Chat log from the meeting on 2017-05-23

From OpenSimulator

Jump to: navigation, search

[11:06] Andrew.Hellershanks There were some questions about BulletSim physics last week.
[11:07] Misterblue Waves: like what?
[11:08] Andrew.Hellershanks I'd have to go back through last weeks notes to remember what exactly. IIRC, it was related to crossing regions with a vehicle.
[11:08] Andrew.Hellershanks The vehicles stopped working after a crossing and the person had to stand up and sit back down again.
[11:10] Misterblue Waves: do you have a link to last week's minutes?
[11:10] Arielle Popstar:
[11:10] Arielle Popstar: around 11.55
[11:10] Andrew.Hellershanks ok, ty Arielle.
[11:11] Arielle Popstar: Logger sewell
[11:12] Andrew.Hellershanks He is currently busy but he may be able to get to this meeting soon.
[11:12] Gavin.Hird I saw on the mailing list today Diva said the changes to permissions are OK to use now. Are they safe for production or still in flux?
[11:14] Andrew.Hellershanks Other issues are being worked on at the moment. There is still at least one object related perms issue to be addressed.
[11:14] Gavin.Hird ok
[11:14] Marcus.Llewellyn The hanging on exit for newer monos seems difficult to nail down.
[11:14] Leighton.Marjoram when that is done will that mean 0.9 will be released?
[11:14] Andrew.Hellershanks I thought that had been addressed some time ago but it appears to have popped up again
[11:15] Marcus.Llewellyn It's still happening on anything newer than 4.0.4, I tried today, but it still hang on exit.
[11:15] Andrew.Hellershanks Leighton, I don't know if that is the only issue holding up 0.9
[11:15] Leighton.Marjoram Oh ok
[11:16] Roland.Francis silently nods and
[11:16] Andrew.Hellershanks Many of the mono 4.x series have caused problems. I thought there was one version of 4 that was ok to use but now I'm hearing perhaps not.
[11:16] Misterblue Waves: Diva has been building a test suite for permissions
[11:17] Gavin.Hird I used the brute force method suggested in #7743 and after taking Ubit's changes this morning it works a bit more graceful. On mono 4.8.1
[11:17] Marcus.Llewellyn I've been using 4.0.4 without any issues so far.
[11:17] Andrew.Hellershanks Yes, she has. That will go a long way to verifying if perms are working as expected. It will also allow testing future versions of code to catch any regressions.
[11:17] Misterblue Waves: the main outcome has been thinking through a lot of the permission hairyness and then writing tests that confirm them
[11:17] Gavin.Hird up to 4.4.2 works fine in my experience
[11:17] Gavin.Hird anything above hangs
[11:18] Misterblue Waves: I don't know if she wrote anything up... but there has been a lot of fixes and cleanups of permissions... if anything, they are much more correct in the latest master
[11:18] Andrew.Hellershanks I had seen hangs on other earlier versions of mono.
[11:18] Arielle Popstar: there is a thread in metro forums mentioning probs with mono on new ubuntu and IPv6
[11:18] Arielle Popstar:
[11:19] Misterblue Waves: mono versions have become a real problem.... we need some devs to work on compatibility with the newer monos
[11:19] Andrew.Hellershanks Misterblue, part of the problem with writing tests for perms is knowing what the proper behaviour should be.
[11:20] Gavin.Hird that Microsoft installs mono 5 with their editors....
[11:20] Misterblue Waves: very much so Andrew
[11:20] Misterblue Waves: there were lots of discussions on #opensim-dev
[11:20] Arielle Popstar: striving for s/l compatibility on perms?
[11:21] Misterblue Waves: mostly trying to have SL compatibilitiy while allowing for HG
[11:21] Andrew.Hellershanks Arielle, not 100% on that. SL has some perms issues.
[11:21] Arielle Popstar: HG is just a transport system
[11:22] Arielle Popstar: are perms changing on HG jumps?
[11:22] Gavin.Hird I think it is the effect on the visitor after the transport we are talking about
[11:23] Gavin.Hird perms change to the extent there are differences between OpenSim versions running on grids, including local forks and adaptations
[11:23] Misterblue Waves: Arielle: not changing prermissions just making sure that extensions to the permissions for being in 'non-home' grids are what one expects
[11:24] Arielle Popstar: nod
[11:24] Gavin.Hird Kitely non export is one example
[11:24] Arielle Popstar: that might be beyond control in view of the potential compatibilities
[11:25] Andrew.Hellershanks Gavin, Kitely market has an export control setting that is not the same as the in-viewer export option.
[11:25] Gavin.Hird not everything on Kitely happens on the marketplace
[11:26] Gavin.Hird one example is free items given away in sims that you cannot bring home
[11:26] Andrew.Hellershanks ok. I thought that might have been the option you were referencing
[11:27] Andrew.Hellershanks Gavin, that would suggest the export bit on the item was not set.
[11:27] Gavin.Hird sure, but it is a local adaptation
[11:27] Gavin.Hird meeting HG travelers
[11:28] Arielle Popstar: so might have to relook at how hypergrid is done. Reading some historical aspects of how HG was brought in, it struck me that the current method is a second choice because at the time, the core devs didnt have the ability to look at or modify viewers
[11:29] Arielle Popstar: which isn't the case now
[11:29] Marcus.Llewellyn OpenSimulator can be configured to disallow object transfer between grids. That's not the same as the Kitely Market export flag. They're very seperate things.
[11:29] Andrew.Hellershanks Marcus, that is what I said. :)
[11:29] Marcus.Llewellyn Oops. I should read this with both eyes. lol
[11:30] Ubit Umarov: Diva forgot to mention the worse issue with mono
[11:30] Gavin.Hird The Kitely export flag on Kitely non marketplace items is proprietary
[11:30] Ubit Umarov: that is its DNS
[11:30] Gavin.Hird they have not shared the implementation with anyone
[11:30] Andrew.Hellershanks Marcus, np. You said it in a bit more detail.
[11:31] Ubit Umarov: 2 issues on that depending on mono version
[11:31] Ubit Umarov: A) does not respect TTL, with that breaking dyndns things
[11:31] Andrew.Hellershanks An object could be full perm but that doesn't automatically mean the object creator wanted to allow people to carry the item to other grids.
[11:32] Ubit Umarov: B) seems there is a issue with invalid reverse.. that kills most real life IPV4
[11:32] Andrew.Hellershanks I would expect the default to be no export allowed on objects.
[11:32] Arielle Popstar: not good for legacy items Andrews
[11:33] Marcus.Llewellyn Full perm only confers the rights explicitly stated. Grid transfer is not one of those rights. To respect copyright. you have to consider all other rights reserved.
[11:33] Andrew.Hellershanks Arielle, if it is a legacy item it could be difficult to contact the original creator to determine their wishes about an object.
[11:33] Arielle Popstar: copyright has no clue about different "grids"
[11:34] Gavin.Hird @Markus, an avatar , who is the legal entity behind ownership of an item has the right to use that item everywhere it is technically possible
[11:34] Arielle Popstar: that is manufactured thing because of Opensim
[11:34] Andrew.Hellershanks Ubit, is the reverse DNS issue a problem with looking up the IP of the main grid server or the IP address of a server running a region?
[11:35] Gavin.Hird connected grids and virtual identities are not legal entities
[11:35] Gavin.Hird only the person behind is
[11:35] Gavin.Hird behind the avatar
[11:35] Marcus.Llewellyn Gavin, in theory yes. In the US, things like the First Party Doctrine ought to apply, But so far case law regarding digital goods doesn't lean that way.
[11:35] Gavin.Hird Half of the grids are in non US locations
[11:36] Marcus.Llewellyn And there is a big difference between an owner's rights and a creators rights. Copyright is about the creators rights, and those often trump owner rights.
[11:36] Andrew.Hellershanks I wish this post beside me wasn't in the way. Makes it hard for me to see everyone properly. Can't remove it without the roof coming down on hour heads. :)
[11:36] Gavin.Hird You said that Andrew
[11:36] Gavin.Hird move the cam
[11:37] Arielle Popstar: anyway what about this mono thing?
[11:37] Andrew.Hellershanks Gavin, I might be able to make that work if I look at people from behind.
[11:37] Marcus.Llewellyn Ubit has been eyeballing the mono thing, and misterblue has at least dipped a toe in, from the looks of the mantis report.
[11:37] Andrew.Hellershanks But there are other posts that will get in the way
[11:38] Andrew.Hellershanks If it is a mono issue a proper fix may be on hold until it gets fixed by the mono developers.
[11:38] Arielle Popstar: so is relevant for those running linux but not windows
[11:38] Andrew.Hellershanks Arielle, yes
[11:38] Marcus.Llewellyn Getting upstream mono to fix things on our behalf has always been easy. >:)
[11:38] Gavin.Hird Linux and macOS
[11:39] Arielle Popstar: most home users on windows probably?
[11:39] Gavin.Hird I doubt it is a pure mono issue given they have released at leat 3 version since it was first detected
[11:39] Misterblue Waves: my fear is that Mono is being 'improved' and we cannot expect perfect backward compatibility
[11:39] Andrew.Hellershanks Arielle, that would be a safe bet but I haven't seen any stats to back it up.
[11:39] Misterblue Waves: in that case, OS will have to change to adapt to the new 'improved' version
[11:39] Gavin.Hird There is a LOT of Microsoft code flowing into mono these days
[11:40] Gavin.Hird so there are changes
[11:40] Andrew.Hellershanks Misterblue, the mono devs may also still be in the middle of making use of the code released by MS.
[11:40] Ubit Umarov: well im really "blind" about those mono issues.. for the moment i can't test with mono
[11:40] Marcus.Llewellyn It is getting harder to build old version of mono. I couldn't build a 3.x version on Ubuntu 17.04 for example.
[11:40] Arielle Popstar: Mels view was apparently is that home users will be history
[11:40] Cinder.Biscuits well that's why targetting .NET core would probably a good thing to work towards.
[11:40] Arielle Popstar: according to one poster
[11:40] Gavin.Hird agreed Cinder
[11:40] Misterblue Waves: I bet they don't have unit tests to make sure everything that worked just-so before working exactly the same after code merges
[11:41] Ubit Umarov: cinder, many opensim users run old machines.. that does limit us
[11:41] Ubit Umarov: we are "stuck" on XP suport
[11:41] Marcus.Llewellyn core is even more of a moving target than mono is. Don't get me wrong, I love core, but porting to it would not be trivial... it makes more than a few breaking changes, and we'd have to be certain of full support for the assemblies we use.
[11:42] Andrew Hellershanks: Code breaking between versions is a common issue. I have seen working programs on my system break after updates.
[11:42] Andrew Hellershanks: Most likely due to lack of test suites
[11:42] Ubit Umarov: (and well , we have tons of code that is actually .net 0,1 ;)
[11:42] Misterblue Waves: OS will have to move on someday
[11:42] Arielle Popstar: move on how?
[11:43] Ubit Umarov: yeap we do move... just can't rush to top
[11:43] Misterblue Waves: I'm not totally sure about Mel's 'home is dead' view.... there are some OS functions that rely on reverse lookup but I don't know if they are cast in stone
[11:43] Ubit Umarov: ( and rushing to top is rushing into tons of new bugs hard to spot)
[11:43] Misterblue Waves: but I haven't looked deeply into the problem
[11:44] Marcus.Llewellyn That we still base test builds on mono 2.10.8 is kinda weird to me. Surely most reasonable people have upgraded either mono or their distro to something that runs a version of mono that's been released much later than 2011.
[11:44] Cinder.Biscuits well i think that's the version centos still ships with.
[11:44] Cinder.Biscuits but it's less than ideal to actually run opensim on that.
[11:45] Gavin.Hird centos or 2.10.8?
[11:45] Arielle Popstar: any major grids use it?
[11:45] Cinder.Biscuits 2.10.8
[11:45] Gavin.Hird it is slow 2.x compared to version 3
[11:46] Ubit Umarov: we still do compile checks on 2.xx
[11:46] Ubit Umarov: that we do need to change :)
[11:46] Cinder.Biscuits gc in mono 2.x is rotten.
[11:46] Marcus.Llewellyn And that's saying it nicely. lol
[11:46] Ubit Umarov: guess no one uses that now
[11:46] Arielle Popstar: it was Justin in past who kept the mono versions from being upgraded
[11:47] Ubit Umarov: well i will also slow that down.. just not that down ;)
[11:47] Misterblue Waves: there was some method to that madness :)
[11:47] Marcus.Llewellyn Well, to be fair, it wasn't that long ago that distro were very poor about updating mono packages. That's changes in the last 4 years or so.
[11:47] Gavin.Hird How about stating minimum requirement for 0.9 is some suitable version of 3?
[11:48] Arielle Popstar: seem to remember Justin mentioning it was debian holding it up back then
[11:48] Andrew Hellershanks: I have access to some machines running CentOS but I can't tell which version of mono is available to them as packages.
[11:48] Ubit Umarov: we need to fix issues with mono 4x... ( the not broken versions of it )
[11:48] Misterblue Waves: is it acceptable to specify a Mono version that is required?
[11:48] Marcus.Llewellyn Debian (which i use) was a major bad actor in mono upgrades, yes.
[11:48] Ubit Umarov: we do need better working GC etc
[11:49] Marcus.Llewellyn They've since improved, and xamarin has it's own mono repos for debian as well.
[11:49] Misterblue Waves: most Linux distributions have Mono 4.x
[11:49] Marcus.Llewellyn Yup
[11:49] Ubit Umarov: boehm doesn't seem to be holding well with the caching etc
[11:50] Andrew Hellershanks: I haven't had a chance to do some code updates to see if the recent changes to OS re: which GC to makes any different to the amount of memory is used by Robust.
[11:50] Marcus.Llewellyn Boehm is still there as an option, but hasn't been the default in ages. Mono 5.0 has an even newer GC. I forget what it is.
[11:51] Ubit Umarov: yes marcus.Llewellyn but those DNS issues prevent osg for example to use mono 4x on robust
[11:51] Ubit Umarov: we do need to find them
[11:52] Gavin.Hird so what should you run Robust on Ubit?
[11:52] Gavin.Hird it runs up to 4.4.2
[11:52] Ubit Umarov: osg is stuck on 3.2.8
[11:52] Sheera.Khan Centos 7.3 has mono-core 4.2.4
[11:52] Sheera.Khan Verfügbare Pakete
Name  : mono-core
Architektur : x86_64
Version  : 4.2.4
Ausgabe  : 7.el7
Größe : 15 M
Quelle  : epel/x86_64
[11:52] Marcus.Llewellyn Curious. Why is that a problem for OSgrid?
[11:53] Ubit Umarov: with 4x regions using dyndns fail
[11:53] Gavin.Hird ah, IC
[11:53] Ubit Umarov: but im talking from reports i get.. can't really test
[11:53] Marcus.Llewellyn 4x regions?
[11:53] Ubit Umarov: ( mono 4x )
[11:53] Marcus.Llewellyn Ah
[11:54] Marcus.Llewellyn I'm running a grid with robust under mono 4.0.4. Seems to run just fine, and I use dyndns.
[11:54] Andrew Hellershanks: Marcus, which version of OS are you using?
[11:54] Ubit Umarov: well Dan and others do report issues
[11:55] Marcus.Llewellyn Oh lord... dunno. lol. Some oldish version of master at the moment.
[11:55] Ubit Umarov: an initial one was the TTL.. mono caching dns entries for ever
[11:55] Cinder.Biscuits that can be changed.
[11:55] Gavin.Hird so the open connect regions with dyndns is an issue when on OSGrid?
[11:55] Andrew Hellershanks: I've got memory issues with Robust using a master from December with mono 3.2.8 and 4.3.2
[11:55] Ubit Umarov: that showed on early 4.X i was told.. hope fixed on more recent ones
[11:56] Cinder.Biscuits i've got a patch somewhere for that even. it's opensim caching those not mono.
[11:56] Gavin.Hird if the run higher than 3.2.8?
[11:56] Ubit Umarov: don't think we do much dns caching
[11:57] Ubit Umarov: hmm if any.. hmm
[11:58] Ubit Umarov: the reverse dns issue mb ours.. i'm just failing to find where :(
[11:58] Misterblue Waves: are the memory issues only with Robust use?
[11:58] Cinder.Biscuits what's the reverse dns issue? is there a mantis?
[11:58] Andrew Hellershanks: Misterblue, yes. I haven't noticed the problem with the regions.
[11:58] Ubit Umarov: yes
[11:58] Ubit Umarov: ( looking )
[11:59] Arielle Popstar: does it have to do with this thread Ubit?
[11:59] Ubit Umarov: ( )
[12:00] Andrew Hellershanks: 7908?
[12:00] Andrew Hellershanks: hehe... Ubit and I found it about the same time.
[12:01] Andrew Hellershanks: Hm... I don't think Logger is going to make it to todays meeting as we will soon be wrapping up for today.
[12:01] SkyFlier.NorthStar I have a simple question about the latest master versions
[12:01] Sheera.Khan @ Arielle: In that case the IPv6 address got entered into the grid database and the clients couldn't connect to it since it wasn't even bound to opensim
[12:01] Arielle Popstar: ahh ok. Thx Sheera
[12:02] SkyFlier.NorthStar these versions seem to have fewer items, dlls
[12:02] Gavin.Hird in bin you mean?
[12:02] SkyFlier.NorthStar yes
[12:02] SkyFlier.NorthStar i have an old master from october 2016 which works fine
[12:03] Sheera.Khan so what I don't quite understand is why an IPv6 address gets entered into the grid database even though opensim isn't even using it
[12:03] Gavin.Hird pick up the release candidate zip and they missing should be there
[12:03] SkyFlier.NorthStar and wanted to upgrade
[12:03] Arielle Popstar: Sheera that was why i wondered if the issues were connected
[12:04] Andrew Hellershanks: Sky, a lot has changed in master since 2016
[12:04] Marcus.Llewellyn snorts.
[12:04] Andrew Hellershanks: :)
[12:04] Gavin.Hird Andrew if you just download a zipped version of the latest master, it does not have all the binaries needed to run
[12:04] SkyFlier.NorthStar fewer dlls are necessary Andrew?
[12:05] Gavin.Hird so you have to get them from elsewhere
[12:05] Gavin.Hird the RC zip has them
[12:05] Andrew Hellershanks: SkyFlier, I have never counted the DLL files.
[12:05] Ubit Umarov: ( there are also mantis for the ttl issues but older and im not finding them ;) )
[12:05] Gavin.Hird the OSGrid zip probably does too
[12:05] Andrew Hellershanks: Gavin, what binaries are missing?
[12:05] Gavin.Hird a bunch
[12:05] SkyFlier.NorthStar so these older dlls are no longer necessary?
[12:05] Ubit Umarov: hmm what is missing other then VS 2008 runtime for ode ?
[12:06] Gavin.Hird try just download the zipped version of master from git
[12:06] Gavin.Hird it will not run
[12:06] Ubit Umarov: what is missing gavin?
[12:06] Andrew Hellershanks: There may be items missing from the package but nothing required to run OS. There may be some things that a typical grid would like to have but that may not be included by default.
[12:07] Gavin.Hird I cannot remember at the top of my head
[12:07] Ubit Umarov: i use git clone and see no issues :(
[12:07] SkyFlier.NorthStar when you get to release version you expect to have pretty much whats in the master now?
[12:08] Andrew Hellershanks: SkyFlier, yes.
[12:08] SkyFlier.NorthStar i guess i was an hour late sorry
[12:08] Ubit Umarov: hmm RC2 is old now
[12:08] Gavin.Hird download the RC and download master and do a diff on the two bin
[12:08] SkyFlier.NorthStar yes it is
[12:08] Arielle Popstar: meeting usually goes 2 hours but some leave after the first
[12:09] Andrew Hellershanks: SkyFlier, I count 124 DLL files but that would include about 3 that are unique to my setup and not part of a stock version of OS.
[12:09] SkyFlier.NorthStar if you say RC2 is old then i dont need it right?
[12:09] Arielle Popstar: .9 RC2? Thats only a couple months i thought
[12:10] Gavin.Hird the missing binaries are things that don't change often if at all
[12:10] Andrew Hellershanks: Gavin, check on what you think is missing and get back to us next meeting.
[12:10] SkyFlier.NorthStar that would be great
[12:10] Gavin.Hird Yeah I'll diff the two and see
[12:11] Andrew Hellershanks: Any other topics for today before we start more people need to leave?
[12:11] SkyFlier.NorthStar in the mean time i will see how the master works for me
[12:12] Andrew Hellershanks: SkyFlier, you will need to check the ini files and update your configuration. There has been a number of changes to the ini files.
[12:12] SkyFlier.NorthStar thank you Andre
[12:12] SkyFlier.NorthStar yes
[12:12] Andrew Hellershanks: yw
[12:12] SkyFlier.NorthStar i always check the new ini files
[12:13] Andrew Hellershanks: SkyFlier, the OSSL enables are kept in a separate file. That might have changed from the master you are using from 2016
[12:13] Arielle Popstar: and ossl.ini
[12:13] SkyFlier.NorthStar is visual studio 2010 still good enough to build the code?
[12:13] Arielle Popstar: should have all the ini's in one folder instead of spread all over :)
[12:13] SkyFlier.NorthStar some said should use vs 2017
[12:13] Andrew Hellershanks: Does anyone here know that who builds under Windows?
[12:13] Cinder.Biscuits the details on that mantis are a little slim. i'm assuming opensim is calling DNS.GetHostEntry()
[12:14] Ubit Umarov: ( and ppl plz report issues also mantis so we can see them )
[12:14] SkyFlier.NorthStar ok Ubit
[12:14] Ubit Umarov: yes it is cinder :(
[12:14] Andrew Hellershanks: Arielle, that makes things easier. I have been keeping my changes in a single separate file now instead of customizing the main files.
[12:15] Ubit Umarov: and we only call DNS.GetHostEntry on local things
[12:15] Ubit Umarov: some i actually removed like that non op NAT code
[12:15] Ubit Umarov: well it was doing nothing anyway...
[12:15] SkyFlier.NorthStar ty Ubit
[12:16] Ubit Umarov: Adam disabled it while he was doing it :)
[12:16] Arielle Popstar: Cinder, any idea why the viewers on some routers pick up the internal router IP instead of the external?
[12:16] Arielle Popstar: causing a mismatch on HG jumps
[12:16] Gavin.Hird but Melanie submitted a commit not very long ago with a setting for how the internal vs external IP should be handled
[12:16] Ubit Umarov: Ari o told you that already
[12:16] Gavin.Hird I never understood how it worked
[12:17] Ubit Umarov: is how some routers do lookback
[12:17] Ubit Umarov: not viewers
[12:17] Arielle Popstar: mhmm but if viewer can be configured differently in those cases....
[12:17] Cinder.Biscuits because that's what the interface is telling them it is. it's not a viewer thing.
[12:17] Ubit Umarov: some routers loopback using internal IP
[12:17] Ubit Umarov: not external one
[12:17] Cinder.Biscuits damn, and it looks like GetHostByAddress is deprecated too. :(
[12:18] Ubit Umarov: yeap seen that also
[12:18] Ubit Umarov: well i still fail to find where we to look for reverse :(
[12:18] Ubit Umarov: but im kinda blind ;)
[12:19] Cinder.Biscuits mono is handling GetHostEntry() wrong. instead of doing what it should be doing, it's discarding any dns result and returning the interface ipv4 address.
[12:19] Cinder.Biscuits looks like they broke it trying to fix ipv6 resolution. :O
[12:20] Gavin.Hird Make resolving the IP on startup optional. Adds ResolveAddress boolean to regions.ini
[12:20] Cinder.Biscuits i'm trying to think of some alternative method that could be used.
[12:20] Ubit Umarov: IPHostEntry rDNS = Dns.GetHostEntry(end);
[12:20] Gavin.Hird Commit: 9ef9dde0f3ea9185bb9aca03d173e5312e1ba308 [9ef9dde
[12:21] Ubit Umarov: we only use it directly on ban checks
[12:21] Ubit Umarov: but that code was not working at all since 2 days ago
[12:21] Cinder.Biscuits the irony here is if ipv6 worked on mono, we wouldn't need any of this NAT crap, :^)
[12:21] Ubit Umarov: ) but mono does use it internally i fear )
[12:22] Arielle Popstar: suppose it would be a challenge to have opensim use ipv6?
[12:22] Sheera.Khan but the IPv6 address isn't stable either...
[12:22] Cinder.Biscuits impossible with the state of support in mono.
[12:22] Ubit Umarov: i would hate to bypass mono on dns
[12:22] Marcus.Llewellyn You've have to drag the viewers along to IPV6, which would also mean a major change to their protocol stack.
[12:22] Sheera.Khan due to the privacy extensions the IPv6 changes regularly - even more often than dyn IPv4
[12:23] Cinder.Biscuits not really. most of the resolution is done in curl, ipv6 in the viewer would almost be free.
[12:23] Cinder.Biscuits you wouldn't need a static ipv6 address, that's what AAAA records are for.
[12:24] Sheera.Khan well then we are back at square one with the reverse DNS problem ^^
[12:24] Cinder.Biscuits no because reverse dns isn't an issue in ipv6.
[12:24] Cinder.Biscuits because there's no need for NAT transversal.
[12:25] Arielle Popstar: every appliance has its own external address?
[12:25] Sheera.Khan DynDNS I meant, sorry
[12:25] Cinder.Biscuits in my experience, every iface gets about 3 addresses.
[12:25] Cinder.Biscuits time-warner gave me 500 ipv6 addresses to use.
[12:26] Sheera.Khan normally you get at least a /64 net
[12:26] Cinder.Biscuits yeah
[12:26] Andrew Hellershanks: one IP per IF. The only time you see "multiples" is if it uses something like wlan.0, wlan.1
[12:26] Ubit Umarov: well we dont have all that much info on the reverse dns issue.. just that its there ;)
[12:27] Ubit Umarov: i can't make Dan break osgrid to test it ;(
[12:27] Andrew Hellershanks: :)
[12:27] Arielle Popstar flüstert: aww :
[12:27] Marcus.Llewellyn It's been broken before. I say make him. ;)
[12:28] Sheera.Khan couldn't the robust server simply use the IP-address the region uses to connect and put it into the database?
[12:28] Andrew Hellershanks: I can't test reverse DNS issues as the machines I've been using to run OS all have proper reverse DNS
[12:28] Sheera.Khan the grid server should see the correct IP, shouldn't it?
[12:28] Ubit Umarov: it sees the direct one
[12:29] Ubit Umarov: thing is that many ISPs don't have correct reverse
[12:29] Andrew Hellershanks: Sheera, it is why I asked Ubit whether the reverse DNS issue is at the upper grid level, or affects servers running regions and their reverse mappings
[12:29] Andrew Hellershanks: He didn't answer so may not have seen the question
[12:30] Ubit Umarov: they buy IP ranges from others, set the FW and ignore the reverse
[12:30] Ubit Umarov: or dynamic IP.. they don't care about reverse.
[12:31] Ubit Umarov: so real world IPv4 reverse cant be used
[12:31] Ubit Umarov: except on mail machines and similar where you want a proper peer
[12:31] Sheera.Khan I have a question there: is there any kind of regular traffic between the region server and robust with the uuid of the region in it?
[12:32] Andrew Hellershanks: Sheera, I think so. How regular.
[12:32] Sheera.Khan then the robust could use those packets to determine the current IP without any DNS at all
[12:32] Andrew Hellershanks: ?
[12:32] Andrew Hellershanks: real world IPv4 reverse can't be used??
[12:32] Ubit Umarov: i said why andrew
[12:33] Andrew Hellershanks: why what?
[12:33] Ubit Umarov: why it can't be used
[12:34] Andrew Hellershanks: oh, you were talking about dyndns
[12:34] Ubit Umarov: more than that
[12:34] Gavin.Hird if you run your own DNS can't you create a reverse record that is fits the bill?
[12:34] Andrew Hellershanks: sure.
[12:34] Ubit Umarov: most ISPs smaller ones specially don't care about reverse tables
[12:34] Andrew Hellershanks: or if you have your own domain name then you have access to the DNS entries for the domain and can set reverse mappings.
[12:35] Sheera.Khan Metro Grid uses an extension which gets rid of the whole DynDNS problem...
[12:35] Ubit Umarov: mine does not care
[12:35] Ubit Umarov: my reverse does not point to my fw
[12:36] Ubit Umarov: well today my fw points to someone else its broken :)
[12:36] Ubit Umarov: AHHH its fixrf Yiiipiii
[12:37] Andrew Hellershanks: Anything else for today?
[12:38] Andrew Hellershanks: Let's leave it at that for today.

Personal tools
About This Wiki