stream and i guess, that would be the most common reason for your problem.

-MAD
MADxHAWK is offline   Reply With Quote
Old 9th June 2011, 16:20   #3
schematicdnb
Junior Member
 
Join Date: Jun 2011
Posts: 6
Same system, windows 2008 based VPS... I was using the external IP I believe, switching to localhost didn't help though.

Right now I seem to have at least made the buffering more tolerable with an ENORMOUS buffer size, i.e. adaptive @ 240 seconds but this of course creates a huge delay between server and listener.

However, I'm still getting this pretty often:

2011-06-09 12:10:09 E msg:[UVOX2] 4157 [UVOX2] 4157 socket send error.
2011-06-09 12:10:09 E msg:[UVOX2] 4157 Waiting 60 seconds

Which shuts everything down for a minute, and then when it comes back up the buffering problem is back and worse than ever regardless of what the buffer is.
schematicdnb is offline   Reply With Quote
Old 9th June 2011, 16:43   #4
DrO
 
Join Date: Sep 2003
Posts: 27,873
i tried to briefly debug a similar issue with the Transcoder on a linux VPS which was showing a similar issue to what you're seeing (based on a newer build of the Transcoder).

i wasn't able to get too far other than it seems that the VPS in that case wasn't getting the time needed for the Transcoder to run properly (leading it to spike the cpu usage and then get shutdown). so it could be the same issue especially if you're having to set the buffer so high to try to keep it filled. but generally, the Transcoder is more sensitive to cpu availability unlike some other 3rd party solutions (something i'd like to get resolved but there's no guarantee).

with the timeout issue, i've logged it as something to make configurable for a future version (as the 60 second timeout isn't too good for such cases and hinders me when working on it as well).

-daz
DrO is offline   Reply With Quote
Old 9th June 2011, 17:05   #5
schematicdnb
Junior Member
 
Join Date: Jun 2011
Posts: 6
The CPU usage on the VPS itself never goes over 15% could it be that another VPS on the same server is hogging the cpu but that doesn't register in mine?

I suppose I'll ticket my VPS host and see what can be done about cpu usage.
schematicdnb is offline   Reply With Quote
Old 9th June 2011, 23:23   #6
DrO
 
Join Date: Sep 2003
Posts: 27,873
anything is possible when it comes to VPS instances especially depending on how the provider has configured things so it could be possible for what cpu usage is reported to be different from what is actually happening. but it's all a bit of guess work sadly and could be something else that i've not come across yet.

-daz
DrO is offline   Reply With Quote
Old 10th August 2011, 18:17   #7
schematicdnb
Junior Member
 
Join Date: Jun 2011
Posts: 6
Just wanted to update this a little: this is still an issue with the current builds of the DNAS (GM build 29) and Transcoder (Beta drop 51).

However, I've just tried running the transcoder from my home PC and connected to the DNAS on my VPS and everything appears to be running smoothly... so that isolates the problem as being the transcoder on the VPS.

I'm going to look into having my VPS migrated to a different node and see if that makes any difference - but is there any chance the issue could be looked into a little more?

Cheers
schematicdnb is offline   Reply With Quote
Old 11th August 2011, 11:45   #8
DrO
 
Join Date: Sep 2003
Posts: 27,873
i can try to find some time but when i don't have the means of a true VPS i'm not sure what i can do - the nearest is my linux VM's though they do cause buffering issues but that's from not having enough cpu time to keep on top of things. unfortunately the Transcoder is somewhat sensitive which usage in a VPS can expose at times.

-daz
DrO is offline   Reply With Quote
Old 12th August 2011, 08:06   #9
schematicdnb
Junior Member
 
Join Date: Jun 2011
Posts: 6
OK as far as I can tell it had something to do with the way the Xen HVM was either set up or possibly Xen itself.

I've switched from the Xen HVM Windows VPS to a OpenVZ Linux VPS (CentOS 6) and am running everything through SSH and so far it's working flawlessly (and on lower specs than the Windows box was). Hurray for things working and saving some money!
schematicdnb is offline   Reply With Quote
Old 12th August 2011, 10:16   #10
DrO
 
Join Date: Sep 2003
Posts: 27,873
hehe, that's good to hear

-daz
DrO is offline   Reply With Quote
Reply
Go Back   Winamp & Shoutcast Forums > Shoutcast > Shoutcast Technical Support

Tags
buffer, skip, trans

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump