From (spam-protected) Wed Jan 6 00:27:34 2010 From: (spam-protected) (Mitar) Date: Wed, 06 Jan 2010 00:27:34 +0100 Subject: [Nodedb-interop] Let's continue In-Reply-To: <4B14402D.6010506@guifi.net> References: <4B14212C.60604@tnode.com> <4B14402D.6010506@guifi.net> Message-ID: <4B43CAE6.7090509@tnode.com> Hi! OK, I have to do my thesis so I did not have much time this last weeks. But I setup a wiki we have been talking about. So that we can continue with work on interoperability. http://interop.wlan-lj.net/ After some thought I have decided for Trac wiki as it is not bloated, i is easily extendable (we can simply integrate source code repository and also other systems into its MVC framework if there will be some day need) and I have good experience with it. Hope it will serve all our purposes well. (Suggestions for improvements are of course welcome. Also if you have ideas for what we could automatize or integrate, for example there is also graphviz support: http://interop.wlan-lj.net/wiki/SandBox) I have configured wiki so that for editing registration is necessary. I think this is OK as it is useful to know who has edited or contributed something so that we can easier discuss things and also limits possible spam. After registration user has almost all rights on wiki, except deletion and editing of read-only pages (there are currently none). If anybody wants to share this administration burden with me, you are welcome. It is also setup in an independent virtual machine if there will be someday need to cooperate on the system itself. I have also made sections I think we could work on. If you think I have missed feel free to tell me. I have also populated wiki with some data we have send to the mailing list. And some of my suggestions. Please feel free to discus them. (One idea I was playing with is also a "best practices" wiki where we could publish things which really worked well, from routing protocols to good antenna designs. (It could be just link of course, it is not necessary to copy things.)) One interesting thing I come up with is a list of IP addresses: http://interop.wlan-lj.net/wiki/IPAddressing/List Which is automatically parsed and duplicated in the second table. I do not think we need a full-fetched allocation system (as offered by Ramon) as we are talking about whole ranges here and not independent nodes (for that I would recommend some other system based on common schema and data aggregation). And also with IPv6 we will not need precise cooperation on IP ranges anymore. I think this is the easiest for people to edit and add themselves and we could just add some checks and issue warnings if there are problems how people entered data. The other I would like to emphasize is our own top-level domain and DNS root servers we could work on? http://interop.wlan-lj.net/wiki/DNS/TLD And now about database schema. I think it would be useful if we now publish everything on the wiki and see what is common ground. For example Ramon's example is very good of what is missing in our schema which comes directly from experience and needs (we have currently only one city with network in Slovenia so we had little use of making a hierarchy for nodes, on other side we have worked on monitoring a lot). As have also Ramon's suggested our work will probably be mostly seeing what everybody has, combining it and name attributes and classes. I also think it would be useful to invite everybody again to participate. For example list of IP ranges is almost empty. So, let's continue! Mitar From (spam-protected) Fri Jan 8 10:58:55 2010 From: (spam-protected) (mickey) Date: Fri, 08 Jan 2010 10:58:55 +0100 Subject: [Nodedb-interop] Freifunk Node Database Message-ID: Hello world, triggered by Alx i've stumbled accross the interop project and i pretty much appreciate this idea. As some of you allready might know i have recently started a new project called Freifunk Node Database. It's main goal is to provide a simple and working solution for maintaining data of our local mesh network. We have had an installation of FFSomething for quite some time running but unfortunatly the users kept complaining about bugs and nobody was able to take care of it properly. This technical hurdle kept newcomers from joining our network and even let formerly active people leave :-( The new database follows the KISS design principle and is (currently) focused on rather "offline" data maintained by the node owners. As i've experienced with other projects like FFSomething and Netmon the aggregation of live data and ideas alike takes away the developers focus from the regular people's needs way too much. Therefore my approach is trying to keep the perspective of the technically unexperienced user as much as possible. In the long run i envision a smooth and fancy application that provides a "new node wizard" with automatic ip address generation and nearby subnet proposal based on geo location. The application is written in Rails because this is what i personally like best for web applications. And there are plenty of plugins that ease even the more complicated details, e.g. geokit, map_layers, et. al. You can find a live "demo" (actually our working copy) at http://db.hannover.freifunk.net The source code lives on http://github.com/knox/nodes I just have contributed some technical info about this project and our local mesh to the interop wiki allready and i am really looking forward for a great dialog within the interop project. Have a nice day, Knox (Freifunk Hannover, Germany) From (spam-protected) Fri Jan 8 14:58:12 2010 From: (spam-protected) (Mitar) Date: Fri, 08 Jan 2010 14:58:12 +0100 Subject: [Nodedb-interop] Let's continue In-Reply-To: <4B43CAE6.7090509@tnode.com> References: <4B14212C.60604@tnode.com> <4B14402D.6010506@guifi.net> <4B43CAE6.7090509@tnode.com> Message-ID: <4B4739F4.108@tnode.com> Hi! I have also added information about DNS to our list: http://interop.wlan-lj.net/wiki/IPAddressing/List Please fill it in. Because I discovered that without DNS there is of little use knowing IPs of some else's network if you want to access content or services there. For example we have access to Freifunk Hannover network because kiberpipa.net (also in Ljubljana) network is peering with them but do not know anything about that network and this peering is of little use - mostly just to show of that we peer! (This is just an example.) So I think we should be really working on letting others know what and how to access things in our networks. For example AWMN (Athens' network) have everything under DNS and with peering with them we can easily find content and services because we can just enter woogle.awmn in our web browser and have a Google like service which indexes many servers they have there. Or VoIP on www.voip.awmn and SIP on sip.awmn. Mitar From (spam-protected) Fri Jan 8 15:29:42 2010 From: (spam-protected) (Mitar) Date: Fri, 08 Jan 2010 15:29:42 +0100 Subject: [Nodedb-interop] Freifunk Node Database In-Reply-To: References: Message-ID: <4B474156.904@tnode.com> Hi! > The new database follows the KISS design principle and is (currently) > focused on rather "offline" data maintained by the node owners. As i've > experienced with other projects like FFSomething and Netmon the aggregation > of live data and ideas alike takes away the developers focus from the > regular people's needs way too much. Therefore my approach is trying to > keep the perspective of the technically unexperienced user as much as > possible. In the long run i envision a smooth and fancy application that > provides a "new node wizard" with automatic ip address generation and > nearby subnet proposal based on geo location. This is very very like what our ideas for our node database are. Please check it here: http://nodes.wlan-lj.net/ You can freely register and try with some test node to see how it works. So the idea is to make very smooth user experience for newcomers (thus also graphical designing everything and integration with wiki so that all additional information and help is easily available) while still allowing complex network analyzes and tools for power users. So our way is somehow like that: - user registers - user adds a node and can input as much of information as he/she knows, all else is automatically set (minimum is in fact just location and type of the router) - user presses a button on the web interface and gets complete firmware image with all configuration in it to his/her e-mail - user flashes the image with GUI flasher program (wizard) (on TODO) - user puts the router on location, plugs it into electricity and plugs it into Internet uplink (VPN) and this is it, plug&mesh So all this (except GUI flasher) already works and it is proven. We are also using MVC framework, but Python one, Django. And have put a lot network (health) monitoring and analysis into it and more will come in next version. But for example all nodes are reporting how they are working. Check graphs: http://nodes.wlan-lj.net/nodes/node/211639ea-a1be-4a2c-a742-1449268d775e Of course system allows also power users to do things differently, in this case it is just a node database. But if you want it can help you maintain your node very easily (and because voluntaries' time is very limited this means you spend less time maintaining nodes and more time building new ones). So even I rather use this system than to do things by hand. (And of course we are configure the system to do exactly the right thing.) So the question I have, as you are just started the project: would you be maybe interested in participating in our node database system and extend it with your ideas while getting everything we already implemented. All code, comments and repository is in English. If somebody else joins development it will be no problem also switching tickets and documentation to English (later will be in any case probably in English). Please go around and check it. If you have any questions feel free to ask them. (I think it could be also on this mailing list as node database design and ideas are I believe very on-topic.) Mitar From (spam-protected) Fri Jan 8 21:22:37 2010 From: (spam-protected) (ZioPRoTo (Saverio Proto)) Date: Fri, 8 Jan 2010 21:22:37 +0100 Subject: [Nodedb-interop] did you know about the meshboard implementation ? Message-ID: Hello, sorry for being too little active in this list. I really would like to help the discussion but I had very little free time after the Ninux Day. If I got it right the goals of this list are two: * Find a proper db-model to describe a generic node of a community network * Identify a management system for mesh network that relies on this db-model ** (eventually merge and optimize already existent implementations) I never tried my self such a management system ... but I want to write about this in my next email ... What I want to point out now is an open source implementation of a node management system for mesh networks: https://dev.art-insite.org/mesh/ It is an open source clone of a proprietary product :http://open-mesh.com/store/ The author was at Ninux Day, he is Italian and his name is Filippo Sallemi . I never tried my self its software but I know Filippo is a good programmer and I know he used this software for quite big mesh deployments he has done for working. I asked Filippo to present his work at Ninux Day but he was afraid his English was not good enought for a oral presentation (of course I think this is not true). Well, I guess you can't miss his voice in this mailing list, so I invite him to subscribe to the nodedb-interop mailing list.... Saverio From (spam-protected) Fri Jan 8 23:02:41 2010 From: (spam-protected) (Ramon Roca) Date: Fri, 08 Jan 2010 23:02:41 +0100 Subject: [Nodedb-interop] did you know about the meshboard implementation ? In-Reply-To: References: Message-ID: <4B47AB81.3040809@guifi.net> I do also apologize for being late... I'm now in Madrid, still having to deal with several legal issues... :(. I'm a bit confused on the focus, maybe we need to have a talk on that, or maybe I'm missing something, but I'm getting the impression that we are talking about user interfaces, apps for database management, even firmware/protocol-oriented apps? I'm much less ambitious, there are already a few of those apps, and might come more, but to me the key point now is to build an architecture to build the capability to interact between apps by describing an XML schema. Mitar, just for reference, if you want to get a report of the Ip ranges used while allocating ip's for each zone, you can get it here: http://guifi.net/es/node/2413/view/ipv4 And the XML export of this looks like that: http://guifi.net/en/guifi/cnml/2413/zones Just by referemce, because this could be incomplet: Although we do export zones in "CNML", and we do manage IP ranges, currently we don't export this attribute, so the schema might have to be enhanced.... I think we should start working on XML/DTD descriptions, and after that, everybody debelop in whatever they do feel more comfortable/creative, however if we miss that step again, the risk is to get just another nice application but not reusable app. I have a dream: A Service Oriented Arquitecture and people developing components for that, able to interoperate because able to speach a common language. so instead of relying on isolated/self developed apps installed on a given server, able to take advantage of the could computing ;) Good 2010 to all!!! Ramon. Al 08/01/10 21:22, En/na ZioPRoTo (Saverio Proto) ha escrit: > Hello, > > sorry for being too little active in this list. I really would like to > help the discussion but I had very little free time after the Ninux > Day. > > If I got it right the goals of this list are two: > > * Find a proper db-model to describe a generic node of a community network > * Identify a management system for mesh network that relies on this db-model > ** (eventually merge and optimize already existent implementations) > > I never tried my self such a management system ... but I want to write > about this in my next email ... > > What I want to point out now is an open source implementation of a > node management system for mesh networks: > > https://dev.art-insite.org/mesh/ > > It is an open source clone of a proprietary product > :http://open-mesh.com/store/ > The author was at Ninux Day, he is Italian and his name is Filippo > Sallemi . I never tried my self its software but > I know Filippo is a good programmer and I know he used this software > for quite big mesh deployments he has done for working. > > I asked Filippo to present his work at Ninux Day but he was afraid his > English was not good enought for a oral presentation (of course I > think this is not true). > > Well, I guess you can't miss his voice in this mailing list, so I > invite him to subscribe to the nodedb-interop mailing list.... > > Saverio > > _______________________________________________ > Nodedb-interop mailing list > Nodedb-interop at lists.funkfeuer.at > http://lists.funkfeuer.at/mailman/listinfo/nodedb-interop > From (spam-protected) Sat Jan 9 17:42:55 2010 From: (spam-protected) (Mitar) Date: Sat, 09 Jan 2010 17:42:55 +0100 Subject: [Nodedb-interop] did you know about the meshboard implementation ? In-Reply-To: <4B47AB81.3040809@guifi.net> References: <4B47AB81.3040809@guifi.net> Message-ID: <4B48B20F.9050605@tnode.com> Hi! > I'm now in Madrid, still having to deal with several legal issues... > :(. Hope they are going well ... > I'm a bit confused on the focus, maybe we need to have a talk on that, > or maybe I'm missing something, but I'm getting the impression that we > are talking about user interfaces, apps for database management, even > firmware/protocol-oriented apps? I think that we are still finding focuses - in plural. For sure one and main is common node database schema. Some other I proposed on the wiki. If they catch OK, if not, also OK. And I think it is OK that we exchange also other ideas (not just schema and attributes there) about our node database systems. To work on collaborative knowledge in this field. > but to me the key point now is to build an architecture to build the > capability to interact between apps by describing an XML schema. I completely agree. This is our primary and most important focus. > Just by referemce, because this could be incomplet: Although we do > export zones in "CNML", and we do manage IP ranges, currently we don't > export this attribute, so the schema might have to be enhanced.... I am not really sure how to read this. A lot of data. :-) So if I understand you are using this IP ranges (not completely): 10.139.0.0/16 172.25.0.0/16 10.228.150.0/24 10.228.129.0/24 10.228.128.0/17 172.16.0.0/14 172.20.0.0/14 172.24.0.0/14 172.28.0.0/14 172.16.0.0/12 10.35.80.0/21 Is this correct? How often do you change this? Would it be enough just to enter this into the wiki? Could you do it? For now maybe we could just enter things into the wiki manually and then later on when we define schema I could make this wiki page dynamic so that it would aggregate data automatically from all networks. (And yes, that list is also just for reference, that is why there is also a contact address so that you can get into the contact for more precise or current information.) > I think we should start working on XML/DTD descriptions, Yes, but we should first try to get everybody who wants to participate in this. No? Maybe we could try to get somebody from Greece? And others? Maybe meshboard could describe also their schema on the wiki? http://interop.wlan-lj.net/ > I have a dream: A Service Oriented Arquitecture and people developing > components for that, able to interoperate because able to speach a > common language. so instead of relying on isolated/self developed apps > installed on a given server, able to take advantage of the could > computing ;) Yes. That is why I also added to the wiki not just schema but also that we should work on data aggregation/exchange. Work on this could simply mean just agree that everybody would use same RESTful way of doing this, for example. Mitar From (spam-protected) Sun Jan 10 10:11:57 2010 From: (spam-protected) (Ramon Roca) Date: Sun, 10 Jan 2010 10:11:57 +0100 Subject: [Nodedb-interop] did you know about the meshboard implementation ? In-Reply-To: <4B48B20F.9050605@tnode.com> References: <4B47AB81.3040809@guifi.net> <4B48B20F.9050605@tnode.com> Message-ID: <4B4999DD.5060403@guifi.net> All the ip provisioning is already automated in our case, avoinding users to get addresses for themselves, that's a jey for growing, if not peopletrend to take big ranges and on ipv4, you are dead. Instead we do provide a small range, and if they grow, of course it expands, therefore is much more efficient. That's not only our experience, I would say that freenetworks.org got collapsed also because of that, all the IP ranges got allocated, regardless if the IPs were in use or not, and therefore, became useless. I think that we do mainly are using a 10.138.0.0/15, but other orgs joined our network later, and had already their own ranges. What we don on the table you have seen is provide the ranges at the zones. The zones have hierarchies (province, county, city, village, neighborhood...), and when a zone starts to require IP addresses, get those addresses form the ranges at the zone level, and if they aren¡t, looks into the zone parents etc... Really, oncve it works automatic, we don't care very much anymore, there is no risk of duplicates, so athought the info is in the database, I don't think that we do have a report of a summary of all the ranges in use, only what you have seen, of course that report can be reworked, but again, and that is another example, would be much better to provide an XML report for that, so later anyone who wants to develop any report in whatever tool of his preference, might simply parse the XML. Ramon. 172.x.x are used for internall links, etc. We still do manage ranges, but there is no need to avoid duplicates.. 10.x.x... are the public addresses Al 09/01/10 17:42, En/na Mitar ha escrit: > Hi! > > >> I'm now in Madrid, still having to deal with several legal issues... >> :(. >> > Hope they are going well ... > > >> I'm a bit confused on the focus, maybe we need to have a talk on that, >> or maybe I'm missing something, but I'm getting the impression that we >> are talking about user interfaces, apps for database management, even >> firmware/protocol-oriented apps? >> > I think that we are still finding focuses - in plural. For sure one and > main is common node database schema. Some other I proposed on the wiki. > If they catch OK, if not, also OK. > > And I think it is OK that we exchange also other ideas (not just schema > and attributes there) about our node database systems. To work on > collaborative knowledge in this field. > > >> but to me the key point now is to build an architecture to build the >> capability to interact between apps by describing an XML schema. >> > I completely agree. This is our primary and most important focus. > > >> Just by referemce, because this could be incomplet: Although we do >> export zones in "CNML", and we do manage IP ranges, currently we don't >> export this attribute, so the schema might have to be enhanced.... >> > I am not really sure how to read this. A lot of data. :-) > > So if I understand you are using this IP ranges (not completely): > > > 10.139.0.0/16 > 172.25.0.0/16 > 10.228.150.0/24 > 10.228.129.0/24 > 10.228.128.0/17 > 172.16.0.0/14 > 172.20.0.0/14 > 172.24.0.0/14 > 172.28.0.0/14 > 172.16.0.0/12 > 10.35.80.0/21 > > Is this correct? How often do you change this? Would it be enough just > to enter this into the wiki? Could you do it? > > For now maybe we could just enter things into the wiki manually and then > later on when we define schema I could make this wiki page dynamic so > that it would aggregate data automatically from all networks. > > (And yes, that list is also just for reference, that is why there is > also a contact address so that you can get into the contact for more > precise or current information.) > > >> I think we should start working on XML/DTD descriptions, >> > Yes, but we should first try to get everybody who wants to participate > in this. No? Maybe we could try to get somebody from Greece? And others? > > Maybe meshboard could describe also their schema on the wiki? > > http://interop.wlan-lj.net/ > > >> I have a dream: A Service Oriented Arquitecture and people developing >> components for that, able to interoperate because able to speach a >> common language. so instead of relying on isolated/self developed apps >> installed on a given server, able to take advantage of the could >> computing ;) >> > Yes. That is why I also added to the wiki not just schema but also that > we should work on data aggregation/exchange. Work on this could simply > mean just agree that everybody would use same RESTful way of doing this, > for example. > > > Mitar > From (spam-protected) Mon Jan 11 10:09:27 2010 From: (spam-protected) (Mitar) Date: Mon, 11 Jan 2010 10:09:27 +0100 Subject: [Nodedb-interop] did you know about the meshboard implementation ? In-Reply-To: <4B4999DD.5060403@guifi.net> References: <4B47AB81.3040809@guifi.net> <4B48B20F.9050605@tnode.com> <4B4999DD.5060403@guifi.net> Message-ID: <4B4AEAC7.6000605@tnode.com> Hi! > All the ip provisioning is already automated in our case, avoinding > users to get addresses for themselves, that's a jey for growing, if not > peopletrend to take big ranges and on ipv4, you are dead. Instead we do > provide a small range, and if they grow, of course it expands, therefore > is much more efficient. I completely agree. And also from maintenance perspective it is much better if this things are automated. Especially in IPv6 would be hard to do this things by hand. > That's not only our experience, I would say that freenetworks.org got > collapsed also because of that, all the IP ranges got allocated, > regardless if the IPs were in use or not, and therefore, became useless. True. But this is the problem of IPv4. I think that we should try to survive IPv4 era and hope for IPv6. Because in reality, 10/8 is simply to small for everybody. OK, we cut put a lot of effort to more precise allocations and could probably pull this thing off (there is space for 16777214 hosts in 10/8), but is it really worth for this few years? > Really, oncve it works automatic, we don't care very much anymore, there > is no risk of duplicates, so athought the info is in the database, I > don't think that we do have a report of a summary of all the ranges in > use, only what you have seen, of course that report can be reworked, but > again, and that is another example, would be much better to provide an > XML report for that, so later anyone who wants to develop any report in > whatever tool of his preference, might simply parse the XML. I agree. It would be very useful also to define schema of IP ranges in use, IP pools, zones, and services available there. Especially services. So, I do not know. Fill the table on the wiki in whatever way you think it is best. We could also make a test case and you (or we together) make an XML proposal for IP ranges used and then for example you would enter XML URL in place of your IP ranges and I would make a reader for it which would fetch it, parse it and fill the lower table accordingly. Mitar From (spam-protected) Tue Jan 12 13:56:03 2010 From: (spam-protected) (Joseph Bonicioli) Date: Tue, 12 Jan 2010 14:56:03 +0200 Subject: [Nodedb-interop] Let's continue In-Reply-To: <4B4739F4.108@tnode.com> References: <4B14212C.60604@tnode.com> <4B14402D.6010506@guifi.net><4B43CAE6.7090509@tnode.com> <4B4739F4.108@tnode.com> Message-ID: <394A9004E6254BBF9DB864187E58D9A2@infolex.gr> Happy New Year to all you guys. I just got back from my vacations and digital isolation in Italy. I also filled some Addressing and DNS info in the wiki. But I must say that here in Greece we have allocated 10.0.0.0/8 to every single prefecture of Greece. Some are not used and a couple of /16s where just left aside as reserved A full address plan is shown here Attica is Athens. Some areas do have VPNs with AWMN OR are interconnected wirelessly. Have a look at this map . Most of the communities here use the WiND Project (Wireless Nodes Database) and you can receive recent stats from all those nets by clicking on their map points. Please fill in your IP Address Plans since it might be interesting to expand this VPN network. Although IPv6 will be the way forward, in some cases we still have a chance to expand with IPv4. We did this recently with wlanlj and it will be interesting to expand this all over Europe. Joseph -----Original Message----- From: nodedb-interop-bounces at lists.funkfeuer.at [mailto:nodedb-interop-bounces at lists.funkfeuer.at] On Behalf Of Mitar Sent: Friday, January 08, 2010 3:58 PM To: nodedb-interop at lists.funkfeuer.at Subject: Re: [Nodedb-interop] Let's continue Hi! I have also added information about DNS to our list: http://interop.wlan-lj.net/wiki/IPAddressing/List Please fill it in. Because I discovered that without DNS there is of little use knowing IPs of some else's network if you want to access content or services there. For example we have access to Freifunk Hannover network because kiberpipa.net (also in Ljubljana) network is peering with them but do not know anything about that network and this peering is of little use - mostly just to show of that we peer! (This is just an example.) So I think we should be really working on letting others know what and how to access things in our networks. For example AWMN (Athens' network) have everything under DNS and with peering with them we can easily find content and services because we can just enter woogle.awmn in our web browser and have a Google like service which indexes many servers they have there. Or VoIP on www.voip.awmn and SIP on sip.awmn. Mitar _______________________________________________ Nodedb-interop mailing list Nodedb-interop at lists.funkfeuer.at http://lists.funkfeuer.at/mailman/listinfo/nodedb-interop -------------- next part -------------- An HTML attachment was scrubbed... URL: