[Opencdn-users] Re: Configuring a Darwin relay from VBrick and Starback
Brought to you by:
aalef
Menu
▾
▴
[Opencdn-users] Re: Configuring a Darwin relay from VBrick and Starback
|
From: Megan T. <tro...@os...> - 2005-03-25 20:36:56
|
Alessandro, I played around a bit today and did some testing. It seems that the RTSP URL for the live streams will ALWAYS be in the form: rtsp://192.148.244.131/.sdp/ViDe/ViDe_X.sdp where the X will be either 1, 2, or 3. The streams will be disconnected and reconnected as each session ends and a new one begins. The ViDe account on our Starbak has only 3 lines and there can only be one active call on each line at a time. Once the files are archived, they will have very different names. I don't know which line will be associated with which talk other than the Monday talk from 2:30 PM to 4:30 PM. That will be "PBS and Higher Education" Maybe Chip can share his cheat sheet with you so you know which X will be which session. Best, Megan At 01:08 PM 3/25/2005, Alessandro Falaschi wrote: >Megan, > >thank you for your analysis. In the meanwhile, I did put on-line this >page http://ocdn.mirror.garr.it/ where a static link to a content coming >from Internet2 Commons Starbak is placed. Better: is a static link to >the request to be sent to OpenCDN control, in order to see a recorded >content from the Starbak, being redistributed by OpenCDN. > >In the same way, if session names can be known in advance, many links >like that, can be listed in the page. > >On the other side, OpenCDN distribution is at its infancy, so that I >don't belive that many people will configure a local Darwin replication >node. But nobody knows. > >As this is no more than a simple experiment, I don't expect to much, but >if some visibility can be gained for the OpenCDN Project, well I will be >happy. > >Also, on Easter and Angel's Monday, I don't think I will work on this >neither. Anyway, thank you again > >Alessandro > >Il ven, 2005-03-25 alle 16:00, Megan Troyer ha scritto: > > Alessandro, > > I won't be the one actually running the Starbak for the event. I believe > > that Chip Eberhart's crew will be doing that. I don't know that he'll > know > > the stream titles ahead of time, but that might already be in the > > plans. I'm not 100% sure how the Starbak configures the RTSP URLs (some > > experimentation may be necessary) so that the URLs can be planned for > ahead > > of time. > > > > I'm not sure if Chip's group will want to take on the "only by you" role > > you describe below. I'm more of an H.323 person and streaming is > something > > that I do only with the Starbak server which is an H.323 device more > than a > > streaming device. I don't have any experience with Darwin and very little > > with Unix. I don't have the time to learn OpenCDN before Monday, not to > > mention I'll be in a car on the way to the ViDe Conference on Monday, so > > I'll be disconnected. (I'm not lucky enough to have that Verizon Cellular > > based ISP for my laptop either.) > > > > As far as I'm concerned you are more than welcome to duplicate the streams > > over your OpenCDN network. I haven't been asked to password protect the > > streams, so they'll be open to anyone who knows the URL of the Starbak > server. > > > > Best, > > Megan > > > > > > At 06:46 AM 3/25/2005, Alessandro Falaschi wrote: > > >Megan, all the others interested, > > > > > >I succeeded in modifying the code (it was really simple), and now at > > >http://labtel.ing.uniroma1.it/oCDN/perl/index.pl is possible to request > > >the feeds form Internet2 and SurfNet to be distributed through OpenCDN > > >:-) Just the Dutch TV stops quite immediately. Remember, that just is > > >the "general" test page, I will put an ad-hoc one elsewhere, when things > > >stabilize. > > > > > >Having a different stream title (and RTSP URL) for session (as Megan > > >says below) is like to watching TV, and having to cycle power at every > > >commercial. But I understand that the name change is required for proper > > >StarBack archiving, so we have to live with it. > > > > > >I think that 'simple' OpenCDN unattended retransmission can be done in > > >these ways: > > > > > >- only by me: > > > * I should know in advance the exact names of the sessions, and place > > >different mount points. OpenCDN is not dealing with time schedules by > > >know, so the publication page will report these too, or link > > >http://commonsvcg.oar.net/ViDe for user information. > > > > > >- only by you: > > > * if you can run an OpenCDN "Origin" daemon entity, you can restart it > > >with proper configuration data when session changes, and the actually > > >published URL will be used by OpenCDN. Or: > > > * you don't bother with OpenCDN entities and configuration, and > > >maintain a local (Darwin) relay which sources the different session > > >names, and provide a unique URL which can be used by OpenCDN > > > > > >More complex schemes can be envisaged (such as periodic parsing of your > > >announcement page by my part, or reception of an RSS feed from StarBack) > > >but they seem really out of our actual reach. > > > > > >In the meanwhile that I set up an ad-hoc page, maybe someone can have > > >some other ideas :-) > > > > > >Best, > > >Alessandro > > > > > >Il ven, 2005-03-25 alle 08:38, Megan Troyer ha scritto: > > > > Alessandro, > > > > When the streams go live, they will be updated for each session with an > > > > appropriate title for each session, so in that sense, the streams > will be > > > > going down regularly. Pair that with the 3 consecutive track > sessions on > > > > Tuesday afternoon and Wednesday, you may have your hands full with > static > > > > set-ups. > > > > > > > > As an additional FYI--the streams, when live, will be available at > this > > > URL: > > > > http://commonsvcg.oar.net/ViDe > > > > > > > > Of course, there's nothing there now and probably will not be there > until > > > > Monday afternoon. > > > > > > > > Best, > > > > Megan > > > > > > > > > > > > At 08:54 PM 3/24/2005, Alessandro Falaschi wrote: > > > > >I said : > > > > > > > > > > > ...when I try to (manually) configure a Darwin relay for the > unicast > > > > > > feed (rtsp://192.87.116.121/stream2), its DESCRIBE /stream2 request > > > > > > receives a RTSP /1.0 400 Bad Request error from VBrick > > > > > > > > > >I figured out why, and the sad thing, is that it happens also with > > > > >internet2 Starback :-( > > > > > > > > > >When my mp4layer client sends an RTSP DESCRIBE request, it says > > > > > > > > > >DESCRIBE > > > > >rtsp://commonsvcg.oar.net/Internet2/I2-RTC-BOF_860/I2-RTC-BOF_860.ce > > > > >RTSP/1.0\r\n > > > > > > > > > >on the contrary, when Darwin attempts to build a relay, it says > > > > > > > > > >DESCRIBE /Internet2/I2-RTC-BOF_860/I2-RTC-BOF_860.ce RTSP/1.0\r\n > > > > > > > > > >i.e. it is missing of the host and protocol part. As evident, although > > > > >correct, this upsets Starback and VBrick. > > > > > > > > > >The good thing is that... by inserting the full URL as a mountpoint in > > > > >the Darwin Relay configuration page, IT WORKS !!!! > > > > > > > > > >The bad thing is that OpenCDN builds relays interpreting the "mount > > > > >point" in a relative fashion - and not absolute. Next thing will > be, to > > > > >see how this can be made compatible with OpenCDN. I think there is not > > > > >much time for modifying the code in this sense, but a static set up on > > > > >top of OpenCDN should work, i.e., with a static relay from there to > > > > >here, and the "Origin" settled here. This means that if the static > feed > > > > >goes down, it can't be automatically set up again (real OpenCDN > Origins > > > > >can, instead). > > > > > > > > > >tomorrow I will try > > > > >-- > > > > >Alessandro Falaschi <al...@in...> > > > > > > > > Megan Troyer 453 Baker Systems > > > > Technology Coordinator 1971 Neil Avenue > > > > OARnet/ OSU--CIO Columbus, OH 43210 > > > > support: 614-247-8066 office: 614-292-8443 > > > > tro...@os... cell: 614-946-9613 > > > (emergencies only) > > > > Megan Troyer 453 Baker Systems > > Technology Coordinator 1971 Neil Avenue > > OARnet/ OSU--CIO Columbus, OH 43210 > > support: 614-247-8066 office: 614-292-8443 > > tro...@os... cell: 614-946-9613 > (emergencies only) Megan Troyer 453 Baker Systems Technology Coordinator 1971 Neil Avenue OARnet/ OSU--CIO Columbus, OH 43210 support: 614-247-8066 office: 614-292-8443 tro...@os... cell: 614-946-9613 (emergencies only) |
