Closed GoogleCodeExporter closed 9 years ago
somehow the message was lost
https://dl.dropbox.com/u/3394987/flowplayer.cluster-3.2.8.zip
it will now dispatch resolving failures correctly back to the player.
the correct config for http resolving is
hosts: [
// server side load balancer
'http://balance.flowplayer.org/video/',
// direct URLs to the servers behind the balancer
'http://pseudo01.hddn.com/vod/demo.flowplayervod1/',
'http://stream.flowplayer.org1/'
]
which are the paths to the url
url: 'Extremists.flv',
no idea how that example came about
Original comment by dani...@electroteque.org
on 26 Jun 2012 at 9:35
I have no idea either, but I also do not understand what you're saying.
Both:
http://pseudo01.hddn.com/vod/demo.flowplayervod1/
http://stream.flowplayer.org1/
do not exist - while those without the terminating "1" do.
http://balance.flowplayer.org/video/
does not exist either.
And the "balancer" is surely supposed to _balance_, and not to error out in
this demo - as opposed to:
http://flowplayer.org/demos/plugins/streaming/clustering.html
which uses an invalid domain on purpose.
Plus the client-side load balancing _seems_ to work in this demo:
http://flowplayer.org/demos/plugins/streaming/cluster-loadbalance.html
Obviously with links to _existing_ directories and files. If you were right,
this one would stall as well.
Original comment by blacktrashproduct
on 26 Jun 2012 at 2:35
thats correct that was just an example to throw an error :)
http://pseudo01.hddn.com/vod/demo.flowplayervod/
http://stream.flowplayer.org/
the paths would be these
Original comment by dani...@electroteque.org
on 26 Jun 2012 at 2:56
Yeah, right, the filename has to be removed, I did not even look for such a
mistake. Perhaps I should remove the fictional "server-side load balancer" too
- but then there's not much difference left to:
http://flowplayer.org/demos/plugins/streaming/cluster-loadbalance.html
except for testing the onConnectFailed event (which, after all might be a good
idea).
Original comment by blacktrashproduct
on 26 Jun 2012 at 4:38
I will take care of this:
https://github.com/flowplayer/site/issues/532
Original comment by blacktrashproduct
on 26 Jun 2012 at 4:42
That's not correct, if you enable the loadBalance config it will turn on random
round robin connectivity similar to what a hardware load balancer or dns would
do but a bit simpler obviously.
The example has the filename in the path hence the issue.
Original comment by dani...@electroteque.org
on 27 Jun 2012 at 12:45
No, I fixed the problem with the filenames.
http://balance.flowplayer.org/video is a bogus address, and therefore the demo
now falls back to client-side clustering. These are the messages now returned
at
http://flowplayer.org/demos/plugins/streaming/cluster-loadbalance-advanced.html:
attempting to connect to: http://balance.flowplayer.org/video
server-side load balancer failed, enabling client-side load balancing
attempting to connect to: http://stream.flowplayer.org
started clip: http://stream.flowplayer.org/Extremists.flv
Which makes sense to me. Don't you need a server-side load balancer to do
server-side load balancing? If not, and if this is supposed to work with a 404
address (???) as load balancer, then there is a bug. Please clarify.
Reopening.
Original comment by blacktrashproduct
on 27 Jun 2012 at 7:37
round robin if configured means random connectivity attempts on the urls it
will allow to put multiple hosts and connect to a random one , it is
replicating what hardware would do just simpler. it looks fine to me now. the
bad addresses there are to demonstrate the fallback i suppose.
Original comment by dani...@electroteque.org
on 27 Jun 2012 at 9:42
it is ok to close now though i will organise releasing this change to send the
resolving failure back to the player.
Original comment by dani...@electroteque.org
on 27 Jun 2012 at 9:42
My comment was meant in the sense that it would be nice to actually see
server-side load balancing in action - as you could say that this is what the
demo is about. As it is we can only demonstrate that it falls back to
client-side clustering on failure.
What has changed and why (as it seems to work)?
Original comment by blacktrashproduct
on 27 Jun 2012 at 11:58
Oh right now I get you , whats the go there ! balance.flowplayer.org is meant
to be on a load balancer aparantly or so it says from the docs hehe. It's non
existant, so perhaps it's theoretical to display the fallback functionality
still. Perhaps change the wording a little to reflect that ?
Original comment by dani...@electroteque.org
on 27 Jun 2012 at 4:22
Well, it does so already, but it's a bit of lame excuse: "It's an imaginary
host and will never answer in reality. In normal situations it would serve
video files when it's up and running."
What was your change?
Original comment by blacktrashproduct
on 28 Jun 2012 at 3:49
lame excuse what do you mean ?
as above
"it will now dispatch resolving failures correctly back to the player. " as in
it will throw a player error not just trigger the onFailure callback.
Can we close this there was nothing wrong in reality just a documentation issue
;)
Original comment by dani...@electroteque.org
on 29 Jun 2012 at 1:46
From the docs
"We have three hosts on our cluster configured in our hosts property. The first
host is our server side load balancer. It's an imaginary host and will never
answer in reality. In normal situations it would serve video files when it's up
and running."
Looks fine to me and clear enough. Its there to demonstrate fallback.
http://flowplayer.org/demos/plugins/streaming/cluster-loadbalance-advanced.html
this page does not load properly on chrome because its loading some bloaty
facebook javascript ??
{fb: "http://connect.facebook.net/en_US/all.js"},
fails to load for me. when i click the image its attempting to click through to
a non existant file, there is a js error but its not displaying :)
I will close this for now as things look good, and will likely get out a new
release for this added failure callback with changes to the other plugins soon.
Original comment by dani...@electroteque.org
on 30 Jun 2012 at 2:02
Original issue reported on code.google.com by
blacktrashproduct
on 22 Jun 2012 at 4:19