*** joearnold has quit IRC | 00:00 | |
*** blakeyeager has quit IRC | 00:01 | |
bwalker7125 | quick question, is there a way to associate a hostname (like i-00000003) to a floating public ip address in openstack? | 00:01 |
---|---|---|
bwalker7125 | so that you could ping the hostname from another server in your network instead of pinging the ip address | 00:02 |
bwalker7125 | or ping the instance hostname from the openstack host | 00:02 |
*** blakeyeager has joined #openstack | 00:03 | |
*** proberts has quit IRC | 00:09 | |
*** dubsquared has left #openstack | 00:10 | |
*** ranger57 has quit IRC | 00:11 | |
*** hggdh_ has quit IRC | 00:16 | |
*** proberts has joined #openstack | 00:19 | |
*** dfg has quit IRC | 00:20 | |
*** adjohn has joined #openstack | 00:21 | |
*** msassak_ has joined #openstack | 00:22 | |
*** proberts has quit IRC | 00:25 | |
*** msassak has quit IRC | 00:26 | |
*** bwalker7125 has quit IRC | 00:27 | |
kbringard | I'm getting this error intermittently: http://paste.openstack.org/show/643/ | 00:31 |
kbringard | best I can tell, it happens when using an API key belonging to a user who is not an admin of the project | 00:31 |
kbringard | and it seems to happen almost all the time after adding a second compute node | 00:32 |
kbringard | any thoughts? | 00:32 |
*** msassak_ has quit IRC | 00:33 | |
*** kashyapc has quit IRC | 00:33 | |
*** hggdh has joined #openstack | 00:34 | |
*** proberts has joined #openstack | 00:34 | |
Ryan_Lane | what are some common reasons why a node would show as not working in nova-manage service list? | 00:34 |
Ryan_Lane | excluding time synch, I already checked that :( | 00:35 |
Ryan_Lane | I added a second node, and the first one disabled itself. | 00:35 |
kpepple | kbringard: you're losing session in cloud.py for some reason | 00:35 |
Ryan_Lane | also, is multi-node not supported in dhcp-flat mode? | 00:36 |
*** joearnold has joined #openstack | 00:39 | |
*** proberts has quit IRC | 00:39 | |
Ryan_Lane | WTF: csock.connect(('8.8.8.8', 80)) | 00:40 |
kpepple | kbringard: it's running through a list of running instances on your cloud, then if there is a fixed_ip on one instance it's trying to relate the fixed_ip to the network it is one to find out if it needs to do some ipv6 work (clouds.py:685-693) | 00:40 |
Ryan_Lane | in _get_my_ip.... | 00:40 |
kpepple | Ryan_Lane: weird ... it's looking for google's dns | 00:42 |
Ryan_Lane | I guess so that it can figure out what the public socket is | 00:42 |
Ryan_Lane | but that's really not a good way to go about that | 00:42 |
Ryan_Lane | in my instance, the nodes don't have access to 8.8.8.8, so it fails | 00:42 |
Ryan_Lane | network_host would be a much better thing to try there | 00:43 |
kpepple | Ryan_Lane: yeah, but you can set that in /etc/nova/nova.conf | 00:43 |
Ryan_Lane | yeah... still a bad default | 00:43 |
*** joearnold has quit IRC | 00:46 | |
Ryan_Lane | oh, great, it fails even if you set a default | 00:47 |
Ryan_Lane | since it still tries to hit the 8.8.8.8 address | 00:47 |
kpepple | Ryan_Lane: yeah, but it should catch it and return 127.0.0.1 | 00:47 |
Ryan_Lane | socket.error: [Errno 101] ENETUNREACH | 00:48 |
*** proberts has joined #openstack | 00:48 | |
Ryan_Lane | it isn't catching for that error | 00:48 |
Ryan_Lane | it's simply a dirty, dirty way to do it. if this was on a classified network, that would throw up flags | 00:49 |
kpepple | Ryan_Lane: it's only catching socket.gaierror | 00:50 |
*** kashyapc has joined #openstack | 00:50 | |
Ryan_Lane | yeah. it works after I changed it to socket.error | 00:50 |
*** rnirmal has joined #openstack | 01:01 | |
*** rnirmal has quit IRC | 01:01 | |
*** smaresca has joined #openstack | 01:03 | |
*** proberts has quit IRC | 01:05 | |
uvirtbot | New bug: #714928 in nova "_get_my_ip in flags.py should not try to contact 8.8.8.8" [Undecided,New] https://launchpad.net/bugs/714928 | 01:06 |
*** blakeyeager has quit IRC | 01:09 | |
uvirtbot | New bug: #714931 in nova "_get_my_ip in flags.py does not catch some socket exceptions" [Undecided,New] https://launchpad.net/bugs/714931 | 01:12 |
*** vvuksan has quit IRC | 01:13 | |
*** proberts has joined #openstack | 01:15 | |
kbringard | kpepple: thanks for the info, I'll keep digging! | 01:26 |
*** kbringard has quit IRC | 01:27 | |
*** maple_bed has joined #openstack | 01:27 | |
*** bcherian has quit IRC | 01:28 | |
*** dragondm has quit IRC | 01:35 | |
*** dendrobates is now known as dendro-afk | 01:47 | |
*** proberts has quit IRC | 01:48 | |
*** bcherian has joined #openstack | 01:49 | |
*** rlucio has quit IRC | 01:51 | |
*** maplebed has quit IRC | 01:55 | |
*** proberts has joined #openstack | 01:58 | |
* jaypipes is exhausted from this multi-tenant discussion... | 02:00 | |
Ryan_Lane | I'm not wanting to think how I'm going to have to implement that in LDAP | 02:01 |
Ryan_Lane | I'm thinking separate ous | 02:02 |
*** kpepple has quit IRC | 02:06 | |
*** kpepple has joined #openstack | 02:06 | |
gholt | jaypipes: Antagonist! I think I just don't want to implement "gimme the containers for accounts x, y, and z" in Swift. :) | 02:07 |
jaypipes | gholt: heh, fair enough :) | 02:08 |
jaypipes | Ryan_Lane: that's kinda where SQL datastores excel ;) | 02:09 |
*** vvuksan has joined #openstack | 02:09 | |
Ryan_Lane | yeah, but I'm assuming we'll be needing ldap support for it at well, right? | 02:10 |
* jaypipes notes it's kinda funny, that the only thing that Nova dev community has taken a stand on is not taking a stand on anything... | 02:10 | |
Ryan_Lane | lol | 02:10 |
gholt | :) | 02:10 |
Ryan_Lane | nested groups is good for stuff like this | 02:11 |
* gholt imagines a group of Stackers that actually start working on a Management project. | 02:11 | |
Ryan_Lane | not really that hard to implement | 02:11 |
jaypipes | Ryan_Lane: not with my proposal to handle that kind of thing in the Nova datastore, but to be fair, I completely understand the opposing view, I've been trying to illuminate some of the tradeoffs we make with that solution. | 02:11 |
Ryan_Lane | dropping LDAP support would be devastating for me. | 02:11 |
jaypipes | Ryan_Lane: there hasn't been *anyone* proposing that... | 02:11 |
Ryan_Lane | ah ok | 02:11 |
jaypipes | Ryan_Lane: I've been proposing storing account (note, *not* authentication/authorization data) relationships in the Nova data store. Others have been arguing that the auth service (i.e. LDAP, external SQL-based auth layer, or similar) should be responsible for that kind of thing... | 02:13 |
Ryan_Lane | I agree the external services should be | 02:13 |
*** littleidea has joined #openstack | 02:13 | |
jaypipes | Ryan_Lane: cool, so I take it you can store and query those relationships in LDAP :) | 02:14 |
Ryan_Lane | otherwise you are forced to use the API to manage those relationships | 02:14 |
jaypipes | rihgt | 02:14 |
Ryan_Lane | yes. exactly | 02:14 |
Ryan_Lane | I'm using those relationships for way more than they are intended | 02:14 |
jaypipes | :) | 02:14 |
Ryan_Lane | for instance, I'm adding posix group attributes to the projects | 02:14 |
jaypipes | oooh. | 02:14 |
*** bcherian has quit IRC | 02:14 | |
Ryan_Lane | so that when a user logs in to their instance, they are also in the posix group for that project | 02:14 |
jaypipes | I would say nice, but it seems hacky :) | 02:14 |
Ryan_Lane | then I'm allowing different sets of authorization based on the groups | 02:15 |
Ryan_Lane | my solution is fully integrated ;) | 02:15 |
jaypipes | heh, interesting :) | 02:15 |
Ryan_Lane | this is what LDAP is meant for though. heh | 02:15 |
Ryan_Lane | so it really isn't hacky, it's extensible | 02:16 |
Ryan_Lane | I'm using ldap dns entries as puppet nodes too, for instance | 02:16 |
Ryan_Lane | and I add the information about the user who created the node as puppet values, so that I can send that user an email when the instance is up | 02:18 |
Ryan_Lane | I add the user's language so that I can send them an email in their localized language | 02:18 |
jaypipes | Ryan_Lane: no, I was referring to storing the posix group attrs in the LDAP entry, not LDAP in general :) I would think it would just be easier to do "useradd user group" instead of storing the group in the LDAP entry :) | 02:18 |
Ryan_Lane | ah | 02:18 |
Ryan_Lane | project entries are groupofnames | 02:19 |
jaypipes | Ryan_Lane: don't worry, much as I hate LDAP's conventions, I recognize its purpose :P | 02:19 |
Ryan_Lane | so just need to add the posixgroup objectclass, and the attributes | 02:19 |
jaypipes | interesting | 02:19 |
Ryan_Lane | and then the systems pick it up automatically :) | 02:19 |
jaypipes | col | 02:19 |
jaypipes | cool | 02:19 |
jaypipes | :) | 02:19 |
jaypipes | this discussion is taking a toll on my typing abilities. | 02:19 |
jaypipes | not with you, I mean the mailing list :) | 02:20 |
Ryan_Lane | yeah. the nice thing about LDAP is that the schema is very strict. the sucky thing about LDAP is that the schema is very strict :) | 02:20 |
Ryan_Lane | heh | 02:20 |
Ryan_Lane | yeah | 02:20 |
jaypipes | Ryan_Lane: first rule of LDAP is don't talk about the LDAP schema. Second rule of LDAP is... | 02:20 |
Ryan_Lane | heh | 02:20 |
Ryan_Lane | the third rule is you never change the schema :) | 02:21 |
jaypipes | :) | 02:21 |
Ryan_Lane | have I mentioned I changed our schema? :) | 02:21 |
jaypipes | hehe | 02:21 |
eday | jaypipes: marathon email threads rock :) | 02:21 |
* eday is pooped | 02:21 | |
Ryan_Lane | I must need to sign back up for the list. I haven't been getting emails | 02:21 |
eday | Ryan_Lane: you on the openstack@ list? nova@ is gone | 02:22 |
Ryan_Lane | ah. that's the reason :) | 02:23 |
jaypipes | eday: :) | 02:23 |
Ryan_Lane | cool. on it now | 02:23 |
gholt | While fun, we might be losing anything productive on that thread at this point. :) | 02:26 |
jaypipes | eday, Ryan_Lane: http://twitpic.com/3xkmhn | 02:28 |
jaypipes | gholt: ^^ | 02:28 |
Ryan_Lane | :D | 02:28 |
creiht | hah | 02:29 |
Ryan_Lane | I get threads like this every week on wikitech-l :) | 02:29 |
creiht | Ryan_Lane: Are most of them from nelson? ;) | 02:29 |
Ryan_Lane | lol | 02:29 |
jaypipes | Ryan_Lane: lol | 02:29 |
Ryan_Lane | no, they are almost always about the same few things though | 02:29 |
*** bcherian has joined #openstack | 02:32 | |
annegentle | woohoo, OSCON proposal submitted - "Documentation in the Open" | 02:37 |
*** bcherian has quit IRC | 02:39 | |
*** littleidea has quit IRC | 02:42 | |
Ryan_Lane | hmmm. odd | 02:49 |
*** proberts has quit IRC | 02:49 | |
Ryan_Lane | my NAT rules on my network node are wrong after upgrading | 02:49 |
Ryan_Lane | it's setting the SNAT to one of my private interfaces, not my public one | 02:50 |
Ryan_Lane | even though my public_interface config option is set properly | 02:50 |
Ryan_Lane | ah. there's a new flag it seems :) | 02:51 |
*** jtran has quit IRC | 02:54 | |
*** jtran has joined #openstack | 02:58 | |
*** proberts has joined #openstack | 02:58 | |
*** jtran has quit IRC | 03:02 | |
*** proberts has joined #openstack | 03:08 | |
*** zaitcev has quit IRC | 03:13 | |
*** proberts has quit IRC | 03:16 | |
jarrod | ALRIGHT. new servers came in just as new compute released | 03:25 |
jarrod | score++ | 03:25 |
*** proberts has joined #openstack | 03:25 | |
*** proberts has quit IRC | 03:29 | |
*** proberts has joined #openstack | 03:38 | |
*** proberts has quit IRC | 03:42 | |
*** littleidea has joined #openstack | 03:46 | |
*** dendro-afk is now known as dendrobates | 03:51 | |
*** guynaor has joined #openstack | 03:51 | |
*** proberts has joined #openstack | 03:52 | |
*** guynaor has left #openstack | 04:01 | |
*** proberts has quit IRC | 04:06 | |
*** littleidea has quit IRC | 04:09 | |
*** hggdh has quit IRC | 04:10 | |
*** hggdh has joined #openstack | 04:11 | |
*** vvuksan has quit IRC | 04:12 | |
*** proberts has joined #openstack | 04:15 | |
*** dirakx has joined #openstack | 04:17 | |
*** proberts has quit IRC | 04:17 | |
openstackhudson | Project nova build #491: SUCCESS in 1 min 30 sec: http://hudson.openstack.org/job/nova/491/ | 04:22 |
openstackhudson | Tarmac: Catching all socket errors in _get_my_ip, since any socket error is likely enough to cause a failure in detection. | 04:22 |
*** proberts has joined #openstack | 04:26 | |
*** hadrian has quit IRC | 04:33 | |
*** proberts has quit IRC | 04:34 | |
*** proberts has joined #openstack | 04:44 | |
*** omidhdl has joined #openstack | 04:44 | |
*** mdomsch has joined #openstack | 04:46 | |
*** omidhdl1 has joined #openstack | 04:48 | |
*** omidhdl has quit IRC | 04:48 | |
*** sebastianstadil has quit IRC | 05:01 | |
*** proberts has quit IRC | 05:01 | |
uvirtbot | New bug: #714986 in nova "'nova-manage network create ...' populates incorrect data in database" [Undecided,New] https://launchpad.net/bugs/714986 | 05:02 |
*** baldben has joined #openstack | 05:03 | |
*** proberts has joined #openstack | 05:10 | |
*** baldben has quit IRC | 05:33 | |
*** proberts has quit IRC | 05:33 | |
*** baldben has joined #openstack | 05:34 | |
*** Hayati has joined #openstack | 05:40 | |
*** proberts has joined #openstack | 05:42 | |
*** mdomsch has quit IRC | 05:43 | |
*** sateesh has joined #openstack | 05:51 | |
*** MarkAtwood has joined #openstack | 05:58 | |
openstackhudson | Project nova build #492: SUCCESS in 1 min 38 sec: http://hudson.openstack.org/job/nova/492/ | 06:02 |
openstackhudson | Tarmac: Checks whether the instance id is a list or not before assignment. This is to fix a bug relating to nova/boto. The AWK-SDK libraries pass in a string, not a list. The euca tools pass in a list. | 06:02 |
*** adjohn has quit IRC | 06:08 | |
openstackhudson | Project nova build #493: SUCCESS in 1 min 36 sec: http://hudson.openstack.org/job/nova/493/ | 06:08 |
openstackhudson | * Tarmac: fixes #713766 and probably #710959, please test the patch before committing it. | 06:08 |
openstackhudson | * Tarmac: Updates to the multinode install doc based on Wayne's findings. Merged with trunk so should easily merge in. | 06:08 |
*** adjohn has joined #openstack | 06:10 | |
*** drico_ has quit IRC | 06:17 | |
*** proberts has quit IRC | 06:17 | |
*** drico_ has joined #openstack | 06:17 | |
*** proberts has joined #openstack | 06:26 | |
*** zenmatt has quit IRC | 06:29 | |
*** Hayati has quit IRC | 06:40 | |
*** ccustine has joined #openstack | 06:42 | |
*** proberts has quit IRC | 06:43 | |
*** pquerna has left #openstack | 06:48 | |
*** proberts has joined #openstack | 06:52 | |
*** miclorb has quit IRC | 07:01 | |
*** proberts has quit IRC | 07:08 | |
*** proberts has joined #openstack | 07:17 | |
*** proberts has quit IRC | 07:18 | |
*** littleidea has joined #openstack | 07:21 | |
*** proberts has joined #openstack | 07:28 | |
*** metoikos_ has joined #openstack | 07:31 | |
*** metoikos has quit IRC | 07:34 | |
*** proberts has quit IRC | 07:34 | |
*** guigui1 has joined #openstack | 07:34 | |
*** Hayati has joined #openstack | 07:34 | |
*** kashyapc has quit IRC | 07:37 | |
*** befreax has joined #openstack | 07:41 | |
*** proberts has joined #openstack | 07:43 | |
*** proberts has quit IRC | 07:47 | |
*** CloudChris has joined #openstack | 07:53 | |
CloudChris | Good morning everybody :) | 07:54 |
*** thimble has joined #openstack | 07:54 | |
*** ewindisch has quit IRC | 07:54 | |
*** proberts has joined #openstack | 07:56 | |
*** proberts has quit IRC | 07:57 | |
*** kashyapc has joined #openstack | 07:58 | |
*** miclorb_ has joined #openstack | 07:58 | |
*** adjohn has quit IRC | 07:58 | |
*** Nacx has joined #openstack | 08:01 | |
jaysun | morning CloudChris | 08:04 |
*** proberts has joined #openstack | 08:07 | |
*** sebastianstadil has joined #openstack | 08:08 | |
*** adjohn has joined #openstack | 08:11 | |
*** proberts has quit IRC | 08:13 | |
*** rchavik has joined #openstack | 08:13 | |
*** rchavik has joined #openstack | 08:13 | |
openstackhudson | Project dashboard-tarmac build #1,811: FAILURE in 22 sec: http://hudson.openstack.org/job/dashboard-tarmac/1811/ | 08:16 |
openstackhudson | Yippie, build fixed! | 08:20 |
openstackhudson | Project dashboard-tarmac build #1,812: FIXED in 2.1 sec: http://hudson.openstack.org/job/dashboard-tarmac/1812/ | 08:20 |
*** rcc has joined #openstack | 08:21 | |
*** proberts has joined #openstack | 08:22 | |
*** littleidea has quit IRC | 08:27 | |
*** proberts has quit IRC | 08:27 | |
*** Nick_ has joined #openstack | 08:30 | |
*** Nick_ is now known as Guest85402 | 08:30 | |
*** ccustine has quit IRC | 08:31 | |
*** sateesh has quit IRC | 08:33 | |
*** miclorb_ has quit IRC | 08:35 | |
*** proberts has joined #openstack | 08:36 | |
*** metoikos_ has quit IRC | 08:38 | |
*** perra has joined #openstack | 08:42 | |
*** perra has quit IRC | 08:43 | |
*** perra has joined #openstack | 08:44 | |
*** omidhdl1 has quit IRC | 08:48 | |
*** omidhdl has joined #openstack | 08:50 | |
*** proberts has quit IRC | 08:50 | |
*** miclorb has joined #openstack | 08:51 | |
*** berendt has joined #openstack | 08:52 | |
*** perra has quit IRC | 08:57 | |
*** proberts has joined #openstack | 08:59 | |
*** irahgel has joined #openstack | 09:01 | |
*** proberts has quit IRC | 09:05 | |
*** rcc has quit IRC | 09:08 | |
*** proberts has joined #openstack | 09:14 | |
*** allsystemsarego has joined #openstack | 09:27 | |
*** allsystemsarego has joined #openstack | 09:27 | |
*** calavera has joined #openstack | 09:27 | |
RJD22 | openstackhudson: what kind of daskboard is it? | 09:35 |
ttx | RJD22: openstackhudson is a bot, won't answer. The dashboard is the OpenStack dashboard, an admin web UI | 09:36 |
RJD22 | ah ok | 09:37 |
*** proberts has quit IRC | 09:39 | |
*** miclorb has quit IRC | 09:41 | |
*** miclorb__ has joined #openstack | 09:42 | |
*** proberts has joined #openstack | 09:48 | |
*** eikke has joined #openstack | 10:07 | |
*** omidhdl1 has joined #openstack | 10:08 | |
*** omidhdl has quit IRC | 10:10 | |
*** eikke has quit IRC | 10:13 | |
*** proberts has quit IRC | 10:14 | |
berendt | why can't i change the status of a bugreport (nova) to wishlist..? | 10:19 |
ttx | berendt: need to be part of the bug supervisor team to change others' bugs, I think | 10:21 |
ttx | berendt: which bug ?* | 10:21 |
berendt | https://bugs.launchpad.net/nova/+bug/715102 | 10:21 |
uvirtbot | Launchpad bug 715102 in nova "creation of a simple file based volume driver in nova/volume/driver.py" [Undecided,New] | 10:21 |
berendt | I think this is a wishlist entry.. :) | 10:21 |
berendt | but it's my own bug | 10:22 |
ttx | berendt: it should even be a blueprint, IMHO | 10:23 |
ttx | berendt: I marked it confirmed/Wishlist | 10:23 |
ttx | berendt: if you want to work on that for Cactus, you should create a blueprint about it | 10:23 |
berendt | ttx: yes? I think there are just a few calls to mount, rm and qemu-img.. (take a look into the linked branch) | 10:23 |
*** proberts has joined #openstack | 10:24 | |
berendt | but sure i can create a blueprint if this is necessary to include it into cactus | 10:24 |
ttx | berendt: not necessary, but certainly helps in tracking what new features end up in the release | 10:24 |
*** omidhdl1 has left #openstack | 10:24 | |
berendt | ttx: ok. i'll read how to create blueprints and will submit one | 10:25 |
ttx | berendt: http://wiki.openstack.org/BlueprintsLifecycle | 10:25 |
uvirtbot | New bug: #715102 in nova "creation of a simple file based volume driver in nova/volume/driver.py" [Wishlist,Confirmed] https://launchpad.net/bugs/715102 | 10:26 |
