Hello, I've been working with this too in trying to voicetrack remotely from a satellite office to a central office with some success. The issue we face is limited bandwidth over a VPN.
Here is the approach I tested: 1. Sync files with a Cloud Service like SpiderOak This type of service ($10/month or so) allows for reasonable prices, backup, and easy distribution across multiple nodes on the WAN. Configure /var/snd on all machines to be "synced" and updated files appear on all nodes almost instantly. Compared to RSync, this service does a much better job managing the network traffic. Also - use MP2 as the back-end for files. 2. VPN for the MySQL database connection 3. Modification of source code for the Voicetracker to reduce MySQL traffic Item #3 is the key to make the experience usuable. I looked at the network traffic of the Voicetracker program when saving, inserting, and modifying a track and there appears to be quite a bit sent. From looking at the code, the entire log seems to be sent to MySQL each and every time there is a slight change to the log. With a fully loaded log, this can take 30-60 seconds of a reasonable link where there is network management in place. These sorts of things are reasonable to expect in most radio envioronments. There was a patch that was released to only transmit the log on modification, or on final save if the actual voicetrack was changed. I modifed several source files on 2.8.1 and was able to get this to work. It still is very chatty with MySQL, and has some times where there is a large delay in sending the full log. All said, this patch makes remote voicetracking possible versus unusable. The basis of the patch is here, it will need to be manually modified to work with newer versisons: https://www.mail-archive.com/rivendell-dev@lists.rivendellaudio.org/msg00421.html RDLogedit was designed for local network access, and thus we see this sort of behavior. Architecturally, it would great to modify RDLogedit to work over a WAN. This would require: 1. Migrating to the Web Services vs. MySQL for data 2. Modification of the Client and Server component to transmit only Deltas of the log 3. Implementation of some sort of locking mechanism for when the log is being edited Short of these modifications, the only thing I could think of was to use the patch suggested above. Maybe we could get this put into the general code base if it tests out alright? I'd be happy to share the modified files with the development team if they would like to consider putting it into the code base - please email me privately. Brian On Sunday, March 2, 2014 12:11 PM, Fernando Della Torre <f...@vdit.com.br> wrote: I Think rsync'ing your /var/snd to the local machine would help a lot. After recording you could rsync back only the voice files. Anyway accessing MySQL over a WAN link it's not the optimal. Regards. 2014-03-02 10:19 GMT-03:00 Morten Krarup Nielsen <morte...@gmail.com>: I've tried option 1. >There's a 20/20 fiber connection in both the studio and at my home. At best it >works a little slow, but I can live with that. Some times the log gets >corrupted though. > >We store our music in wav. It would work faster if you work with MP3. I >thought about using Bittorrent Sync for /var/snd but haven't done any >experimentation yet. > > > > >2014-02-28 16:17 GMT+01:00 Keith Thelen <kthe...@kanabec.net>: > > >Hello all! >> >>I know there's no official remote voicetracking feature at this point. But >>from digging through past posts, I gather there's people out there who have >>managed to assemble something that accomplishes the same function, more or >>less. >> >>As I see it, there's at least two ways of doing this: >> >>1) Provide the person who will be voicetracking with a computer running >>Rivendell, let them connect over a VPN, tweak things to obtain acceptable >>results. (There seems to be a few ways in which people have done this… local >>copy of the library, a patch that reduces log writes, etc. What are the best >>techniques here?) >> >>2) Provide the person who will be voicetracking with a copy (printed, text, >>PDF, whatever) of the log in question, and receive a pile of individual >>voicetracks which must then be imported and placed. (This seems like an ugly >>solution, but on a technical level it's the simplest. Who, if anyone, is >>doing this? What tricks have you used to reduce the labor involved?) >> >>Any hints would be appreciated… don't want to reinvent the wheel if I can >>help it. >> >> >> >>--------------- >>Keith Thelen >>Kanabec Systems >> >>
_______________________________________________ Rivendell-dev mailing list Rivendell-dev@lists.rivendellaudio.org http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev
