[00:03] hi guys, could someone lead me to some documentation for capacity planning for an ubuntu cloud? [00:06] <\sh> jan247: eventually http://www.ubuntu.com/system/files/UbuntuEnterpriseCloudWP-Architecture-20090820.pdf ? [00:07] <\sh> http://www.google.de/search?sourceid=chrome&ie=UTF-8&q=capacity+plan+for+ubuntu+enterprise+cloud <- first hit on friend google [00:12] how clean are complete dist upgrades? when lucid comes out will it really be as clean and simple as a regular apt-get dist-upgrade? [00:12] chocamo: ideally [00:13] is it better to just backup certain config files and do a clean install with new release? [00:13] how much of your servers software is fromt eh repos [00:14] and what version you coming from [00:15] to get a good idea first hand, start up a vm and give it a shot imo [00:15] main things to watch for is non repo software that you 'installed' [00:15] ya ok [00:15] thanks === robbiew is now known as robbiew_ [01:04] hello, i got a problem with a ubuntu 8.04 lts running as a virtual machine on hyper v [01:05] every day or two the clock gets stuck in a 5 second loop and i have to restart the vm to get it going again [01:05] i have removed ntp ntpd from startup, i have also remove ntp script from cron.daily [01:06] the problem always occurs at 00 minutes (so at 12:00:00, or 13:00:00 and so on) [01:08] i just saw that there is a cron job that runs rdate -s every hour to sync the clock to an external rdate server... could this be causing trouble? [01:17] <\sh> zoran119: where did you see this rdate script, /etc/cron.hourly or /etc/cron.d ? [01:19] A cron job to run rdate is pretttty stupid. [01:19] \sh: the rdate script was in root's crontab.... 'crontab -e' [01:20] \sh: i have removed that cron job now [01:21] zoran119: sounds like whoever set up your VM was a fool. [01:22] twb: why is it such a bit issue? [01:22] zoran119: because cron events are triggered based on time. [01:22] twb: infinite loop possibility? [01:22] So if you have a time event that causes the clock to be set back, it'll loop forever [01:23] Although I guess it *ought* to work if the inetd it's talking to isn't also hokey... [01:23] twb: Doesn't cron have some smarts about triggering the same job multiple times? I believe it does the right thing on daylight savings time shifts. [01:24] erichammond: maybe it does. [01:24] Certainly rdate is deprecated in favour of NTP for other reasons. [01:24] <\sh> zoran119: that's not installed by default, right? actually I don't know any package which installs something into roots crontab [01:24] \sh: not by default no... it was added manualy [01:25] Probably by whoever built the pre-built guest image, or by the equivalent of vmware-tools. [01:25] <\sh> using ntp on our esx vms does work and doesn't crash...it helps to maintain a sane time sync on our vms [01:26] \sh: ntp won't make large steps by default. [01:26] <\sh> twb: iburst? [01:26] \sh: hmm? [01:28] <\sh> twb: you meant with "large steps" that ntp doesn't sync your time at startup of ntp, or did I miss your meaning? [01:28] <\sh> grmpf...I hate unittest [01:30] ? [01:30] lifeless: hiya [01:30] kirkland: hey [01:30] <\sh> adding the keyword "iburst" on your server line in ntp.conf it syncs the time directly after startup (just like ntpdate ; /etc/init.d/ntp start [01:31] !paste [01:31] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [01:33] getting conflict in linux-virtual install http://paste.ubuntu.com/392171/ [01:34] linux-image-virtual refuses to copy in its kernel [01:42] how can I cancel an install with conflicts? [01:42] its frozen at Inst/Unpacked [01:52] New bug: #535439 in samba (main) "Windows Network Computer are not shown" [Undecided,New] https://launchpad.net/bugs/535439 [02:24] so I accidentally toasted /var/lib on my server box. I have a bunch of stuff configured on it that I don't want to lose. I have enough space on my external to copy everything that's not toasted over prior to a reinstall. how should I proceed to minimize pain and effort? it would also help if I could minimize wailing of gnashing of teeth, but at this point I'm flexible. [02:27] invisime: dd the entire partition is safest [02:27] * invisime goes to read man dd. [02:28] so i'm having issues with rtorrent. keeps segfaulting, and it seems to be my config. i _swear_ nothing changed with the config file, but when i run rtorrent with the -n switch, it works fine... [02:30] sorry, it's been a while since i've used irc! [02:30] arrrghhh: strace it [02:31] wow i've never used strace [02:31] seems to dump a ton of info to the termina.l [02:33] is there a particular way i should be running rtorrent with strace? i see the segfault at the end, but there's an endless amount of pretty much garbage proceeding it. [02:35] twb: how do I tell which /dev/ entry is mounted as root? [02:35] i think it may be the xmlrpc calls that is causing rtorrent to segfault... hrm. [02:37] invisime: /proc/mounts, perhaps? [02:38] invisime: if you don't know that much already, it's probably dangerous for you to be trying to do anything at all [02:38] twb: well, clearly. I mean I already accidentally deleted /var/lib :P [02:39] eek === heynow_ is now known as heynow [02:40] arrrghhh: put the strace output in a file, then examine the file [02:40] twb, i'm not sure what this output is... but perhaps it'll make sense in a file. [02:41] arrrghhh: I expect you to at least LOOK at the manpage [02:42] for what, strace? ok. the output does make more sense in a file as well, but i still don't get why it's segfaulting. [02:43] impressive, i'm surprised i haven't heard of strace before. [02:43] It's a log of the system calls made by the rtorrent process during its lifetime. [02:43] Interpreting what is actually happening is a skill you'll have to pick up over time. [02:44] i see "The SCGI socket has not been bound to any address and likely poses a security risk" - i didn't think it would cause a segfault. [02:44] OK, so now we approach the problem from the other direction. What release are you running? Did you enable any third-party repos? Did you install any software by hand (instead of via apt)? [02:45] xml-rpc i did have to compile by hand... [02:45] rtorrent is from the repo's [02:45] Especially if you've done something silly like installing karmic's rtorrent into hardy, that would be a dead giveaway. [02:45] uhm i hope not [02:45] i'm running karmic [02:45] How does xml-rpc fit into this? [02:46] i'm thinking that's what is segfaulting rtorrent [02:46] What evidence do you have of this? [02:46] well, that error. and when i access rtgui it *seems* to crash rtorrent. [02:47] This is obviously something new since I last used rtorrent, when it had an ncurses GUI. [02:47] i had to compile xml-rpc by hand because the advanced tree that fixed a bug i was having rtgui [02:47] oh it still uses ncurses. but it can accept xml-rpc calls so other front-ends can control it. [02:49] What do you mean "because the advanced tree"? [02:49] it's been a while since i went thru it, lemme find the link. essentially any torrent over 4gb would show the incorrect size unless i used the advanced tree of xml-rpc [02:52] well i found the directions, but not the explanation... [02:52] i guess it has a little blurb of why in the intro [02:52] http://code.google.com/p/rtgui/wiki/CompilingRtorrent [02:53] "The original problem is caused by the standard version of XMLRPC-C that is shipped with Ubuntu." [02:54] So you installed xmlrpc-c AND libtorrent AND rtorrent from source? [02:55] i believe i just did xmlrpc-c from source. it has been a while, but as i recall libtorrent and rtorrent came from the repo's. [02:56] So your libtorrent and rtorrent are failing because the version of xmlrpc-c they expect isn't the version that's installed. [02:56] that... would make sense. [02:57] The *right* thing would be to wait for Ubuntu to fix the issue, probably in the next release, since AFAICT it's not a show-stopper. [02:57] The next least-bad thing would be to roll a .deb for your xmlrpc-c svn snapshot, then apt-get build-dep, apt-get source --build and install libtorrent, then rtorrent. [02:58] no, not a show-stopper. but how do i get back to a functioning rtorrent? i'm assuming i have to purge the custom install of xml-rpc and reinstall the one from the repos? [02:58] getting conflict in linux-virtual install http://paste.ubuntu.com/392171/ [02:59] arrrghhh: Ideally by running "make uninstall" in the xmlrpc-c source dir and praying its uninstall code isn't completely fucked, then by running "aptitude reinstall xmprpc-c" or whatever the package name is. [02:59] hrm. [02:59] ok [02:59] ChmEarl: it looks like you simply can't have both kernels installed at once. [03:03] I am trying to setup wireless as a backup interface (don't ask), and I have wpa_supplicant connecting fine, but other problems: static ip doesnt work, tried dhcp but "no dhcpoffers received" [03:04] twb, aaaaaand if the make uninstall fails? [03:05] twb, I tried to uninstall the exiting server image, but the conflict is interfering. How can I cancel that install? [03:05] existing [03:06] i think there is a force option [03:07] -f is force, but the apt-get always comes back and says run "apt-get install -f" [03:07] and I'm trying to do a remove? [03:09] nevermind mates, I got it. I put both of the depends on the line together :) [03:09] woohoo finally [03:10] ChmEarl: dpkg -P [03:10] that does a cancel? [03:10] Oh, right. [03:10] ChmEarl: never mind, you fixed it already. [03:10] a purge? [03:10] Yes, -P is purge. [03:11] I was stuck until you got me trying a diff approach, thanks [03:13] twb- it worked.. uninstalled one, then installed the virtual type [03:14] the initrd was lowered from 7MB->4MB [03:14] If you care about saving 3MB, you shouldn't be running Ubuntu [03:16] ha, seriously [03:16] kirkland, i was wondering if packages that use upstart jobs commands like update-rc.d won't work as always. === RoAk is now known as RoAkSoAx [03:17] clear [03:18] twb, sorry to bug you with this, especially since i caused the problem myself... but the make uninstall didn't work. doesn't seem to exist... is there anything else i can do? === zz_monteith is now known as monteith [03:48] looking in grub.cfg, the root is (hd0,5) is the 5 0-based or 1-based [04:24] twb, so is there anything else i can do? the make uninstall failed. [04:25] arrrghhh: either reinstall, or put up with a messy system [04:25] ChmEarl: depends which version of grub :-/ [04:25] ChmEarl: in GRUB Legacy, everything counts from zero. IIRC in GRUB 2, disks count from zero and slices (partitions) count from 1. [04:26] can i get rid of the xml-rpc stuff at least? i have the tar i compiled it from. [04:26] arrrghhh: shrug [04:26] yea... i knew all this custom stuff would bite me in the end. [04:27] arrrghhh: it's just a phase [04:27] arrrghhh: you'll get over it [04:28] somehow i doubt ubuntu is going to 'fix' the problem with xml-rpc anytime soon. so a clean system will put me back to where i was before, getting sizes that were negative. [04:29] IMO that is a Good Thing [04:29] Hello, I accidentally broke my network somehow in my ubuntu 804 install. I was attempting to setup a PXE server, but when i went to setup dhcpd.conf, there was some firestarter script in there. I deleted the config, and uninstalled firestarter just in case. I started to setup the config file, and got sidetracked, and just deleted all the options. I restarted the dhcpd3 and everything was working fine. When i reboote [04:29] Launchpad bug 804 in eric "Bugged by pyQT api update (dup-of: 803)" [Medium,Fix released] https://launchpad.net/bugs/804 [04:29] t I tried pinging the router, and got an error.. I checked all the regular network settings and disabled any route table, and everything looks fine.. I checked ip tables and noticed that it was set to deny all traffic.. I reset the iptables, and still no luck.. Any ideas on the next step i should take to troubleshoot? i think it has something to do with firestarter uninstall [04:29] Launchpad bug 803 in eric "Bugged by pyQT api update" [Medium,Fix released] https://launchpad.net/bugs/803 [04:32] well then i'd just try to redo the stupid advanced tree of xml-rpc to fix the issue just to have it broken again by an update. [04:36] iptables -P INPUT -j ACCEPT, maybe? [04:36] can't remember the exact syntax [04:37] pastebin 'iptables -L -v' [04:57] jayvee, sorry I didnt know you were talking ... [04:57] I did an iptables -F ; iptables --flush; and iptables-save.. When i reboot it shows the same configuration of deny all is in there. [04:58] not sure what to check for startup scripts [05:02] sekyourbox: pastebin the output of "find /etc/init /etc/event*/ /etc/rc?.d/ -ls". [05:02] sorry no internet on that machine [05:02] lol twb [05:02] sekyourbox: I don't care how you do it. [05:13] okay was that just me? [05:13] or was that a netsplit from hell [05:13] sekyourbox: yup [05:14] sekyourbox: oh, as in just you, you were the only one that left from my PoV [05:14] lol, are you playing with me? [05:14] Netsplit *.net <-> *.split quits: sekyourbox [05:14] lol [05:15] re spawn [05:23] finally got linux-image-virtual kernel running as PV guest in Xen [05:24] guest is Karmic 9.10 server (root=1G) and host is Xen 4.0 on SuSE 11.2 [06:37] New bug: #535533 in qemu-kvm (main) "gdb fails in arm chroot" [Undecided,New] https://launchpad.net/bugs/535533 [07:12] hi. easy question. how to disable ip_v6 on some ethernet port ? say eth0 [07:34] not os easy as I supposed... [07:44] mealstrom, sysctl net.ipv6.conf.eth0.disable_ipv6=1 [07:45] thanks [07:45] but why do you want to disable ipv6? [07:45] there were some unfixed bugs with ipv6 and dhcp or something like this. just want to close external interface [07:47] if I were you, I'd seriously consider deploying IPv6 for real some time. [07:53] 'some unfixed bugs with ipv6' [07:54] mealstrom: Disable ipv4 as well! There are various products on IPv4 that have unfixed bugs! [07:54] (: [07:54] nice idea [07:54] internal network is using ipv6 ) [07:55] Point-to-point serial connections are the only safe path to the future :) [07:56] But more seriously, why isn't IPv6 the default for the virbr0 network in libvirt? [08:08] persia: Because ipv6 is unfortunatly not taken very seriously by some people [08:09] Jeeves_: Is there a technical reason we can't set up libvirt do to both IPv4 and IPv6 by default? [08:09] You'd need some address space [08:10] I'd think you'd need to add a /64 to the libvirt interface [08:10] We're using something in 192.168/16 now for IPv4. Couldn't we use something in fc00::/7 for IPv6 ? [08:11] (or maybe fec0::/10 ) [08:11] persia, because the VMs can't connect to the Internet that way [08:11] IPv6, unlike v4, doesn't have NAT for that sort of thing. [08:12] yes, you can add global addresses as well, but it's a little sticker [08:12] s/sticker/stickier/ [08:12] Ah, so we'd need to install an IPv6->IPv4 gateway by default, and that breaks people with IPv6, etc. [08:12] persia: Uh? What? [08:12] You just need ipv6, not an ipv6->ipv4 gateway [08:13] Jeeves_: I either need IPv6 NAT or a gateway to let the VMs route to the internet unless I have real address space. [08:13] I'm trying my best to get Canonical to offer all packages on ipv6, so you won't need ipv4 at all! :) [08:13] persia: get real address space? :) [08:13] Seriously, get a tunnel somewhere [08:14] Or just native, but that's probably more complicated [08:14] Jeeves_: What packages don't work with IPv6? Surely those can be just patched, rather than waiting for Canonical to get to it. [08:14] Jeeves_: Also, while getting address space isn't that hard, it doesn't solve the "what gets shipped by default" issue. [08:15] * persia tries not to carry local patches or configuration changes, if possible, as this makes install/replication of new stuff easier [08:15] persia: archive.ubuntu.com and security.ubuntu.com don't have AAAA-records [08:16] Oh, so there's no IPv6 mirror. I understand. Is there anything missing in the packages, or is it just a mirror thing? [08:17] things like Postfix don't come configured for IPv6 out of the box [08:17] So following things like this http://www.sixxs.net/wiki/Postfix (which I wrote) becomes necessary. [08:17] jayvee: Is there a known working patch to the default config that would sort that, or do we run into address space issues again? [08:18] Well IMO it's something that should change upstream as well. [08:18] It shouldn't require hard-coding of your IPv6 address space [08:19] I'm in complete agreement with that. Is there a patch that makes sense to make to default config? (Doesn't really matter where it gets applied, once it exists) [08:19] If you don't specify mynetworks, I believe it is automatic, so if the defaults were changed to 'all' instead of 'ipv4', it should "just work". [08:19] Only disadvantage is that no spam DNSBLs support IPv6 yet. [08:19] One did, but it's now defunct. [08:19] Not sure if a patch exists. [08:20] * persia tends to find that ideas get more tractions when patches exist and are promoted [08:20] But the lack of DNSBL for IPv6 makes it awkward :( [08:20] Well DNSBLs aren't configured by default anyway, so there's one way to justify it. [08:20] Good point. [08:21] Also I haven't seen a single piece of spam or abuse on IPv6 yet. [08:21] That's a sign of lack of adoption, really. [08:21] Yeah. [08:21] Also, avahi doesn't come configured with IPv6 turned on by default. [08:22] I could once say that about Usenet :) [08:22] One really unfortunate thing about avahi is that when you do enable IPv6 in /etc/avahi/avahi-daemon.conf, you see duplicate services. [08:22] Is there a local-link address space for IPv6? I didn't think there was an equivalent to 169.254 [08:22] yes [08:22] fe80::/12 or something [08:22] So, for example, in the VNC client, you see two of every server that has IPv4 + IPv6 [08:23] omg ... I've just said how to disable ipv6 on one nic :) [08:23] * jayvee looks it up [08:23] fe80::/10 [08:23] mealstrom: Sure, but you've hit on a discussion that needs happening :) [08:23] persia: http://www.iana.org/assignments/ipv6-address-space/ipv6-address-space.txt [08:23] btw, Ubuntu postfix doesn't come with ipv6-enabled by default [08:24] 2000::/3 (i.e. 2000-3fff) is the "global" address space [08:24] Jeeves_, yeah, I just said === Disconneu is now known as Disconnect [08:24] Right. So postfix is easy: just needs a minor tweak (and some discussion on the mailing list) [08:24] fc00::/7 is closest to the RFC 1918 addresses (e.g. 10.0.0.0/8) [08:24] Well, and 172.16... and 192.168... [08:24] yaeh [08:25] persia: $ host www.debian.org [08:25] avahi needs some way to know when we're talking to a unique machine to avoid duplicated services. Can we do something with arp to help with that? [08:25] www.debian.org has IPv6 address 2001:388:1034:2900::26 [08:25] that's hosted in Australia. :-D [08:25] persia: nothing to do with what you just said, but IPv6 doesn't use ARP [08:25] heh. I wonder who tracks that record most closely :) [08:26] it uses Neighbour Discovery [08:26] you could turn off ipv4 in avahi, but that breaks compatibility with every released ubuntu to date [08:26] not sure how OS X does it [08:26] OS X has had full IPv6 zeroconf support since 2003 [08:26] Probably ignores IPv6 if names are duplicated or something. [08:27] there is also a major flaw in the internals of glibc that prevent mdns from working with link-local (fe80::/10) addresses [08:27] There's an open bug about that? [08:27] yes [08:27] it's been open for years and years [08:27] Anyone working on it? [08:27] doubtful [08:27] it would break too much software [08:28] it's more of a design flaw [08:28] reason being is that link-local addresses also require a "scope ID" [08:28] so I can ping a global address without a scope ID, like so: [08:28] $ ping6 2001:44b8:1::1 [08:28] but to ping a link-local address, I need to specify the ethernet interface [08:28] $ ping6 fe80::21b:fcff:fe25:42e7%eth0 [08:29] now that's obviously structured differently in memory [08:29] but basically glibc doesn't have the provision to return the scope in the dns functions, from what I understand [08:29] actually, I think that's more nss-mdns's problem than avahi [08:30] Sounds it. [08:30] so that doesn't work in an ad-hoc setting [08:30] whereas 169.254 addresses work fine with avahi right here rightn ow [08:31] OK. Reading about NDP, it won't get us information we can reliably compare with something else to determine host identity. [08:31] the one thing that is the same is the hostname, though [08:31] not necessarily, but I agree it's likely safe to assume that. [08:31] nss-mdns also isn't enabled for ipv6 by default [08:31] But I think that belongs in the avahi UIs, rather than anywhere else. [08:32] in /etc/nssswitch.conf, you need to change 'mdns4' to 'mdns' [08:33] in ubuntu 8.04, samba doesn't support ipv6, but in 8.10 and up, it is fully supported and interoperable with Windows Vista, which also supports SMB over IPv6 [08:33] however, LLMNR isn't supported [08:33] which is basically the IPv6 equivalent of nmblookup that microsoft invented [08:33] Right. [08:33] hello, I'm trying to get ubuntu to boot from an older debian kernel - I have already downgraded to grub-1 but I keep getting error 13 when trying to boot the old kernel [08:33] LLMNR is rumoured to be supported by avahi one day [08:33] LLMNR is not ubuntu's problem [08:34] https://bugs.launchpad.net/ubuntu/+source/nss-mdns/+bug/94940 [08:34] Launchpad bug 94940 in avahi "mdns listed in nsswitch.conf causes excessive time for dns lookups" [Undecided,Confirmed] [08:34] I think there ought to be a wiki page listing known issues and referencing open bugs, etc. [08:34] Looking around, I don't think one exists. [08:34] There's https://wiki.ubuntu.com/IPv6Integration but that's an incomplete spec. [08:34] http://lists.freedesktop.org/archives/avahi/2007-February/000959.html [08:35] There's also https://wiki.ubuntu.com/IPv6 but that's user documentation. [08:35] Would you mind creating https://wiki.ubuntu.com/IPv6/OutstandingIssues or similar? [08:35] Jeeves_: Maybe you also have some useful stuff to add there? [08:36] if we get it all in one place, I suspect we'll be well poised to get patches where they need to be to make it work. [08:36] isn't "outstanding issues" and the like what launchpad is for? [08:36] FireCrotch: Yes, it is [08:36] persia: the wiki isn't a bug tracker :) [08:36] FireCrotch: Launchpad does a great job of capturing individual issues, or plans to resolve classes of issues. It does less well at organising them. [08:37] Jeeves_: I know, but it gives us structure, so we can identify which issues block other bits, etc. and build a strategy to solve things. [08:37] Jeeves_: I don't think just adding an "ipv6" tag to bugs gives us that. [08:37] there's already the "IPv6 Task Force" in launchpad which does bugger all [08:37] Indeed [08:38] Maybe that team needs a refresh :) [08:38] (excuse my Australian) [08:38] persia: Feel free to join [08:39] Jeeves_: I need to get a new router first :) My router doesn't support IPv6. [08:39] (and I know almost nothing about the area) [08:39] you can learn a lot even by just running a tunnel on your PC [08:40] apt-get install gw6c [08:40] gets you on the IPv6 internet with a single IP through an anonymous tunnel provided by freenet6 [08:40] works through NATs with no config [08:41] Installed, and I'm sure I'll learn. [08:41] persia: does http://ipv6.google.com/ work? [08:41] I still think it's worth trying to organise the list of issues if we want them solved. [08:42] I think the main issues aren't actually technical — they're political [08:42] jayvee: Now that I've installed gw6c, yes. [08:42] for example, I was disgusted when I was reading the Ubuntu Enterprise Cloud documentation [08:42] I'm fairly certain the main issues are political. That's why I think it needs organisation. [08:42] it was all IPv4 [08:43] Organisation is key to political discussion, and well-organised plans can easily overcome apathy. [08:44] hmm, I guess you have more faith in bureaucracy than I do :) [08:45] It's not that. I just consider bureaucracy a tool. [08:45] Like any tool, one needs to use it the right way. [08:45] Documentating what needs doing, and in what order it needs doing builds a plan. [08:45] Having such a plan makes it easier to create the necessary patches. [08:45] Getting those patches applied is just legwork on the various mailing lists, bug trackers, etc. [08:46] Having a plan, and garnering support at the distribution level helps provide incentive and demonstration of testing to upstreams. [08:46] Having a plan and garnering upstream support helps support applying patches at the distribution level when upstream balks. [08:47] I'm just gonna test whether nginx supports it by default [08:48] actually, I already have apache [08:48] hmm, someone else want to install nginx on karmic or lucid and tell me what "netstat -64ln | grep 80" says? [08:50] jayvee: Do you need a real install, or will the result from a liveCD meet your needs? [08:50] live CD [08:50] hey, don't go to too much trouble [08:54] Argh! kvm is coredumping on every launch for me right now. Sorry. [08:54] (and worse yet, the "report a problem" bit won't complete) [08:54] no worries at all [08:55] I'm wiping the dust off one of my VMs [08:55] When kvm isn't segfaulting, I usually find it trivially easy to instantiate a new VM from a liveCD. [08:56] (and tend to recommend that procedure rather than maintaining VMs for scratch tests) [08:56] A fresh install tends to avoid any yet-unfixed upgrade bugs. [08:57] persia: well here we go: https://wiki.ubuntu.com/IPv6/OutstandingIssues [08:58] Wasn't there a glibc bug? [08:58] Jeeves_: Do you know of anything else outstanding? [09:00] Actually I think that a section should be added with regards to services like archive.ubuntu.com and www.ubuntu.com not having AAAA records. [09:00] <_ruben> i thought postfix posed you a question wether to listen on ipv4 or ipv6 or both .. perhaps only during reconfiguration or so [09:01] _ruben: It does, but only on reconfigure. [09:01] _ruben: The idea would be to change the default. [09:02] <_ruben> persia: ah ok [09:02] There's a bug where ssh x-forwarding doesn't work on a box where ipv6 isn't enabled, but ssh listen on :: [09:02] Template: postfix/protocols [09:03] ttx: on bug 524147, is it true that one CC always controls exactly one Availability Zone? So you can't have an AZ with multiple CCs for redundancy within that AZ? [09:03] Launchpad bug 524147 in eucalyptus "UEC NC failed to fetch preseed.conf from CC using lucid-server-amd64-20100218" [Medium,Confirmed] https://launchpad.net/bugs/524147 [09:03] TeTeT: yes it's true [09:04] TeTeT: you can't have multiple CCs for the same "cluster" [09:04] TeTeT: there is a HA module in the closed-source eucalyptus, not sure what it does though [09:05] Jeeves_: How can ssh listed on :: when IPv6 is disabled? [09:06] Let me rephrase that. [09:06] ipv6 is enabled but there aren't any ipv6 addresses available [09:07] Ah. Right. [09:07] * persia suspects that's a bug in the definition of "ipv6 is enabled" [09:08] ttx: ok, so you should plan your CC and CLC to be redundant when going for a production environment, only the NC can be left brittle [09:09] persia, Jeeves_: there is also the possibility for currently working IPv6 functionality to be broken for IPv4 in the future [09:09] Debian recently announced that bindv6only would be enabled by default [09:09] currently, if you listen on ::, it includes the ::ffff:0.0.0.0 compabitility addresses for IPv4 [09:09] so if your server app is IPv6-enabled, it is IPv4 enabled [09:10] so you only have to support one stack [09:10] persia: There allready is a bug about that [09:10] TeTeT: CLC and Walrus are the one SPOF, CC+SC you could consider losing them and still have a few clusters running elsewhere [09:10] and lots of servers currently depend on that behaviour, and will then lose IPv4 functionality when that option is enabled [09:10] TeTeT: depends on the SLA you want to offer with the private cloud [09:10] reason why the option is changing is because it's not RFC-compliant, as far as I've heard [09:13] unrelated, but CentOS has a bug in its installed. It says "neighbour advertisement" when it should say "router advertisement". and actually I think it should be saying "router discovery" in that particular instance, but I forget. [09:13] s/installed/installer/ [09:14] Jeeves_: I can't find that bug. All I find for ssh ipv6 are bugs #281882 and #407173 [09:14] Launchpad bug 281882 in openssh "ssh hangs in initial handshaking when using IPv6" [Undecided,Confirmed] https://launchpad.net/bugs/281882 [09:14] Launchpad bug 407173 in openssh "openssh: Please set traffic class on IPv6 packets" [Wishlist,Confirmed] https://launchpad.net/bugs/407173 [09:15] persia: Just a sec [09:17] there is a bug that I don't know whether it was reported or not, but if I use an SSH SOCKS proxy, I cannot access IPv6–enabled websites [09:17] whether SSH is running over IPv4 or IPv6 [09:17] all I get is "connection denied" or something — can't remember exactly [09:18] persia: 434799 [09:18] buf #434799 [09:18] bug #434799 [09:18] Launchpad bug 434799 in openssh "X11 forwarding via SSH does not work after upgrade to karmic" [Low,Confirmed] https://launchpad.net/bugs/434799 [09:19] https://bugs.launchpad.net/ubuntu/+bugs?field.tag=ipv6 probably needs more, but I've added that. [09:21] New bug: #535583 in nut (main) "Excessive logging by apcsmart program" [Undecided,New] https://launchpad.net/bugs/535583 [09:49] jayvee: Were you talking abut bug #239701 earlier, or a different one in glibc ? [09:49] Launchpad bug 239701 in glibc "getaddrinfo fails with numerical IPv6 values" [Undecided,New] https://launchpad.net/bugs/239701 [09:58] persia, nope [09:58] Heh. OK. [09:59] I'm referring to the fact that resolving something like "rillian.local" can't return "fe80::21b:fcff:fe25:42e7%eth0". Only things like "2001:44b8:7df3:b970::23". [09:59] I found a bunch more "please enable ipv6" bugs. I'm not sure there is a clear solution for them yet. Added a link to the bugtracker on the page. [09:59] On OS X, it works perfectly. [09:59] ping6 rillian.local works both on a link-local only and a global network. [09:59] on OS X [10:00] jayvee: On osx, you're never sure wether you will use ipv4 or ipv6 [10:00] Right. So we need to reach feature parity :) [10:00] It depends on what answer comes in first [10:00] I'm not sure whether it's an RFC-compliance issue. Maybe OS X is breaking RFCs to provide that. But I'm not sure. [10:00] But OS X has supported that since v10.3, which was released in 2003. [10:00] My v10.3 Panther system supports IPv6 better than Windows 7 and Ubuntu. ;) [10:01] New bug: #325111 in ntp "ntpq output truncates IPv6 addresses" [Unknown,Confirmed] https://launchpad.net/bugs/325111 [10:01] ejabberd doesn't have IPv6 enabled by default [10:01] is it worth talking about universe packages, or only main packages? [10:02] I spose main is more important. ejabberd is in universe. [10:02] It's worth talking about everything. [10:03] The current definition of "main" is only that stuff in main builds against stuff in main. It doesn't currently correspond to translations support, upload restrictions, security support, etc. [10:03] A better way to think about things is "what is available by default takes priority". [10:06] well CUPS only listens on 127.0.0.1:631 [10:06] so http://[::1]:631/ doesn't work — not that that's a problem affected by the IPv4 apocalypse [10:07] localhost should resolve to both ::1 and 127.0.0.1 [10:07] currently it only resolves to 127.0.0.1 [10:07] also, Second Life doesn't work if you have an IPv6 address for a nameserver in /etc/resolv.conf [10:07] but that's offtopic — it's not in Ubuntu [10:09] I do believe bind9 is IPv6-enabled by default. [10:10] I think maybe some sample config in comments for IPv6 wouldn't go wrong in /etc/network/interfaces [10:10]