berendt | ttx: thanks.. maybe i have time to read it later.. have to prepare my rhcsa exam.. | 10:26 |
*** proberts has quit IRC | 10:27 | |
*** MarkAtwood has quit IRC | 10:30 | |
*** MarkAtwood has joined #openstack | 10:31 | |
*** proberts has joined #openstack | 10:36 | |
*** proberts has quit IRC | 10:39 | |
berendt | soren: https://bugs.launchpad.net/nova/+bug/683137 | 10:44 |
uvirtbot | Launchpad bug 683137 in nova "CA code doesn't get installed by "setup.py install"" [Medium,Triaged] | 10:44 |
berendt | soren: any news there? would be nice to have the solution in the code base.. we have to fix the bug at the moment manually in our SPEC file.. | 10:44 |
*** proberts has joined #openstack | 10:48 | |
soren | berendt: I'm going to rewrite all that CA stuff anyways. | 10:50 |
*** proberts has quit IRC | 10:50 | |
soren | berendt: In Python. So that it doesn't need all this special love and care. | 10:50 |
*** thimble has quit IRC | 10:53 | |
*** CloudChris has quit IRC | 10:56 | |
*** CloudChris has joined #openstack | 10:56 | |
*** arun_ has joined #openstack | 10:58 | |
*** proberts has joined #openstack | 10:59 | |
*** miclorb__ has quit IRC | 11:08 | |
*** proberts has quit IRC | 11:11 | |
* soren lunches | 11:15 | |
*** proberts has joined #openstack | 11:20 | |
*** guigui3 has joined #openstack | 11:25 | |
*** guigui1 has quit IRC | 11:26 | |
*** guigui3 has quit IRC | 11:27 | |
*** ramkrsna has joined #openstack | 11:27 | |
*** ramkrsna has joined #openstack | 11:27 | |
*** MarkAtwood has quit IRC | 11:35 | |
*** proberts has quit IRC | 11:35 | |
*** MarkAtwood has joined #openstack | 11:35 | |
*** sebastianstadil_ has joined #openstack | 11:39 | |
*** sebastianstadil has quit IRC | 11:42 | |
*** sebastianstadil_ is now known as sebastianstadil | 11:42 | |
*** proberts has joined #openstack | 11:44 | |
*** proberts has joined #openstack | 11:54 | |
*** Nacx has quit IRC | 11:56 | |
*** PeterHansen has joined #openstack | 11:59 | |
*** PeterHansen has quit IRC | 12:00 | |
*** proberts has quit IRC | 12:01 | |
*** sebastianstadil has quit IRC | 12:07 | |
*** proberts has joined #openstack | 12:10 | |
*** Nacx has joined #openstack | 12:13 | |
*** adjohn has quit IRC | 12:15 | |
*** ctennis has quit IRC | 12:18 | |
*** proberts has quit IRC | 12:22 | |
*** CloudChris has quit IRC | 12:24 | |
*** CloudChris has joined #openstack | 12:24 | |
*** thimble has joined #openstack | 12:28 | |
*** proberts has joined #openstack | 12:31 | |
*** proberts has quit IRC | 12:33 | |
*** ctennis has joined #openstack | 12:34 | |
*** ctennis has joined #openstack | 12:34 | |
*** proberts has joined #openstack | 12:43 | |
*** vvuksan has joined #openstack | 12:43 | |
*** perra has joined #openstack | 12:49 | |
*** heden has joined #openstack | 12:51 | |
berendt | ttx: can you please check my blueprint for the FileDriver: https://blueprints.launchpad.net/nova/+spec/filedriver | 12:53 |
*** heden has quit IRC | 12:53 | |
*** dprince has joined #openstack | 12:53 | |
*** heden has joined #openstack | 12:54 | |
ttx | berendt: set yourself as drafter and assignee, set "dendrobates" as approver, set "cactus" as the series goal, and set definition to "Pending approval" -- should be alright | 12:54 |
*** dprince has quit IRC | 12:55 | |
*** dprince has joined #openstack | 12:56 | |
*** reldan has joined #openstack | 12:56 | |
berendt | ttx: done | 12:56 |
ttx | berendt: looks good, thanks | 12:57 |
*** Hayati has quit IRC | 13:00 | |
*** fabiand_ has joined #openstack | 13:07 | |
*** proberts has quit IRC | 13:10 | |
*** proberts has joined #openstack | 13:19 | |
*** toddmorey has joined #openstack | 13:28 | |
*** ewindisch has joined #openstack | 13:28 | |
*** westmaas has joined #openstack | 13:31 | |
vvuksan | i am deploying Nova in an existing corporate network where I don't control the underlying network infrastructure | 13:32 |
vvuksan | I can request pool of internal IPs to use for my VMs | 13:32 |
vvuksan | so instead of using VLANManager I was thinking of using floating IPs | 13:33 |
vvuksan | i am unclear on the exact floating IP setup | 13:34 |
vvuksan | ie. how to bind certain floating IP block to certain set of compute hosts | 13:35 |
vvuksan | is that even possible? | 13:35 |
*** proberts has quit IRC | 13:38 | |
*** proberts has joined #openstack | 13:47 | |
*** citral has joined #openstack | 13:47 | |
*** hadrian has joined #openstack | 13:50 | |
*** dendrobates is now known as dendro-afk | 13:51 | |
*** dendro-afk is now known as dendrobates | 13:51 | |
*** jtimberman has quit IRC | 13:53 | |
*** jtimberman has joined #openstack | 13:54 | |
citral | hi, on Ubunutu 10.10 I am following http://wiki.openstack.org/NovaInstall/Bexar which went good, and it proceeds into http://wiki.openstack.org/RunningNova, but there the step sudo nova-manage project zipfile <user> gives me a problem. What should I do here? No such file or directory: '/usr/local/lib/python2.6/dist-packages/nova-2011.1-py2.6.egg/nova/..//CA' | 13:57 |
vvuksan | the way I resolved it | 13:58 |
vvuksan | one sec | 13:58 |
citral | I tried a symbolic link from nova/test/CA | 13:58 |
vvuksan | http://wiki.openstack.org/NovaInstall/CentOSNotes | 13:59 |
vvuksan | look for | 13:59 |
vvuksan | rsync -av /opt/nova/CA/ /var/lib/nova/CA/ | 13:59 |
vvuksan | assuming that the bexar code is in /opt/nova | 13:59 |
vvuksan | you also may need to set the path for ca_path | 13:59 |
*** fabiand_ has quit IRC | 14:01 | |
*** fabiand_ has joined #openstack | 14:01 | |
*** hub_cap has joined #openstack | 14:03 | |
*** rcc has joined #openstack | 14:06 | |
*** ppetraki has joined #openstack | 14:06 | |
rcc | hey guys | 14:06 |
rcc | https://bugs.launchpad.net/nova/+bug/708025 | 14:06 |
uvirtbot | Launchpad bug 708025 in nova "VLAN field in networks table should be constrained" [Low,Confirmed] | 14:06 |
rcc | what should be better, put the fix in nova-manage or in the VlanManager class? | 14:07 |
rcc | i.e. deeper the better? I'm more inclined to fix bad arguments at the pertinent file is getting them | 14:10 |
vvuksan | i'd say fix the underlying class | 14:10 |
*** befreax has quit IRC | 14:11 | |
rcc | k, i'll just do that thx | 14:11 |
citral | vvuksan: thanks, I symlinked /opt/nova-2011.1/CA/ to /usr/local/lib/python2.6/dist-packages/nova-2011.1-py2.6.egg/CA and then the error goes away. | 14:14 |
*** rchavik has quit IRC | 14:14 | |
vvuksan | i'd set ca_path instead | 14:15 |
vvuksan | but that would work | 14:15 |
citral | ah ok | 14:16 |
*** fsaad has joined #openstack | 14:25 | |
*** zenmatt has joined #openstack | 14:34 | |
uvirtbot | New bug: #715180 in nova "vlan100 and br100 still available after stopping nova-network: restart of nova-network not possible, br100 in state DOWN after starting nova-network" [Undecided,New] https://launchpad.net/bugs/715180 | 14:36 |
*** kashyapc has quit IRC | 14:37 | |
*** Ryan_Lane is now known as Ryan_Lane|away | 14:39 | |
citral | Question. I am still following http://wiki.openstack.org/RunningNova , is my impression correct that I cannot just use uec-publish-tarball without some setting up of the euca tools? (It doesn´t work right away.) Is it a better approach to try a manual registration process of the image? | 14:42 |
annegentle | citral: hey thanks for updating that wiki page. Actually the uec-publish-tarball is part of the cloud-utils package. | 14:44 |
annegentle | citral: check out https://help.ubuntu.com/community/UEC/BundlingImages9.10 | 14:44 |
*** f4m8 is now known as f4m8_ | 14:48 | |
*** ramkrsna has quit IRC | 14:52 | |
citral | annegentle: thanks for the pointer. uec-publish-tarball does not work out of the box for me (I installed the cloud-utils). It says it is unable to run euca-describe-images, which in turn gives a [Errno 111] Connection refused. The wiki doesn´t cover configuring euca2ools - is that necessary? | 14:57 |
*** reldan has quit IRC | 14:59 | |
annegentle | citral: that error sounds more like one of the nova- services isn't running. | 15:02 |
citral | annegentle: I think you´re right :) let me investigate that | 15:03 |
annegentle | citral: ok, go for it :) | 15:03 |
*** ccustine has joined #openstack | 15:05 | |
*** littleidea has joined #openstack | 15:11 | |
*** msassak has joined #openstack | 15:11 | |
*** rnirmal has joined #openstack | 15:24 | |
*** rnirmal has quit IRC | 15:26 | |
*** rnirmal has joined #openstack | 15:27 | |
*** hub_cap has joined #openstack | 15:27 | |
*** abecc has joined #openstack | 15:32 | |
*** iammartian has joined #openstack | 15:33 | |
*** reldan has joined #openstack | 15:34 | |
*** gondoi has joined #openstack | 15:40 | |
*** gondoi has quit IRC | 15:42 | |
*** gondoi has joined #openstack | 15:45 | |
citral | I got further now, and the previous error is fixed. But this error leaves me puzzled: http://paste.openstack.org/show/646/ | 15:47 |
*** magglass1 has quit IRC | 15:59 | |
*** dendrobates is now known as dendro-afk | 15:59 | |
*** dendro-afk is now known as dendrobates | 16:01 | |
*** dragondm has joined #openstack | 16:01 | |
annegentle | jaypipes ^^ is register_image() a glance call? Not sure I can troubleshoot further, sorry citral. | 16:01 |
cjreyn | can anyone tell me which of the ubuntu images here http://uec-images.ubuntu.com/releases/ work with bzr649 ? | 16:01 |
jaypipes | annegentle: looking... | 16:02 |
jaypipes | cjreyn: at least karmic and up should work. | 16:02 |
jaypipes | annegentle: no, register_image() is in euca-register, which is used by uec-publish-tarball. The issue is that the image has not been associated with a ramdisk (see the line that says ramdisk: none) | 16:03 |
*** Hayati has joined #openstack | 16:03 | |
jaypipes | annegentle: or at least, that's what it *looks like* :) | 16:04 |
jaypipes | devcamcar: check out citral's paste above...my explanation seem correct? | 16:04 |
vvuksan | cjreyn: should be any of themn | 16:07 |
vvuksan | cjreyn: although I'd say it's not to hard to bundle your own image | 16:07 |
cjreyn | mmm, I just tried 9.10 and console output shows its stuck at: * Setting preliminary keymap... * Waiting for EC2 meta-data service | 16:09 |
vvuksan | cjreyn: this is what I use to pack up RHEL images http://wiki.openstack.org/CreatingRHELImages | 16:09 |
*** perra has quit IRC | 16:09 | |
vvuksan | cjreyn: are you sure your setup works ? | 16:09 |
vvuksan | cjreyn: did you try the ttylinux images? | 16:09 |
cjreyn | well images boot, including the tiny image ttylinux-uec-amd64-12.1_2.6.35-22_1.tar.gz | 16:10 |
cjreyn | the bexar admin manual states that .tiny images cannot be ssh'd to | 16:10 |
vvuksan | can you connect into ttylinux after it boots? | 16:10 |
cjreyn | nope | 16:10 |
vvuksan | hmm | 16:10 |
vvuksan | i don't think that's correct | 16:11 |
*** blakeyeager has joined #openstack | 16:11 | |
vvuksan | can you ping it at least? | 16:11 |
cjreyn | nope. I just figured the problem was because of the image instantiation type | 16:11 |
*** kashyapc has joined #openstack | 16:11 | |
cjreyn | I condifured the br100 device manually | 16:11 |
vvuksan | ok | 16:12 |
*** Ryan_Lane|away is now known as Ryan_Lane | 16:12 | |
cjreyn | mm, the bridge is up, and has vnet0 and eth1 (my main interface) attached. | 16:13 |
*** dprince has quit IRC | 16:13 | |
*** dprince has joined #openstack | 16:14 | |
cjreyn | what are the -flat_injected and -public_interface flags in nova.conf I acn;t find any docs on these | 16:15 |
*** Guest85402 has quit IRC | 16:15 | |
*** rootwyrm has joined #openstack | 16:15 | |
rootwyrm | 'morning | 16:15 |
*** baldben has quit IRC | 16:17 | |
*** DanEllis has quit IRC | 16:18 | |
*** DanEllis has joined #openstack | 16:18 | |
*** toddmorey has quit IRC | 16:20 | |
*** magglass1 has joined #openstack | 16:24 | |
jaypipes | dprince: I promise I'll be getting to your bug this morning. sorry for the delay. | 16:24 |
jaypipes | rootwyrm: morning. | 16:24 |
rootwyrm | 'morning jaypipes. I'm interested in finding the POCs for OpenStack. | 16:26 |
rootwyrm | not sure who to talk to though. | 16:26 |
*** troytoman-away is now known as troytoman | 16:27 | |
*** thimble has quit IRC | 16:28 | |
*** abecc has quit IRC | 16:28 | |
*** abecc has joined #openstack | 16:31 | |
*** CloudChris has quit IRC | 16:31 | |
*** CloudChris has joined #openstack | 16:31 | |
annegentle | rootwyrm: POC = Proof Of Concept? | 16:31 |
rootwyrm | Point of Contact | 16:31 |
annegentle | rootwyrm: let me know your interest and I'll connect you | 16:32 |
* annegentle sounds like a switchboard operator | 16:32 | |
ttx | or Project Oversight Committee | 16:32 |
ttx | yay overloaded acronyms | 16:32 |
annegentle | ttx: yes that was my other guess :) | 16:32 |
rootwyrm | I'm the business developer for eRacks Open Source Systems. My CEO wants to get on board with the project and use OpenStack for our mass-NAS systems. | 16:32 |
*** westmaas has quit IRC | 16:33 | |
jaypipes | rootwyrm: POC? proof of concepts? point of contacts? | 16:33 |
rootwyrm | ... Point of Contact | 16:33 |
*** kashyapc has quit IRC | 16:33 | |
jaypipes | rootwyrm: depends on what you're looking to do. development? deployment? partnering? etc, etc :) | 16:34 |
rootwyrm | help with deployment, but primarily partnering. | 16:34 |
jaypipes | rootwyrm: are you interested in contributing your code as open source? | 16:34 |
rootwyrm | we | 16:34 |
*** guigui3 has joined #openstack | 16:34 | |
rootwyrm | we're going to help provide hardware for the software. | 16:34 |
*** calavera has quit IRC | 16:35 | |
jaypipes | rootwyrm: ah, I see. I would contact Jim Curry (jim@openstack.org) or Mark Collier (mark@openstack.org). They head up the business unit around OpenStack partnerships and can point you in the right direction. For developer contributions and interests, feel free to post to the mailing list (https://launchpad.net/~openstack) | 16:36 |
rootwyrm | nice. I'll write both Jim & Mark then. | 16:37 |
jaypipes | rootwyrm: cheers :) | 16:37 |
rootwyrm | If OpenStack decides they need some specialized hardware for any of its development, send them my way. I'll give a developer discount on our systems to help out. | 16:38 |
*** troytoman is now known as troytoman-away | 16:38 | |
ttx | jaypipes: about missing translations, I merged lp:~nova-core/nova/bexar-translations into my bexar branch -- any reason why this wouldn't be sufficient ? | 16:38 |
rootwyrm | but, our eRacks/NAS50 150TB 9U system might be rather nice to work with, or our 42U 1 Petabyte system. | 16:38 |
*** abecc has quit IRC | 16:39 | |
*** citral has quit IRC | 16:41 | |
*** abecc has joined #openstack | 16:42 | |
*** CloudChris has quit IRC | 16:42 | |
jaypipes | ttx: heh, because the translations aren't in there as binary files. | 16:43 |
jaypipes | ttx: still working on that...please don't merge it into bexar trunk yet :) | 16:43 |
ttx | jaypipes: aye | 16:43 |
cjreyn | ok, when I'm booting a tty image, it fails to get an IP via DHCP. Is there a way I can check that the DHCP server on the compute node is working, and bound to the bridge IP? | 16:43 |
ttx | jaypipes: this i18n stuff is a bit opaque to me :) | 16:44 |
* ttx would rather beat all the French into speaking properly english. | 16:44 | |
imsplitbit | ttx: "properly english"? | 16:45 |
jaypipes | hehe | 16:45 |
ttx | imsplitbit: heh | 16:46 |
uvirtbot | New bug: #715263 in glance "glance-combined not included in setup.py" [Undecided,New] https://launchpad.net/bugs/715263 | 16:46 |
vvuksan | cjreyn: on the compute node type iptables -t nat -l | 16:49 |
vvuksan | cjreyn: sorry it's -L and not -l | 16:49 |
*** blakeyeager has quit IRC | 16:51 | |
*** johnpur has joined #openstack | 16:51 | |
*** ChanServ sets mode: +v johnpur | 16:51 | |
*** kashyapc has joined #openstack | 16:51 | |
*** littleidea has quit IRC | 16:53 | |
cjreyn | cheers. Will these chains change when images are instantiated? | 16:55 |
*** dfg has joined #openstack | 16:55 | |
jaypipes | vishy: oh noes. thread re-ignited ;) | 16:56 |
ttx | jaypipes: excellent postmortem feedback, thanks | 16:58 |
jaypipes | ttx: welcome. | 16:59 |
vvuksan | cjreyn: look for line that says 169.254 ? | 16:59 |
vvuksan | is it there ? | 17:00 |
*** blakeyeager has joined #openstack | 17:00 | |
cjreyn | vvuksan: yes, in the preouting chain: DNAT tcp -- anywhere 169.254.169.254 tcp dpt:www to:192.168.40.2:8773 | 17:02 |
vvuksan | that's good | 17:02 |
cjreyn | so 40.2 is my bridges IP | 17:02 |
vvuksan | cjreyn: what kind of networking are you using ? Flat or FlatDHCP / | 17:02 |
cjreyn | FlatDHCP | 17:02 |
vvuksan | do a ps auxww | grep dnsmasq | 17:02 |
vvuksan | is it there ? | 17:03 |
cjreyn | yep | 17:03 |
vvuksan | says something like static ? | 17:03 |
cjreyn | but the listen address relates to the virbr0 interface IP | 17:03 |
vvuksan | not right | 17:04 |
vvuksan | likely you are missing path to nova-dhcpdbridge | 17:04 |
vvuksan | add something like this | 17:05 |
vvuksan | --dhcpbridge=/usr/bin/nova-dhcpbridge | 17:05 |
vvuksan | then make sure you restart nova-compute | 17:05 |
vvuksan | then try instantiating the instance again | 17:05 |
vvuksan | and actually one thing to do is | 17:06 |
vvuksan | run nova-dhcpbridge by hand make sure it doesn't error out | 17:06 |
vvuksan | also i'd kill dnsmasq listening on virbr0 | 17:06 |
vvuksan | that's the dnsmasq started by libvirt | 17:06 |
cjreyn | kill dnsmasq first? | 17:07 |
vvuksan | yeah | 17:07 |
vvuksan | nova-compute should start it | 17:07 |
*** troytoman-away is now known as troytoman | 17:08 | |
*** troytoma_ has joined #openstack | 17:08 | |
cjreyn | ok, so it doesnt restart when nova-compute restarts. my nova.conf does contain the --dhcpbridge=/usr/bin/nova-dhcpbridge flag | 17:08 |
cjreyn | to start it by hand? | 17:09 |
vvuksan | no | 17:09 |
vvuksan | just leave it | 17:09 |
vvuksan | i assume /usr/bin/nova-dhcpbridge exists ? | 17:09 |
cjreyn | yes | 17:10 |
vvuksan | nova do euca-run | 17:10 |
vvuksan | now :-) | 17:10 |
*** Hayati has quit IRC | 17:10 | |
cjreyn | yeah fired up an image... | 17:11 |
vvuksan | now watch it | 17:11 |
vvuksan | while it's booting check ps for dnsmasq | 17:11 |
vvuksan | has it started ? | 17:11 |
cjreyn | yeah it started, but console print form the VM again shows: udhcpc (v1.17.2) started | 17:12 |
cjreyn | Sending discover... | 17:12 |
cjreyn | Sending discover... | 17:12 |
cjreyn | Sending discover... | 17:12 |
cjreyn | No lease, forking to background | 17:12 |
cjreyn | starting DHCP forEthernet interface eth0 [ OK ] | 17:12 |
*** troytoman is now known as troytoman-away | 17:12 | |
*** troytoma_ has quit IRC | 17:12 | |
cjreyn | the debug print from the console in the vm does an ifconfig -a, and no IP has been assigned | 17:12 |
*** nelson has quit IRC | 17:13 | |
*** nelson has joined #openstack | 17:13 | |
*** baldben has joined #openstack | 17:13 | |
cjreyn | no dnsmasq on the compute node | 17:13 |
nelson | http://alsted.wikimedia.org/wikipedia/commons/thumb/a/aa/000_Finlanda_harta.PNG/75px-000_Finlanda_harta.PNG is coming from a 3-node swift cluster. | 17:13 |
*** forge has joined #openstack | 17:14 | |
*** proberts has joined #openstack | 17:14 | |
creiht | nelson: how goes it? | 17:14 |
*** forge is now known as Guest91310 | 17:14 | |
nelson | Along with 21GB of its closest friends. :) | 17:14 |
berendt | does somewhere exists a nice overview graphic for the swift architecture? i only now the one for nova the moment | 17:15 |
nelson | berendt: I haven't seen one. | 17:15 |
creiht | is that still with squid or are you trying varnish yet? | 17:15 |
creiht | nelson: is that public content, or still in test? | 17:15 |
nelson | creiht: that's going through squid. it's public in the sense that anybody who knows which files are in our 'aa' shard is free to fetch them. :) | 17:16 |
creiht | hehe | 17:16 |
vvuksan | cjreyn: try to run nova-dhcpbridge by hand | 17:17 |
vvuksan | see if it errors out | 17:17 |
vvuksan | i remember having issues with it starting | 17:17 |
cjreyn | kill the current instance 1st? | 17:17 |
vvuksan | forgot what I did | 17:17 |
nelson | creiht: that's including 鳴子峡レストハウス.JPG | 17:17 |
*** guigui3 has quit IRC | 17:17 | |
vvuksan | cjreyn: do dnsmasq is running or not ? | 17:17 |
cjreyn | ha ha, cool thanks for this! | 17:17 |
creiht | berendt: at what level are you looking for? | 17:17 |
cjreyn | should I stop the instance 1st? | 17:17 |
*** guigui1 has joined #openstack | 17:18 | |
nelson | creiht: e.g. http://alsted.wikimedia.org/wikipedia/commons/thumb/a/aa/%E9%B3%B4%E5%AD%90%E5%B3%A1%E3%83%AC%E3%82%B9%E3%83%88%E3%83%8F%E3%82%A6%E3%82%B9.JPG/800px-%E9%B3%B4%E5%AD%90%E5%B3%A1%E3%83%AC%E3%82%B9%E3%83%88%E3%83%8F%E3%82%A6%E3%82%B9.JPG | 17:18 |
berendt | creiht: like this one for nova: http://nova.openstack.org/_images/Novadiagram.png | 17:18 |
creiht | nelson: hehe | 17:19 |
nelson | aka http://alsted.wikimedia.org/wikipedia/commons/thumb/a/aa/鳴子峡レストハウス.JPG/800px-鳴子峡レストハウス.JPG | 17:19 |
annegentle | berendt: my take is that swift doesn't have a messaging system so it's not nearly as complex. Does text help? See http://swift.openstack.org/overview_architecture.html | 17:20 |
nelson | and it sure looks like the name of the image is the same as what's over the door in the image. | 17:20 |
creiht | berendt: yeah I need to add that to my list | 17:20 |
creiht | sorry we don't have a good overview of that yet | 17:21 |
nelson | annegentle: but some people grokk things visually better. | 17:21 |
berendt | annegentle: i know that text.. but for example a graphic for the rings and the communication between the swift nodes would be nice | 17:21 |
creiht | I've seen some presentations that have something similar though, let me see if I can dig one up | 17:21 |
nelson | creiht: I'd like to see it, too, if you would. | 17:21 |
* creiht nods | 17:22 | |
nelson | I have my own mental model of a visual representation of the architecture. | 17:22 |
berendt | creiht: thants | 17:22 |
annegentle | berendt: agreed, I'll repurpose what creiht digs up for the docs, visual images are always helpful. | 17:23 |
*** maplebed has joined #openstack | 17:24 | |
creiht | berendt, nelson: well the diagrams are not all that great :/ | 17:26 |
creiht | annegentle: Are you in SA any time this week? | 17:26 |
nelson | oh dear. | 17:26 |
*** dendrobates is now known as dendro-afk | 17:27 | |
berendt | creiht: ok | 17:27 |
*** drico_ has quit IRC | 17:28 | |
*** ramkrsna has joined #openstack | 17:28 | |
*** ramkrsna has joined #openstack | 17:28 | |
*** zenmatt has quit IRC | 17:30 | |
*** CloudChris has joined #openstack | 17:30 | |
*** zenmatt has joined #openstack | 17:30 | |
annegentle | creiht: yup, on thursday | 17:30 |
*** rlucio has joined #openstack | 17:30 | |
annegentle | creiht: let's draw on the back of a napkin! | 17:30 |
*** zenmatt has joined #openstack | 17:31 | |
* annegentle heads to lunch where there will be napkins in abundance | 17:32 | |
*** fabiand_ has quit IRC | 17:35 | |
*** drico has joined #openstack | 17:36 | |
*** dprince has quit IRC | 17:37 | |
wedjat | hello swift fans :) | 17:37 |
wedjat | does someone know which HTTP request do i need to update metadatas on a container ? for an object i just can use a HTTP POST request with the Metadata as a header, but the same doesn't work for containers | 17:38 |
*** zenmatt has joined #openstack | 17:39 | |
*** kbringard has joined #openstack | 17:39 | |
kbringard | hey guys, I have a question | 17:39 |
kbringard | I have a second compute node setup | 17:40 |
kbringard | I can see that it's registering with rabbitmq | 17:40 |
kbringard | but it never becomes available to nova | 17:40 |
kbringard | am I missing something? | 17:40 |
jaypipes | wedjat: maybe you need to PUT, not POST? | 17:40 |
kbringard | the scheduler and network controllers are running on the same machine as the CC, the only thing running on the second node is compute | 17:41 |
*** zenmatt has quit IRC | 17:42 | |
RJD22 | I'm using the deploy .py script to install a cloud but it doesn't work as expected | 17:43 |
RJD22 | it hangs after the first comman | 17:43 |
RJD22 | command* | 17:44 |
RJD22 | ssh -o StrictHostKeyChecking=no -t <username-here>@cloud-controller 'sudo -u root sh -c "apt-get -y --force-yes install python-yaml"' | 17:44 |
wedjat | jaypipes: PUT or POST, the two methods work for that operation :) | 17:44 |
RJD22 | it runs this command and nothing happens | 17:44 |
wedjat | jaypipes: it was a problem with the X-Container-metadata' which need to be well written | 17:45 |
RJD22 | is there noone who might know what is going wrong? | 17:47 |
*** joearnold has joined #openstack | 17:47 | |
vvuksan | kbringard: hey | 17:48 |
kpepple | RJD22: sorry, haven't used the deploy script | 17:48 |
kbringard | hey vvuksan | 17:49 |
kbringard | how are ya? | 17:49 |
vvuksan | kbringard: good. Do select * from services | 17:49 |
vvuksan | see if it registered there | 17:49 |
kbringard | I did, it is | 17:49 |
RJD22 | nvm it's working | 17:49 |
RJD22 | but I don't know why :P | 17:49 |
kbringard | it's marked as enabled (or rather, disabled = 0) | 17:49 |
kbringard | nova-manage service list shows it there, but it has XXX instead of :-) | 17:50 |
vvuksan | yeah | 17:50 |
*** blueadept has joined #openstack | 17:50 | |
vvuksan | i had the same | 17:50 |
kbringard | I even blasted it from the DB and recreated the entire queue | 17:50 |
kbringard | it all reregistered automagically | 17:50 |
kbringard | but it still doesn't let me use it | 17:50 |
*** dendro-afk is now known as dendrobates | 17:51 | |
*** eikke has joined #openstack | 17:51 | |
vvuksan | however it will still deploy | 17:51 |
kbringard | well, when I shut down the other compute node, VMs get stuck in scheduling | 17:51 |
kbringard | until I bring back the other compute node | 17:52 |
kbringard | the scheduler is only trying to cast them to the compute node that it thinks is working | 17:52 |
vvuksan | it should take about a minute for it to pronounce the node dead | 17:52 |
blueadept | anyone know what google's logo is about today? | 17:52 |
kbringard | blueadept: Jules Verne | 17:53 |
kbringard | wrote 20k Leagues Under the Sea | 17:53 |
blueadept | ah | 17:53 |
blueadept | i thought it had to do with that | 17:53 |
kbringard | and Journey to the Center of the Earth, etc | 17:54 |
kbringard | if you click on the logo | 17:54 |
kbringard | it does a search for what it's about | 17:54 |
RJD22 | running into a new error | 17:54 |
blueadept | i was too busy messing with the controller to know to click on the image | 17:54 |
kbringard | haha, yea, me too | 17:54 |
kbringard | it's fun | 17:54 |
RJD22 | err: Could not retrieve catalog from remote server: hostname was not match with the server certificate | 17:54 |
vvuksan | kbringard: are you using floating IPs ? | 17:55 |
RJD22 | it's when trying to setup puppet | 17:55 |
kbringard | vvuksan: I don't think so... at least not intentionally | 17:55 |
kbringard | although, I did't set my network information in my nova.conf, since it's all in the DB | 17:55 |
vvuksan | kbringard: I'm thinking of using floating IPs instead of VLAN stuff | 17:55 |
vvuksan | i'm just unclear how it works | 17:56 |
*** guigui1 has quit IRC | 17:57 | |
creiht | annegentle: yeah come on over to the swift area and we can work on a diagram | 18:01 |
*** sebastianstadil has joined #openstack | 18:01 | |
*** blakeyeager has quit IRC | 18:01 | |
creiht | wedjat: I don't remember off the top of my head, but I think it should be a POST similar to object metadata | 18:02 |
vvuksan | kbringard: i tested your scenario | 18:04 |
creiht | wedjat: yeah, make a post to the container, and each of your metadata items should start with X-Container-Meta- | 18:04 |
vvuksan | kbringard: you may be running into the timeout issue | 18:05 |
vvuksan | it takes about a minute or so before a node is marked down | 18:05 |
kbringard | well, even still | 18:05 |
kbringard | I shut down all compute nodes, and wait for the status to show them all as down | 18:06 |
kbringard | then I start the compute node on the machine that isn't the CC | 18:06 |
kbringard | and it never shows it as available | 18:06 |
vvuksan | hmm | 18:07 |
kbringard | same thing with the scheduler | 18:07 |
vvuksan | i'd file a bug | 18:07 |
kbringard | the services on the machine that isn't the CC seem to be registering, but never become "available" | 18:07 |
wedjat | creiht: yeah, it works perfectly, as it should ! :) | 18:09 |
wedjat | creiht: i'm searching now how to create and use groups in swift, it should be a little tricky | 18:09 |
kbringard | I wonder if this is a networking issue... like the heartbeat can't connect to realize the stuff is up | 18:10 |
*** m_3 has quit IRC | 18:11 | |
*** edyoon has quit IRC | 18:11 | |
*** m_3 has joined #openstack | 18:12 | |
vvuksan | cjreyn has the same issue I have had with nova-dhcpbridge setting 10.0.0.2 for dhcprange | 18:13 |
*** m_3 has quit IRC | 18:13 | |
*** m_3 has joined #openstack | 18:14 | |
*** Nacx has quit IRC | 18:15 | |
*** eikke has quit IRC | 18:17 | |
*** pvo has joined #openstack | 18:18 | |
*** ppetraki has quit IRC | 18:22 | |
*** dprince has joined #openstack | 18:32 | |
Ryan_Lane | kbringard: are the times on your nodes sync'd? | 18:33 |
