/srv/irclogs.ubuntu.com/2010/11/16/#ubuntu-cloud.txt

smoserkiall, it doesn't do that in default installations00:46
smoseri believe that you can have it be the dns server, but that is not default install, and without it, it cannot answer dns requests00:47
kiallno problem .. one other question - should instances be able to communicate with each other via their public IPs then?00:47
kiall(I'm seeing some strange NAT "stuff" happening at the mo ;))00:47
kiallmeaning there is no way for 2 instances to talk without knowing their private IPs (not exactly ideal...)00:48
smoserboth public and private should work for instance-to-instance traffic00:51
kiallthats what I thought .. but until 30 seconds ago, that simply didnt work .. and now is looking a bit wrong (aka "ping public-ip" .. "reaply from private ip", and an instance cant ping its own public IP).. The change that got me that far was removing the bridge on the CC priv net, and switching to a normal eth interface(00:53
MakereI'm having trouble running instances06:52
Makerethey get stuck on pending06:52
Makerecan't get the bundled images to run at all09:24
Makerelet's put this otherway, has ANYONE managed to ANY images downloaded from the website and HOW?09:45
MakereI'm losing all hope09:53
kiallMakere, i scrapped the "store" images .. they simply didnt work!10:51
kiallDownloaded the 10.10 uec image from uec-images.ubuntu.com and used the uec-publish-tarball command to push it up..10:52
kiallworked like a charm10:52
kiall(oh and - I had to increase the m1.small settings to get it to work with 64 bit images.. they dont work out of the box ... go figure  )10:52
Makerekiall: well the uec-images on didn't work for me12:28
Makeregot stuck on pending :(12:29
=== xfaf is now known as zul
Makereno errors on any log, nc, nc14:18
Makerecc14:19
Makerekiall, you runnin lucid or maverick?14:19
kiallmaverick14:19
Makerecontrollers on seperate pc's?14:20
Makereone or more clusters?14:21
Makerejust [Brunning thevm's without ramdisk?14:21
kialla combined CLC/CC/SC/Walrus and a few nodes... (so 1 cluster)14:21
kialland no ramdisk for the 10.10 uec-images EMI14:22
MakereI'm trying 2 clusters with 1 cloud controller and walrus and 2 cluster controllers/sotrage controllers14:23
kiallHavent done a split setup yet .. but I hope your starting simple and going with 1 cluster and 1 NC first ;)14:24
Makereso I guess the problem has to do something with splitting them :p14:24
kiall...and going with 1 clc, 1 cc, 1 nc*14:24
=== dendrobates is now known as dendro-afk
kiallanyway .. I thing I found was, once I had tried a load of different things .. stop. wipe. reinstall with your new knowledge ;)14:27
kiall1 thing i found*14:27
kiallAlso .. I found the package install "buggy" and ended up using the CD install (hell - it even asked more UEC specific setup questions .. go figure)14:27
=== dendro-afk is now known as dendrobates
Makereyea14:44
MakereI did CD install14:44
Makereproblem is that we've done it multiple times now14:45
Makereand this is not the first time we get the problem14:45
Makerekiall: have you updated the packages?14:45
Makereon the nodes or the "Controller"14:45
kiallMakere, yea - apt-get update && apt-get dist-upgrade && reboot after the CD install .. then setup bridges etc, edit config, clean restart the euca services, and reboot .. - thats about all i needed to get a VM booted (there was more .. but VMs booted)15:02
Makeresetup bridges?15:10
Makere:O15:10
Makereno mention of that in the installation guide15:11
kialllol ... im sure it mentions it loads of times ;)15:11
Makerehttps://help.ubuntu.com/community/UEC/CDInstall atleast this one doesn't15:11
kiallhttp://www.google.com/url?sa=t&source=web&cd=1&ved=0CBcQFjAA&url=http%3A%2F%2Fcssoss.wordpress.com%2F2010%2F05%2F10%2Feucalyptus-beginner%25E2%2580%2599s-guide-%25E2%2580%2593-uec-edition-chapter-7-%25E2%2580%2593-network%25C2%25A0management%2F&rct=j&q=eucalyptus%20beginner%27s%20guide%20networking&ei=Np_iTKb0MIHtsgbz3snkCw&usg=AFQjCNGOWI5oW4sBUfT8-AcILsaCl8_hSQ&sig2=vbyB86w94lQLeUdxEo-9tw&cad=rja <-- was a good howto on the network setup that15:12
kialldoesnt leave anything out15:12
kiallwow... that URL wasnt exactly the one I was going for ;)15:12
kiallbut it should work ;P15:12
kiallLOL .. it actually doesnt mention bridges ..≥15:13
Makereyea15:13
Makere:)15:13
kiallyea .. just go to chapter one of that link (and do it the CD install way instead from my exp) .. ;)15:13
Makereso annoying >_>15:14
Makerethat guide looks like million times better15:15
Makerethanks15:15
MakereI wonder if I should start from the beginning15:16
Makerezzz takes 2 hours to setup with 2 clusters15:16
MakereI guess usb installation would be faster15:16
Makereoh wow15:17
Makerethat ubuntu community guide is so shit compared to this15:17
kiallcan anyone with a UEC instance running check if they can ping that instances public IP from itself?16:05
kiall(the euca community cloud cant, neither can my setup .. but I believe it should be able to)16:05
smoserkiall, i'll check16:11
kiallcool :) Lets see If i should quit my job or not ;)16:12
Makerelol16:20
smoserwell, ping of 'public-ipv4' fails16:20
kiallright so .. at best thats an inconsistency with EC2, and worst a bug :)16:23
kiallI'll dig into iptables rules and see whats missing/wrong and file a bug16:23
=== You're now known as ubuntulog
smoseri tested in 10.04 host, not 10.10 it could have changed.16:32
smoserbut it is a bug if its an inconsistency with ec216:32
kiallI've tested with 10.10, but I can't guarantee my setup is correct16:33
* kiall can *never* remember what hairpin NAT iptables rules should look like .. google time16:34
kiallfunny - it looks like the iptables rules are in place for hairpin NAT ...16:37
smoserkiall, i did you a favor, can you do one for me ?16:38
kiallsure16:38
smoserec2metadata --local-ipv416:38
smoserin an instance16:38
kialloo unavailable16:38
smoseryeah.16:38
smosereucalyptus metadata service doesn't like a trailing slash16:39
smoserhttp://169.254.169.254/2008-02-01/meta-data/local-ipv4/16:39
smoseris what ec2metadata asks for16:39
kiallYea - without the / it works spot on16:39
smoserhttps://bugs.launchpad.net/ubuntu/+source/cloud-utils/+bug/67614416:41
uvirtbotLaunchpad bug 676144 in cloud-utils "ec2metadata should not add trailing / to requests" [Undecided,New]16:41
=== You're now known as ubuntulog_
=== You're now known as ubuntulog
jmgallowayanyone here good with ssh?16:51
jmgallowayI am getting an error when trying to connect to my walrus and cluster controller machines..16:51
kiallsmoser, got the missing rule (evenually) ;)17:22
smoseroh. great. open bug and put it in there.17:22
kiall"-t nat -A POSTROUTING -s $VNET_PRIV_NET/16 -d $INSTANCE_PRIV_IP -j MASQUERADE" seems to do the trick17:22
kiallinfact .. I wonder if one of the existing rules can be changed (now that I look at it and it works...)17:23
kiallYup - one of the existing MASQ rules has a ! $PRIV_NET which is the real issue it seems...17:26
kiallsmoser, worlds worst worded bug report ;)17:37
kiallhttps://bugs.launchpad.net/eucalyptus/+bug/67616717:37
uvirtbotLaunchpad bug 676167 in eucalyptus "Hairpin NAT on CC disabled - EC2 inconsistency" [Undecided,New]17:37
smoserkiall, fyi, if you wanted to a.) be nice and b.) push your fix in17:48
kialltrying to figure out this hole bzr thing ;)17:49
kiallwhole*17:49
smoseryou could branch17:49
smoserbzr branch lp:ubuntu/eucalyptus/natty17:49
smoserwait17:49
smoserbzr branch lp:ubuntu/natty/eucalyptus17:49
smosercd eucalyptus17:50
smoserquilt new kiall-fix-hairpin17:50
smoserquilt edit path/to/new/file17:50
smoserquilt refresh17:50
smoserdch -i17:50
smoser# edit the new message17:50
smoserbzr commit17:50
kiallaha .. git is way easier ;)17:50
smoserwell, thats not really git versus bzr17:51
smosermore bzr packaging (using quilt)17:51
smoserbut i will argue no further against git (as i tend to lean that way also)17:51
kialllol17:51
kiallOMG @ dch -i .. what a timesaver17:53
kiallcomitted rev 174 .. I presume I should find a patch file here somewhere/17:55
kiall?17:55
smoserkiall, oh. i guess i shoudl have said you'd have17:58
smoserto bzr add debian/patches17:58
smoseri hope17:58
smoseryou probalby didn't get hte patch added17:58
kiallNope :)17:58
kialladd and commit again?17:58
kiallor some form of bzr history rewrite ;)17:59
smoseryou can bzr uncommit18:02
smoserthat is probably what i would do18:02
smoseruncommit18:02
smoserthen, add18:02
smoserthen 'debcommit'18:02
smoseroh, and make sure your debian/changelog has '(LP: #BUGNUM)'18:02
smoserthat will tag the branch that you fixed that bug18:02
smoserand hten when you push the branch to somewhere (like 'bzr push lp:~yournamehere/ubuntu/maverick/eucalyptus/fix-public-addr-ping'18:03
smoser)18:03
smoserthat will get automatically linked to the bug18:03
smoserthen, you bother Daviey and tell him to not be lazy18:03
Davieyo/18:04
kiallRight so .. that should be pushed to lp:~kiall/ubuntu/natty/eucalyptus/fix-hairpin-nat ;)18:04
Davieykiall: What is the bug #?18:04
kiall67616718:05
Davieybug #67616718:05
uvirtbotLaunchpad bug 676167 in eucalyptus "Hairpin NAT on CC disabled - EC2 inconsistency" [Undecided,New] https://launchpad.net/bugs/67616718:05
kiallBTW .. It reads like it was translated via google translate a few times .. I apologize ;)18:05
Davieykiall: interesting... is that a regression?18:05
kiallsmoser, tested 10.04, i tested 10.10 .. so I dont think so18:06
Davieykiall: ok, thanks!18:06
Davieykiall: looks good....18:07
Daviey2 comments...18:07
kialldooh18:08
kiall1 see one of them18:08
kiallI see*18:08
Davieyfor maverick, as it is post release, the pocket needs to be "maverick-proposed" not "maverick" (top line of debian/changelog)18:08
Davieysecondly, it would be awesome if the patch had DEP-3 "patch tagging"18:08
Daviey"http://dep.debian.net/deps/dep3/"18:09
kiallRight so .. First time doing anything other than a quick PPA ;)18:09
DavieyYou can get an idea of what patch tagging looks like by looking at the other patches18:09
Davieykiall: You've done great (and the hard part)!  I don't mind fixing them up before merging..18:09
Davieyotherwise, awesome if you do it :)18:10
kialldont forget to remove the extra "network, slashnet, " I left in .. when re-doing after a clean clone ;)18:10
kiall(thats the 1 I spotted)18:10
Davieykiall: the other thing, the version number for maverick-proposed should be "2.0+bzr1241-0ubuntu4.1"18:10
kiallaha I'll blame that one on "dch -i" making me thing it knew better ;)18:11
Davieylol18:11
Davieykiall: seriously awesome work tho...  i'm over the moon you've contributed this18:12
kialllol .. 1 line patch .. cmon ;) (I work on some open source projects .. 1 lines aren't that hard to come by!)18:13
Davieykiall: we've traditionally not had great success getting contributions outside the usual suspects..18:13
Davieythat is the main reason i'm pleased18:13
DavieyOK... i need to go and do some family time..18:14
kiallsure .. cya18:14
* kiall gets back to my real work after 2 days on/off wondering what I had configured wrong ;)18:14
Davieykiall: one more comment... can you add an impact statement to the bug....  why this is importiant it's fixed in stable versions?18:14
Davieykiall: needs to be fixed in natty first, then either you - or me should backport it to maverick and lucid18:15
kiallsure ..18:15
Daviey\o/18:15
Davieykiall: ok.. speak soon o/18:16
kiallImpact statement added (At least I think I wrote an impact statement ;) .. maybe! ) .. right. home, then sleep. cya!18:24
kialloh god. reboot a server before I leave .. and ...18:39
kiall# shutdown -r now18:39
kiall-bash: /sbin/shutdown: Input/output error18:39
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!