Old 22nd October 2012, 15:47   #1
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
Anticipating hardware needs

I'm trying to anticipate hardware needs for my new little talk radio station. We've got a big broadcast coming up tomorrow (the schedule is irregular at this point).

I'm trying to estimate how many listeners I should assume my server can handle for a broadcast so I can set appropriate limits. I've got a Linux cloud server with 4 processors and 4 GB RAM. It shares space with another site that gets about 50,000 unique visitors per month, but overall load average is only around 0.70.

Bandwidth shouldn't be an issue because we don't broadcast all the time. We do small casts an hour or so at a time.

We're broadcasting at mono 64 kbs, 44.1 kHz.

Any ideas how to calculate listener resource usage?
vahost is offline   Reply With Quote
Old 22nd October 2012, 17:10   #2
DrO
 
Join Date: Sep 2003
Posts: 27,873
that's really not an easy thing to work out as the other processes on the machine will also effect how many listeners the DNAS can cope with. really with what you've got, unless you're getting thousands of listeners (which has taken the large stations years to get), you really just need to ensure that you're not going to be bandwidth limited as the DNAS (v1 and v2 [when fixed] ) can deal with 1000+ on modest hardware, it's the bandwidth that causes the main issue.

so that probably isn't the answer being looked for but it's tricky to say x will allow you to do y as there's too many unknown factors to give something like that. even OS updates give variations in performance on the machine. really the only thing i can best suggest is to get the DNAS running and locally make a number of curl connections to act as test listeners and see how far it goes before the DNAS, etc cannot cope (usually known by the stream starting to skip).

-daz
DrO is offline   Reply With Quote
Old 22nd October 2012, 17:15   #3
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
That's still helpful, though, the 1000+ figure. All I was looking for was a ballpark, and that gives me an idea of the ballpark. Thanks!
vahost is offline   Reply With Quote
Old 22nd October 2012, 17:22   #4
DrO
 
Join Date: Sep 2003
Posts: 27,873
really the best way to know is to just stress-test the server and see what it does. like i can get 1500+ on one quad-core VM without issue but only safely 600 on a raspberry pi or only a few hundred at best on a low powered single-core VM.

-daz
DrO is offline   Reply With Quote
Old 22nd October 2012, 17:58   #5
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
Well, with any luck, a stress test is going to happen.

Is there a way to change the maximum listener limits in my sc_serv instance without killing and then restarting the instance?
vahost is offline   Reply With Quote
Old 22nd October 2012, 20:48   #6
DrO
 
Join Date: Sep 2003
Posts: 27,873
the v2 DNAS can do that - just edit the config file and use the reload admin page method.

-daz
DrO is offline   Reply With Quote
Old 22nd October 2012, 22:14   #7
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
What is the reload admin page method?
vahost is offline   Reply With Quote
Old 22nd October 2012, 22:17   #8
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
Oh! Never mind. I found it. I didn't even know that was there!

Is there an equivalent way to update sc_trans with a new item in a playlist? Is that what "Reload Reserved List(s)" is all about?
vahost is offline   Reply With Quote
Old 22nd October 2012, 22:18   #9
DrO
 
Join Date: Sep 2003
Posts: 27,873
like using SIGUSR1?

-daz
DrO is offline   Reply With Quote
Old 22nd October 2012, 22:22   #10
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
Quote:
Originally Posted by DrO View Post
like using SIGUSR1?

-daz
So does kill SIGUSR1 [PID] actually kill and recreate the process very quickly? Does this result in a lost transcoder signal that causes the audio stream to fail?
vahost is offline   Reply With Quote
Old 22nd October 2012, 22:23   #11
DrO
 
Join Date: Sep 2003
Posts: 27,873
it doesn't kill the sc_trans process - being done via 'kill' is probably what is causing your confusion. it's just a signal the sc_trans process can detect and when it does it reloads the main playlist.

the full list sc_trans supports is at http://wiki.winamp.com/wiki/SHOUTcas...tional_Signals

-daz
DrO is offline   Reply With Quote
Old 22nd October 2012, 22:29   #12
vahost
Junior Member
 
Join Date: Sep 2012
Posts: 25
Great, thank you!
vahost is offline   Reply With Quote
Reply
Go Back   Winamp & Shoutcast Forums > Shoutcast > Shoutcast Technical Support

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