Ryan_Lane | kbringard: also, which networking mode are you using? | 18:33 |
kbringard | Ryan_Lane: I'm using VLAN mode, but I think the time issue is the problem | 18:34 |
kbringard | smart man | 18:34 |
Ryan_Lane | RJD22: that means the puppet client is trying to connect to the puppet server, but the puppet server's CA doesn't match the server name you have configured | 18:34 |
Ryan_Lane | kbringard: I ran into it a few weeks ago :) | 18:34 |
kbringard | i didn't even think to check... they were off by about 5 mins | 18:34 |
Ryan_Lane | RJD22: are you using an IP for the server name? | 18:34 |
Ryan_Lane | that's enough to do it | 18:34 |
Ryan_Lane | kbringard: mine was off by like 2 minutes, and that was too much | 18:35 |
Ryan_Lane | seems NTP is a necessity | 18:35 |
kbringard | yep, that fixed it | 18:35 |
kbringard | you are the man | 18:35 |
Ryan_Lane | \o/ | 18:35 |
Ryan_Lane | RJD22: for puppet server you must use a FQDN for the name :) | 18:36 |
vvuksan | Ryan_Lane: do you know much about floating ip setup ? | 18:39 |
Ryan_Lane | only that nova does it for me automatically :) | 18:39 |
Ryan_Lane | I'm using FlatDHCP | 18:40 |
Ryan_Lane | with multiple NICs | 18:40 |
*** CloudChris has quit IRC | 18:40 | |
Ryan_Lane | so, if you are using that networking mode, and have two interfaces, I can likely help you | 18:40 |
Ryan_Lane | I know nothing about vlan mode, and it doesn't really work in flat mode | 18:41 |
kbringard | Ryan_Lane: I'm using VLAN mode with multiple interfaces and it works great | 18:42 |
Ryan_Lane | it's working perfectly for me in flatdhcp mode too :) | 18:43 |
kbringard | very nice | 18:44 |
*** dendrobates is now known as dendro-afk | 18:48 | |
*** dendro-afk is now known as dendrobates | 18:48 | |
*** bwalker7125 has joined #openstack | 18:50 | |
*** reldan has quit IRC | 18:53 | |
kbringard | vvuksan: how's that CentOS 5.5 OpenStack stuff working for you? | 18:57 |
vvuksan | kbringard: works good | 18:58 |
kbringard | nice | 18:58 |
kbringard | RHEL6 was pretty easy to get installed, but I am concerned about a couple of things (like, no qemu-nbd... and when Cent6 will come out) | 18:59 |
*** hggdh has quit IRC | 18:59 | |
*** hggdh has joined #openstack | 18:59 | |
vvuksan | kbringard: the issues I'm trying to resolve are floating IPs | 19:02 |
vvuksan | and also creation of disk.local | 19:02 |
*** irahgel has left #openstack | 19:03 | |
sirp_ | jaypipes: looks like nova trunk is now pointed to cactus, mind doing the same for glance? | 19:03 |
*** MarkAtwood has quit IRC | 19:04 | |
*** jakedahn has joined #openstack | 19:09 | |
openstackhudson | Project nova build #494: SUCCESS in 1 min 34 sec: http://hudson.openstack.org/job/nova/494/ | 19:12 |
openstackhudson | Tarmac: Automates the setup for FlatDHCP regardless of whether the interface has an ip address. | 19:12 |
openstackhudson | FlatDHCP only has worked until now if the underlying interface that you bridge into has no ip address. This branch will do the necessary setup to allow it to work by: | 19:12 |
openstackhudson | * Creating a bridge with the private address for dhcp | 19:12 |
openstackhudson | * Moving any existing ips from the interface onto the bridge | 19:12 |
openstackhudson | * Adding the interface to the bridge | 19:12 |
openstackhudson | * Recreating the default route if it was deleted by moving the interface | 19:12 |
openstackhudson | It will additionally add a route to compute hosts for bridges that it adds to the compute hosts. This seems to be necessary in some cases where the default route uses a different interface. | 19:12 |
*** diegoparrilla_ has joined #openstack | 19:13 | |
*** burris has quit IRC | 19:14 | |
*** diegoparrilla_ is now known as diegoparrilla | 19:14 | |
*** iammartian has quit IRC | 19:19 | |
*** fabiand_ has joined #openstack | 19:20 | |
*** iammartian has joined #openstack | 19:20 | |
*** ramkrsna has quit IRC | 19:20 | |
nelson | In swift/common/wsgi.py is the call to sleep() a call to time.sleep() or eventlet.sleep() ? | 19:20 |
nelson | Only way to find out is to searching for the "from evenelet import sleep". | 19:21 |
nelson | Why oh why do people feel the need to use "from/import"? | 19:21 |
* nelson will whinge for money. | 19:21 | |
*** pvo has quit IRC | 19:22 | |
*** dendrobates is now known as dendro-afk | 19:24 | |
*** nelson has quit IRC | 19:29 | |
*** nelson has joined #openstack | 19:29 | |
ttx | Team meeting in 90 minutes in #openstack-meeting | 19:31 |
RichiH | ttx: open for all? | 19:32 |
ttx | RichiH: of course | 19:32 |
RichiH | _team_ meeting not per se, no :) | 19:32 |
ttx | RichiH: we do open design and open development. so the team is open :) | 19:32 |
RichiH | ttx: there's that, agreed | 19:33 |
RichiH | soren stressed that point during his talk | 19:33 |
RichiH | still, asking before joining a channel is good style :) | 19:33 |
ttx | argh, FOSDEM fallback already :) | 19:33 |
RichiH | i heard about openstack on thursday and went to fosdem on friday | 19:34 |
*** CloudChris has joined #openstack | 19:34 | |
ttx | RichiH: got a presentation on Sunday and went on the IRC channel on Tuesday ? | 19:34 |
nelson | "from eventlet.green import urllib2" is destructive of readability. Is urllib2 the urllib2 module or is it eventlet's urllib2? | 19:35 |
*** pvo has joined #openstack | 19:35 | |
ttx | RichiH: meeting agenda is at http://wiki.openstack.org/Meetings | 19:35 |
RichiH | ttx: and read up on it on saturday | 19:36 |
nelson | And yet that's one of the examples of how they suggest you should code using eventlet. | 19:36 |
* nelson could whinge all day long. | 19:36 | |
dprince | ttx: Can you assign #713144 to dprince (and or titan). | 19:36 |
ttx | bug 713144 | 19:37 |
uvirtbot | Launchpad bug 713144 in nova "Openstack API: GET on /servers/<id> returns empty 'addresses'" [High,Confirmed] https://launchpad.net/bugs/713144 | 19:37 |
dprince | ttx: yes sir | 19:37 |
dprince | ttx: thanks. | 19:37 |
RichiH | kpepple: feel free to ask me for local time & client directly | 19:37 |
ttx | dprince: done. I'm surprised you can't do it yourself | 19:38 |
*** pvo has quit IRC | 19:38 | |
kpepple | RichiH: my annoying irc client does that automagically ... sorry | 19:39 |
RichiH | kpepple: yah, no worries. it was just weird, is all | 19:40 |
*** pvo has joined #openstack | 19:41 | |
RichiH | btw, as there were way too people around soren at the time to ask him directly: is there anything in openstack to mitigate high i/o loads? on our current platforms, the only real problem is i/o | 19:43 |
*** dendro-afk is now known as dendrobates | 19:45 | |
*** johnpur_ has joined #openstack | 19:46 | |
*** proberts has quit IRC | 19:46 | |
*** proberts has joined #openstack | 19:47 | |
RichiH | also, from what soren presented, there is nothing to prevent an attacker from creating a lot of hashes of storage objects so that they will all end up on the same server | 19:48 |
*** johnpur_ has quit IRC | 19:48 | |
RichiH | that's the classic problem when basing a ring's DHT on outside input | 19:48 |
RichiH | one possible, and almost trivial, solution would be to expand the node mapping in those regions of the ring that have a higher load | 19:49 |
RichiH | though finding collisions on the first four bytes of a md5sum might (soon) be simple enough to create identical indices, in which case expanding the hash key length might make sense | 19:51 |
*** dprince has quit IRC | 19:51 | |
*** dprince has joined #openstack | 19:53 | |
uvirtbot | New bug: #715372 in glance "Add --config-file option" [Medium,In progress] https://launchpad.net/bugs/715372 | 19:56 |
creiht | RichiH: In swift, the hashes are salted, so the user couldn't do that unless they knew the salt | 19:57 |
*** bwalker7125 has quit IRC | 19:58 | |
Ryan_Lane | how does one set up the ajax terminals? | 19:59 |
*** sebastianstadil_ has joined #openstack | 20:00 | |
*** sebastianstadil has quit IRC | 20:03 | |
*** sebastianstadil_ is now known as sebastianstadil | 20:03 | |
*** brd_from_italy has joined #openstack | 20:08 | |
*** joearnold has quit IRC | 20:16 | |
*** fabiand_ has quit IRC | 20:20 | |
*** fabiand_ has joined #openstack | 20:20 | |
*** ppetraki has joined #openstack | 20:25 | |
Ryan_Lane | do I need to enable serial consoles via the xml template for libvirt? | 20:26 |
Ryan_Lane | or does the default one do that? | 20:26 |
Ryan_Lane | and if the serial console is enabled, I can test it using virsh console, correct? | 20:27 |
rlucio | Ryan_Lane: afaik the bexar libvirt xml files have support for serial, but you also have to have serial console access setup in the VM | 20:31 |
rlucio | see the nova/tools/nova-debug script | 20:31 |
rlucio | for config | 20:31 |
RichiH | creiht: can the user find out the hashes? | 20:31 |
Ryan_Lane | rlucio: thanks | 20:31 |
creiht | RichiH: nope | 20:31 |
Ryan_Lane | cat >t/etc/init/ttyS0.conf <<TTY_EOF | 20:32 |
Ryan_Lane | rlucio: I'm assuming that part ^^ | 20:32 |
* Ryan_Lane sticks that in his user-data for upstarts | 20:33 | |
Ryan_Lane | do I need to have direct access to the compute nodes for ajax-term support? | 20:35 |
Ryan_Lane | or will it proxy via the network node? | 20:35 |
RichiH | creiht: ah, k. else, it would be trivial to find the salt | 20:38 |
ttx | Team meeting in 15 minutes in #openstack-meeting ! | 20:45 |
*** abecc_ has joined #openstack | 20:46 | |
*** abecc has quit IRC | 20:46 | |
*** abecc_ is now known as abecc | 20:46 | |
*** troytoman-away is now known as troytoman | 20:48 | |
creiht | RichiH: indeed :) | 20:48 |
ttx | Ryan_Lane: there is some explanation on the merge proposal at https://code.launchpad.net/~anso/nova/ajaxterm/+merge/45068 | 20:50 |
