| 11:34.12 | *** join/#asterisk-doc leifmadsen (n=Leif@asterisk/documenteur-extraordinaire/blitzrage) |
| 11:34.12 | *** mode/#asterisk-doc [+o leifmadsen] by ChanServ |
| 12:32.32 | *** join/#asterisk-doc aae (n=aae@217.145.120.198) |
| 12:32.52 | aae | Hi, can anyone here help with the problem of the "root" user responding as unavailable on trying to log in? |
| 13:49.09 | *** join/#asterisk-doc leifmadsen (n=Leif@asterisk/documenteur-extraordinaire/blitzrage) |
| 13:49.10 | *** mode/#asterisk-doc [+o leifmadsen] by ChanServ |
| 14:10.37 | *** join/#asterisk-doc RobH (n=RobH@rob.tech.wikimedia.org) |
| 14:41.40 | leifmadsen | jsmith!!! |
| 14:41.42 | leifmadsen | feeling better? |
| 14:47.05 | jsmith | No, but I've got work to do |
| 14:47.47 | leifmadsen | jsmith: :( |
| 18:33.06 | *** join/#asterisk-doc leifmadsen (n=Leif@asterisk/documenteur-extraordinaire/blitzrage) |
| 18:33.06 | *** mode/#asterisk-doc [+o leifmadsen] by ChanServ |
| 18:58.23 | *** join/#asterisk-doc RobH (n=RobH@rob.tech.wikimedia.org) |
| 19:38.13 | *** join/#asterisk-doc RobH (n=RobH@cpe-173-169-30-118.tampabay.res.rr.com) |
| 19:39.50 | leifmadsen | ping! |
| 19:40.33 | leifmadsen | so I want to create some documentation on distributed agents for a queue where you have 2 (or more) asterisk systems. I'm just curious what kind of topology I should discuss as I don't want to build something too "complex" all right off the bat |
| 19:40.47 | leifmadsen | I tend to come up with schemes that are more complex than they need to be to illustrate my point |
| 19:41.35 | jsmith | I'd go with this use case: |
| 19:41.44 | jsmith | Jacob runs a call center. |
| 19:41.55 | jsmith | Jacob has two Asterisk systems (we'll call them A and B) |
| 19:42.21 | jsmith | Jacob wants his call center distributed across both systems, so that if one system fails, it doesn't disrupt his entire call center. |
| 19:43.34 | leifmadsen | jsmith: do you mean the same queue on both systems, or different queues on different systems? |
| 19:43.48 | leifmadsen | i.e. sales on A and tech support on B ? |
| 19:48.00 | jsmith | leifmadsen: I'd go with one queue initially... |
| 19:50.51 | leifmadsen | jsmith: well that doesn't really work because there is no way of communicating the queue position for a caller as they come into the queue. For example, if someone enters the queue on A (incoming caller, not agent), and they are position 1, then someone enters the queue on B and they are position 2, then there is no correspondence between the systems to state which position the caller is in |
| 19:51.50 | leifmadsen | I'm thinking of having the case where I have two queues, tech support 1 and tech support 2, but have phones on both systems, and agents who could potentially login to either (or both) tech 1 and tech 2 |
| 19:52.03 | jsmith | OK. |
| 19:52.22 | jsmith | Distributed agents, but not distributed queues, in other words |
| 19:52.56 | leifmadsen | right, we don't have distributed queues in Asterisk |
| 19:53.15 | leifmadsen | note my first question, "...documentation on distributed agents for a queue..." |
| 19:53.16 | leifmadsen | :) |
| 19:53.32 | leifmadsen | I'd love distributed queues, but we don't have them yet :( |
| 20:08.21 | leifmadsen | anything i'm missing from this outline? |
| 20:08.22 | leifmadsen | http://www.pastebin.ca/1665356 |
| 20:17.40 | jsmith | http://www.pastebin.ca/1665373 |
| 20:20.13 | leifmadsen | which part did you change? The intro? |
| 20:25.20 | ctooley | leifmadsen, couldn't queue position be communicated between the boxes via DUNDi? |
| 20:26.26 | leifmadsen | ctooley: possibly, but there is no hook to do that afaict |
| 20:26.52 | leifmadsen | russellb and I had discussed that previously about adding something near the beginning of Queue() that did something like that |
| 20:27.10 | leifmadsen | some sort of event would have to be pushed across the servers in the cluster to distribute that information |
| 20:27.27 | ctooley | Sounds like a textbook example of what DUNDi is designed for. |
| 20:28.14 | leifmadsen | well... I'm not 100% convinced DUNDi is the best tool for that, especially since we have an event framework with OpenAIS for distributing events |
| 20:28.25 | leifmadsen | and soon XMPP via the issue opened on the issue tracker |
| 20:28.36 | leifmadsen | (XMPP will allow distributed device state over WAN links) |
| 20:29.44 | ctooley | well, whether it's DUNDi, XMPP or OpenAIS isn't really all that much of an issue. SIP NOTIFY or MESSAGE packets could do the same thing. I don't know all that much about Queues as I've never implemented them, but it seems like a reasonably logical thing to have. |
| 20:33.38 | leifmadsen | ctooley: I agree |
| 20:33.45 | leifmadsen | it's just a matter of convincing someone to create the patch :) |
| 21:57.02 | jsmith | ctooley: It can be done with DUNDi... I did it that way three or four years ago :-p |
| 21:57.11 | jsmith | ctooley: It's a little tricky to do, though :-p |