Ryan_Lane | ttx: great. thanks :) | 20:51 |
*** TheHutch has joined #openstack | 20:51 | |
Ryan_Lane | I've been tracing through the code | 20:51 |
* ttx thinks that should been turned into doc :) | 20:51 | |
Ryan_Lane | heh | 20:51 |
Ryan_Lane | likely | 20:51 |
Ryan_Lane | or at least a wiki page | 20:51 |
ttx | annegentle: ^ | 20:51 |
*** fabiand_ has quit IRC | 20:51 | |
Ryan_Lane | eh? there's a new service for this? do I need to install it? I don't see it in the package lists | 20:52 |
*** dprince has quit IRC | 20:53 | |
ttx | Ryan_Lane: it's been lost in https://bugs.launchpad.net/nova/+bug/714678 | 20:53 |
uvirtbot | Launchpad bug 714678 in nova "Missing items in Bexar release tarball" [Critical,In progress] | 20:53 |
*** dprince has joined #openstack | 20:53 | |
ttx | and accesorily was never installed by the ubuntu packaging | 20:53 |
* annegentle reads | 20:53 | |
Ryan_Lane | :D | 20:53 |
Ryan_Lane | that'll make it hard to use. heh | 20:53 |
ttx | Ryan_Lane: that's what you get when you forget to add your binaries to setup.py :P | 20:54 |
Ryan_Lane | indeed :) | 20:54 |
*** westmaas has joined #openstack | 20:54 | |
ttx | Ryan_Lane: you can test from branch | 20:54 |
*** Ryan_Lane has quit IRC | 20:58 | |
ttx | Join us now: meeting starting NOW in #openstack-meeting ! | 20:59 |
*** bcwaldon has quit IRC | 21:01 | |
*** RJD22 is now known as RJD22|away | 21:01 | |
*** bcwaldon has joined #openstack | 21:01 | |
*** troytoma_ has joined #openstack | 21:02 | |
*** Ryan_Lane has joined #openstack | 21:03 | |
*** vvuksan has quit IRC | 21:06 | |
*** vvuksan1 has joined #openstack | 21:06 | |
*** allsystemsarego has quit IRC | 21:06 | |
*** troytoman is now known as troytoman-away | 21:06 | |
*** RJD22|away is now known as RJD22 | 21:07 | |
*** dprince has quit IRC | 21:10 | |
*** dprince has joined #openstack | 21:11 | |
*** pandemicsyn has quit IRC | 21:12 | |
*** pandemicsyn has joined #openstack | 21:12 | |
openstackhudson | Project swift build #191: SUCCESS in 29 sec: http://hudson.openstack.org/job/swift/191/ | 21:16 |
openstackhudson | Tarmac: fix st <command> --help hangs | 21:16 |
*** joearnold has joined #openstack | 21:17 | |
RichiH | once the first release that's usable by rackspace is made, will i be able to a) upgrade a cloud to new versions while it's live and b) run existing machines on newer cloud servers? | 21:20 |
*** jakedahn has quit IRC | 21:20 | |
*** EdwinGrubbs has joined #openstack | 21:21 | |
*** EdwinGrubbs has joined #openstack | 21:21 | |
rlucio | Ryan_Lane: sry i was out at lunch.. did you get the whole serial console thing ironed out? | 21:21 |
creiht | RichiH: For object storage (swift), I can say yes, as we already have to do that for Cloud Files | 21:22 |
*** DanEllis has quit IRC | 21:24 | |
*** aryko_ has joined #openstack | 21:27 | |
*** aryko_ is now known as aryko | 21:28 | |
*** dirakx has quit IRC | 21:28 | |
*** pvo has quit IRC | 21:30 | |
*** MarkAtwood has joined #openstack | 21:31 | |
uvirtbot | New bug: #715424 in nova "nova-manage : create network crashes when subnet range provided is not enough" [Undecided,New] https://launchpad.net/bugs/715424 | 21:31 |
*** dprince has quit IRC | 21:33 | |
*** pvo has joined #openstack | 21:34 | |
*** ewindisch has quit IRC | 21:35 | |
*** eikke has joined #openstack | 21:36 | |
uvirtbot | New bug: #715427 in nova "nova-manage : create network doesn't check if the fixed_ips are already existing" [Undecided,New] https://launchpad.net/bugs/715427 | 21:36 |
vvuksan1 | what service is responsible for kicking off nova-dhcpbridge in flatdhcp mode ? | 21:37 |
vvuksan1 | i have nova-compute running on the machine but it doesn't seem to kick off dnsmasq | 21:38 |
*** pvo has quit IRC | 21:39 | |
*** miclorb has joined #openstack | 21:39 | |
*** eikke has quit IRC | 21:40 | |
*** abecc has quit IRC | 21:41 | |
*** brd_from_italy has quit IRC | 21:42 | |
*** baldben has quit IRC | 21:46 | |
berendt | vvuksan1: nova-dhcpbrige should be started by dnsmasq | 21:49 |
vvuksan1 | not the other way around ? | 21:50 |
*** troytoma_ has quit IRC | 21:51 | |
soren | Right. | 21:51 |
berendt | vvuksan1: dnsmasq --strict-order --bind-interfaces --conf-file= --domain=example.com --pid-file=/var/lib/nova/networks/nova-br100.pid --listen-address=192.168.0.1 --except-interface=lo --dhcp-range=192.168.0.3,static,120s --dhcp-hostsfile=/var/lib/nova/networks/nova-br100.conf --dhcp-script=/usr/bin/nova-dhcpbridge --leasefile-ro | 21:51 |
*** ctennis has quit IRC | 21:51 | |
soren | nova-dhcpbridge is a script called by dnsmasq to inform nova about state transitions and such. | 21:51 |
*** baldben has joined #openstack | 21:52 | |
vvuksan1 | so i got it working somehow on one node | 21:52 |
vvuksan1 | however on the other node it's not working | 21:53 |
*** berendt has quit IRC | 21:54 | |
uvirtbot | New bug: #715443 in nova "EC2 APIs Create Keypairs/Security Groups etc above database column limits gives wierd results" [Undecided,New] https://launchpad.net/bugs/715443 | 21:56 |
*** kpepple has left #openstack | 21:57 | |
*** kpepple has joined #openstack | 21:57 | |
*** ppetraki has quit IRC | 21:58 | |
*** dendrobates is now known as dendro-afk | 21:59 | |
*** grapex has joined #openstack | 22:01 | |
vvuksan1 | soren: question is which daemon actually invokes nova-dhcpbridge ? | 22:02 |
vvuksan1 | is it nova-compute ? | 22:02 |
vvuksan1 | or something else | 22:02 |
*** aryko has quit IRC | 22:03 | |
*** ctennis has joined #openstack | 22:04 | |
*** ctennis has joined #openstack | 22:04 | |
*** CloudChris has quit IRC | 22:07 | |
vishy | vvuksan1: it is dnsmasq | 22:08 |
vishy | vvuksan1: (dnsmasq is started by nova-network) | 22:08 |
vvuksan1 | ok that is what I was looking for :-) | 22:09 |
vvuksan1 | so on compute nodes I need nova-compute and nova-network | 22:09 |
vvuksan1 | right ? | 22:09 |
*** abecc has joined #openstack | 22:09 | |
vvuksan1 | also if I have both of those running and dnsmasq is not being started up | 22:11 |
vvuksan1 | where would I look :-/ | 22:11 |
*** abecc has quit IRC | 22:11 | |
vishy | it doesn't get started up | 22:12 |
vishy | it gets called whenever a lease changes | 22:12 |
*** dendro-afk is now known as dendrobates | 22:12 | |
vvuksan1 | so nova-compute kicks off a machine | 22:12 |
vvuksan1 | but there is no dnsmasq running so it never gets an IP :-/ | 22:12 |
vishy | ah | 22:12 |
vishy | nova-network kicks off dnsmasq | 22:12 |
*** mlai has joined #openstack | 22:12 | |
vvuksan1 | and nova-network is running | 22:13 |
vvuksan1 | but nothing happens | 22:13 |
vishy | which networking mode? | 22:13 |
vvuksan1 | flatdhcp | 22:13 |
vvuksan1 | unless nova-network didn't register itself ? | 22:13 |
vishy | have you set --network_host | 22:13 |
vishy | ? | 22:13 |
vvuksan1 | in the services table | 22:13 |
vishy | in flat modes, network host is based on the flag | 22:14 |
vvuksan1 | i did not set network_host | 22:14 |
vishy | for example nova-compute --network_host=<some hostname where nova-network is running> | 22:14 |
vishy | or set it in nova.conf on the compute hosts | 22:14 |
vvuksan1 | --network_host=127.0.0.1 ? | 22:14 |
vishy | no | 22:15 |
vishy | hostname | 22:15 |
vvuksan1 | ah | 22:15 |
vishy | is nova-network running on the same box as nova-compute? | 22:15 |
vvuksan1 | yes | 22:15 |
vishy | that is odd, it should work by default then | 22:15 |
vvuksan1 | this may be something that is not a typical setup | 22:15 |
vvuksan1 | and hopefully you can shed some light | 22:15 |
vishy | do you get any errors in nova-network log? | 22:16 |
vvuksan1 | no errors | 22:16 |
vvuksan1 | but let me get this straight | 22:16 |
vishy | and you get nothing for ps ax | grep dnsmasq? | 22:16 |
vishy | it should start dnsmasq when it starts | 22:16 |
vvuksan1 | for flatdhcp there should be a single nova-network daemon on the whole network | 22:16 |
vvuksan1 | correct ? | 22:16 |
vishy | yes | 22:16 |
vvuksan1 | yeah I dont' see dnsmasq | 22:16 |
vvuksan1 | so that is my problem | 22:16 |
vvuksan1 | basically what I'm doing | 22:16 |
vishy | yeah that is odd | 22:16 |
vvuksan1 | I have two hosts | 22:16 |
vvuksan1 | i don't control the infrastructure only hosts | 22:17 |
*** baldben has quit IRC | 22:17 | |
vvuksan1 | it's all internal work so no external IPs needed | 22:17 |
vvuksan1 | i was gonna use br100 -> 192.168.0.0/24 | 22:17 |
vvuksan1 | then use 10.x for floating IPs | 22:17 |
vvuksan1 | would that even work ? | 22:18 |
vishy | yes | 22:18 |
*** pvo has joined #openstack | 22:18 | |
vvuksan1 | but then I would have to have different projects on different hosts ? | 22:18 |
vvuksan1 | so that nova-network can be unique for each host ? | 22:19 |
vishy | no, projects only correlate with networks in vlan mode | 22:19 |
vvuksan1 | but there can only be one nova-network in flatdhcp | 22:19 |
vishy | well you could have more than one, but i would recommend only 1 | 22:20 |
vvuksan1 | how do I tell nova-network to launch dnsmasq foron each host ? | 22:20 |
vishy | it doesn't | 22:20 |
vishy | it only runs on the network host | 22:20 |
vvuksan1 | right | 22:21 |
vvuksan1 | but let's say you have multiple physical machines | 22:21 |
vvuksan1 | you can get IPs assigned for eth0 | 22:21 |
vvuksan1 | you don't care what IP is on br100 as long as floating IP on eth0 gets mapped to it | 22:22 |
vvuksan1 | question is can that be done | 22:22 |
vishy | not with running dnsmasq on multiple hosts | 22:23 |
vvuksan1 | i'm fine with that :-) | 22:23 |
vishy | at least not with some complicated custom networking | 22:23 |
vvuksan1 | i just don't see dnsmasq getting kicked off on host2 | 22:23 |
vishy | is it getting kicked off on host1? | 22:24 |
vvuksan1 | it looks like it | 22:24 |
vvuksan1 | it does | 22:24 |
vishy | so on host2 | 22:24 |
vvuksan1 | and now I see the missing IPs being added on host1 | 22:24 |
vishy | network_host=host1 | 22:24 |
kbringard | shouldn't the uptables rules be forwarding packets to the host running nova-network? | 22:25 |
vishy | and it should work | 22:25 |
vvuksan1 | cause I killed nova-compute on host1 | 22:25 |
kbringard | and I meant iptables | 22:25 |
vishy | which iptables rules? | 22:25 |
kbringard | oh wait, I'm thinking of vlan mode | 22:25 |
kbringard | sorry | 22:25 |
kbringard | ignore me :-) | 22:25 |
vvuksan1 | vishy: that's a bit of problem :-) | 22:27 |
vvuksan1 | vishy: cause that will add IPs on br100 on host1 | 22:27 |
vvuksan1 | and not host2 | 22:27 |
vvuksan1 | and I don't want to be the DHCP server on eth0 | 22:27 |
vishy | vvuksan1: why is that a problem? | 22:27 |
vishy | nova-network is the gateway for all of the vms | 22:28 |
vvuksan1 | right | 22:28 |
vvuksan1 | but shouldn't than network_host be host2 | 22:28 |
vvuksan1 | as that is the machine I want VMs kicked off on ? | 22:28 |
vishy | no | 22:28 |
vishy | network_host is the name of the host that is running nova-network | 22:29 |
vishy | whichever host is running nova-network | 22:29 |
kbringard | so only one host can be running nova-network, right? | 22:29 |
kbringard | one host per availability zone | 22:29 |
vishy | set --network_host on both boxes to that hostname | 22:29 |
vvuksan1 | what if both host1 and host2 are running nova-network ? | 22:29 |
vishy | then you need to give a different network to each host | 22:30 |
vvuksan1 | different dhcp network ? | 22:30 |
vishy | and instances won't be able to talk to each other over private ips unless you set up iptables/routing rules between the hosts | 22:31 |
vvuksan1 | which is fine | 22:31 |
vvuksan1 | they'll talk over floating IPs | 22:31 |
vishy | floating ips may work | 22:31 |
vvuksan1 | how do you assign different networks to different machines ? | 22:32 |
vvuksan1 | additionally regarding floating IPs do you assign a pool of floating IPs on per machine basis ? | 22:32 |
vishy | create two networks | 22:33 |
vishy | change the host in the db | 22:33 |
Ryan_Lane | rlucio: not yet, but I need to wait until the packages include the console services, really | 22:33 |
vishy | vvuksan1: yes to floating ips | 22:33 |
*** proberts has quit IRC | 22:34 | |
vvuksan1 | vishy: this is all starting to make sense now :-) | 22:34 |
Ryan_Lane | though I'm going to see if I can connect to a console using that upstart provided :) | 22:34 |
vishy | nova-manage floating create <host> <range> | 22:34 |
*** laurensell has left #openstack | 22:34 | |
vvuksan1 | vishy: nice | 22:34 |
vvuksan1 | vishy: thanks a lot :-) | 22:35 |
vvuksan1 | i shall write this up | 22:35 |
vishy | yw | 22:35 |
vvuksan1 | vishy: speaking of which, kbringard wrote up his VLAN setup http://wiki.openstack.org/VlanNetworkSetup | 22:35 |
vishy | cool | 22:36 |
kbringard | we are essentially using projects as a way to logically name and tag the broadcast domains netops gives us | 22:36 |
vishy | kbringard: why are you hacking the define instead of just passing --vlan_interface=eth1 in the flagfile or command line? | 22:38 |
*** aryko has joined #openstack | 22:38 | |
*** aryko has joined #openstack | 22:38 | |
*** baldben has joined #openstack | 22:38 | |
*** aryko has quit IRC | 22:38 | |
kbringard | vishy: because I didn't know I could specify vlan_interface on the command line :-) | 22:39 |
vvuksan1 | quick | 22:39 |
vvuksan1 | fix the doc :-) | 22:39 |
vishy | all flags can be set using --xxxx on the command line | 22:39 |
vishy | or in the conf file | 22:39 |
vishy | that is the point of the flags :) | 22:39 |
kbringard | well right, I just didn't realize that was a flag I guess | 22:40 |
kbringard | I need to run, I'll bbl | 22:40 |
*** kbringard has quit IRC | 22:40 | |
vishy | kbringard: might be nice to port that setup to a new manager at some point | 22:40 |
vishy | oops | 22:40 |
vvuksan1 | i'll remind him :-) | 22:41 |
*** TheHutch has quit IRC | 22:44 | |
Ryan_Lane | so, if the instance is running "/sbin/getty -L 115200 ttyS0 xterm", and the qemu process launched it with "-append root=/dev/vda console=ttyS0", I should be able to use "virsh console <domainnumber>" right? | 22:44 |
*** evtoews has joined #openstack | 22:44 | |
evtoews | #openstack-meeting | 22:44 |
Ryan_Lane | I'm getting "No console available for domain" | 22:45 |
soren | vvuksan1: dnsmasq invokes it. | 22:45 |
dendrobates | evtoews: you missed it | 22:45 |
vvuksan1 | soren: vishy was able to explain it | 22:45 |
vvuksan1 | soren: i am using things somewhat atypically | 22:46 |
*** rootwyrm has quit IRC | 22:46 | |
vishy | Ryan_Lane: console stuff is a little tricky | 22:50 |
Ryan_Lane | I'm seeing that :) | 22:50 |
vishy | i had a debug script that modified libvirt.xml to make it work | 22:50 |
Ryan_Lane | yeah | 22:50 |
Ryan_Lane | I'm looking at that | 22:50 |
Ryan_Lane | in tools | 22:50 |
Ryan_Lane | this is defined in my xml: <console type='pty' tty='/dev/pts/2'> | 22:51 |
Ryan_Lane | with source path, and target properly defined | 22:51 |
Ryan_Lane | err | 22:51 |
Ryan_Lane | port | 22:51 |
vishy | you might need to pull out the existing ports in | 22:52 |
vishy | stuff | 22:52 |
Ryan_Lane | eh? | 22:52 |
Ryan_Lane | the other two definitions cause problems with it? | 22:52 |
vishy | i think the additions for ajaxwebconsole | 22:52 |
vishy | may cause issues | 22:52 |
vishy | i would pull out the other serial defs | 22:53 |
Ryan_Lane | that's what I'm trying to get working :) | 22:53 |
vishy | oh | 22:53 |
*** ChanServ sets mode: +v pandemicsyn | 22:53 | |
vishy | yes console doesn't work | 22:53 |
vishy | with both in | 22:53 |
Ryan_Lane | ahhh | 22:53 |
vishy | it isn't able to figure out that there are two | 22:53 |
vishy | that is why ajaxwebconsole uses some sneaky trick to find the char device directly | 22:53 |
Ryan_Lane | oh. so it's either webconsole, or virsh? | 22:53 |
vishy | correct | 22:54 |
vishy | well | 22:54 |
vishy | you can use webconsole + virsh if you get rid of the file console for getconsoleoutput | 22:54 |
Ryan_Lane | eh | 22:54 |
Ryan_Lane | well, that would mean people can't see the console output without actually connecting to a console, right? | 22:54 |
vishy | it is the file based console + the pts console that breaks virsh | 22:54 |
Ryan_Lane | I see | 22:54 |
Ryan_Lane | that's a pain | 22:54 |
vishy | it was the easiest way to get it in | 22:55 |
vishy | alternatively we needed to create a little proxy of somesort that forwards data from the serial port and also logs to a file | 22:55 |
Ryan_Lane | yeah. it's no big deal. I was just going to test using virsh for now, since the packages don't have the web console in them right now :) | 22:56 |
vishy | but it seems to work properly with 2 entries as long as you don't mind virsh console not working | 22:56 |
Ryan_Lane | I can live without virsh console | 22:56 |
Ryan_Lane | is there any way to enable vnc access? | 22:57 |
evtoews | dendrobates: oops. just taking a look through the meeting log now. thx. http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-02-08-21.00.log.html | 22:57 |
vishy | we missed web-console in the release? | 22:57 |
vishy | Ryan_Lane: yessir | 22:57 |
Ryan_Lane | the proxy service, for sure | 22:57 |
*** pvo has quit IRC | 22:57 | |
Ryan_Lane | I was thinking about adding guacamole support at some point | 22:57 |
vishy | you can also add a vnc console in the devices section: | 22:58 |
vishy | # <graphics type='vnc' port='5901' listen='<host_ip>' passwd='<password>'/> | 22:58 |
vishy | i did that when i was testing windows | 22:58 |
Ryan_Lane | could add it as a console driver, now that there is a class for it :) | 22:58 |
Ryan_Lane | guacamole would be useful for windows | 22:58 |
Ryan_Lane | or for ubuntu, if it has X, and such | 22:59 |
* vishy doesn't know guacamole | 22:59 | |
Ryan_Lane | http://guacamole.sourceforge.net/ | 22:59 |
Ryan_Lane | html5 based vnc client | 22:59 |
Ryan_Lane | using canvas | 22:59 |
Ryan_Lane | has a proxy to vnc | 23:00 |
vishy | oh nice, yeah we were talking about doing html5 vnc | 23:00 |
vishy | the problem is that the vnc proxy is xen specific | 23:00 |
Ryan_Lane | may be a little heavy weight, since you need to run a tomcat servlet for each proxied instance | 23:00 |
vishy | we need something that is more general | 23:00 |
Ryan_Lane | guacamole is a proxy too | 23:00 |
vishy | does it have security built in like what we did for ajaxconsole? | 23:01 |
Ryan_Lane | username/password I believe | 23:02 |
Ryan_Lane | can likely make it work | 23:02 |
Ryan_Lane | by setting it in the xml file, and passing it in via url | 23:02 |
*** blueadept has quit IRC | 23:03 | |
Ryan_Lane | username can be generic, then can set the password to the token | 23:03 |
*** joearnold has quit IRC | 23:04 | |
*** opengeard has joined #openstack | 23:05 | |
*** joearnold has joined #openstack | 23:06 | |
*** hub_cap has quit IRC | 23:11 | |
*** reldan has joined #openstack | 23:15 | |
*** evtoews has quit IRC | 23:23 | |
*** Hayati has joined #openstack | 23:28 | |
*** opengeard has quit IRC | 23:34 | |
*** rlucio has quit IRC | 23:35 | |
*** burris has joined #openstack | 23:36 | |
*** rnirmal has quit IRC | 23:36 | |
*** joearnold has quit IRC | 23:37 | |
*** rlucio has joined #openstack | 23:37 | |
*** diegoparrilla has quit IRC | 23:43 | |
*** gondoi has quit IRC | 23:46 | |
*** opengeard has joined #openstack | 23:47 | |
*** reldan has quit IRC | 23:49 | |
*** joearnold has joined #openstack | 23:55 | |
*** johnpur has quit IRC | 23:57 | |
*** reldan has joined #openstack | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!