| *** maplebed has joined #openstack | 00:00 | |
| *** rlucio has quit IRC | 00:01 | |
| *** aimon has joined #openstack | 00:08 | |
| *** jimbaker has quit IRC | 00:13 | |
| *** daleolds has quit IRC | 00:19 | |
| *** Cybodog has quit IRC | 00:19 | |
| *** joearnold has quit IRC | 00:21 | |
| *** rsampaio has joined #openstack | 00:23 | |
| *** jimbaker has joined #openstack | 00:24 | |
| *** jimbaker has quit IRC | 00:34 | |
| *** dubsquared1 has quit IRC | 00:46 | |
| *** jimbaker has joined #openstack | 01:03 | |
| *** pvo has quit IRC | 01:12 | |
| *** ar1 has joined #openstack | 01:15 | |
| *** schisamo has joined #openstack | 01:20 | |
| *** schisamo has quit IRC | 01:28 | |
| *** maplebed has quit IRC | 01:28 | |
| *** maplebed has joined #openstack | 01:29 | |
| *** maplebed has quit IRC | 01:33 | |
| *** burris has joined #openstack | 01:42 | |
| *** jc_smith has quit IRC | 01:56 | |
| *** Cybodog has joined #openstack | 02:06 | |
| *** dragondm has quit IRC | 02:45 | |
| *** dubsquared has quit IRC | 02:52 | |
| *** rsampaio has quit IRC | 02:53 | |
| *** rsampaio has joined #openstack | 02:54 | |
| *** Gibby_away is now known as Gibby | 02:56 | |
| termie | eday: i was hoping to hear more about the status as well, since i was not at the summit i have no idea whether it was discussed | 03:01 |
|---|---|---|
| termie | eday: i put in a few hours of work and got most of it done in that branch, but there are a few hours more to do | 03:02 |
| *** hazmat has quit IRC | 03:04 | |
| eday | termie: we pretty much all agreed with it. eventlet, and choose a daemon lib | 03:18 |
| eday | for the workers it may even make sense to go multi-threaded or mulit-process and don't use eventlet, since signal handling might be easier | 03:19 |
| *** rsampaio has quit IRC | 03:22 | |
| termie | what is a 'daemo lib' ? | 03:35 |
| termie | eday: ^ | 03:35 |
| termie | sorry, daemon lib | 03:35 |
| termie | the point of it was largely to stop daemonizing | 03:36 |
| eday | oh, I thought it was to standardize on module that manages start/stop/pidfiles/... too | 03:36 |
| eday | since right now it's a mix of twisted and 'daemon' module | 03:37 |
| termie | nah, there are a bunch of those that all do the same things, i just htink we should provide scripts for uipstart and init.d | 03:37 |
| termie | right, we won't provide any of that anymore | 03:37 |
| eday | ok, thats fine | 03:37 |
| termie | that is for system processes | 03:37 |
| termie | to handle | 03:37 |
| eday | so yeah, everyone agreed I think. there may be some notes on the linked etherpad | 03:38 |
| termie | cool, i'll try to find it | 03:39 |
| eday | oh, hmm. not sure where it is :) | 03:39 |
| termie | i wish we were using git where i am familiar with doing complicated things to get an old branch up to date | 03:39 |
| eday | maybe there wasn't one | 03:40 |
| eday | bzr merge lp:nova :) | 03:40 |
| termie | right ;)( | 03:40 |
| eday | gotta run, ttyl | 03:40 |
| termie | doesn't look like anybody has even touched the blueprint | 03:40 |
| termie | cya | 03:40 |
| termie | thanks for the info | 03:41 |
| termie | is there a write up of stuff from the design summit anywhere? | 03:43 |
| termie | i don't see anything listen on openstack.org | 03:43 |
| *** Gibby has quit IRC | 03:53 | |
| *** kashyapc has joined #openstack | 03:56 | |
| *** abecc has quit IRC | 04:05 | |
| *** hornbeck has quit IRC | 04:32 | |
| *** mdomsch has joined #openstack | 04:32 | |
| eday | termie: most blueprints have an etherpad attached to them | 04:33 |
| eday | termie: at least the ones that were discussed | 04:33 |
| *** kashyapc has quit IRC | 04:41 | |
| *** AimanA is now known as HouseAway | 04:51 | |
| *** suchitp has joined #openstack | 04:56 | |
| *** suchitp has left #openstack | 04:56 | |
| *** pvo has joined #openstack | 04:59 | |
| *** ChanServ sets mode: +v pvo | 04:59 | |
| *** kashyapc has joined #openstack | 05:04 | |
| *** hornbeck has joined #openstack | 05:05 | |
| *** ArdRigh has quit IRC | 05:09 | |
| *** Ryan_Lane has joined #openstack | 05:11 | |
| *** hornbeck has quit IRC | 05:14 | |
| *** suchitp has joined #openstack | 05:15 | |
| *** suchitp has left #openstack | 05:15 | |
| *** hazmat has joined #openstack | 05:22 | |
| *** suchitp has joined #openstack | 05:24 | |
| *** suchitp has left #openstack | 05:24 | |
| *** aimon has quit IRC | 05:26 | |
| *** aimon has joined #openstack | 05:26 | |
| mtaylor | soren: ping | 05:27 |
| *** suchitp has joined #openstack | 05:35 | |
| *** suchitp has left #openstack | 05:35 | |
| *** f4m8_ is now known as f4m8 | 05:39 | |
| *** sirp1 has quit IRC | 05:42 | |
| *** suchitp has joined #openstack | 05:45 | |
| *** suchitp has left #openstack | 05:45 | |
| *** aimon has quit IRC | 05:47 | |
| *** aimon has joined #openstack | 05:47 | |
| *** mdomsch has quit IRC | 05:53 | |
| eday | mtaylor: is it possible to get a more verbose tarmac log from the hudson job? mainly I'm looking for which revno it tried to merge, to see if it actually got the latest | 05:54 |
| *** Apurv has joined #openstack | 05:57 | |
| *** ramkrsna has joined #openstack | 06:01 | |
| *** ramkrsna has joined #openstack | 06:01 | |
| openstackhudson | Project nova build #249: SUCCESS in 1 min 0 sec: http://173.203.104.204:8080/job/nova/249/ | 06:10 |
| *** perestre1ka has joined #openstack | 06:26 | |
| *** perestrelka has quit IRC | 06:30 | |
| *** hazmat has quit IRC | 06:30 | |
| *** rchavik has joined #openstack | 06:36 | |
| mtaylor | eday: uh- I think so? | 06:38 |
| *** miclorb has quit IRC | 06:56 | |
| *** almaisan-away is now known as al-maisan | 07:01 | |
| *** notmyname has quit IRC | 07:07 | |
| *** notmyname has joined #openstack | 07:07 | |
| *** ChanServ sets mode: +v notmyname | 07:07 | |
| *** BK_man has quit IRC | 07:14 | |
| *** ArdRigh has joined #openstack | 07:31 | |
| *** pvo has quit IRC | 07:37 | |
| *** allsystemsarego has joined #openstack | 07:47 | |
| *** GasbaKid has joined #openstack | 07:48 | |
| *** ibarrera has joined #openstack | 07:51 | |
| *** guigui has joined #openstack | 07:55 | |
| *** al-maisan is now known as almaisan-away | 08:23 | |
| *** aimon has quit IRC | 08:26 | |
| *** aimon has joined #openstack | 08:26 | |
| *** jc_smith has joined #openstack | 08:40 | |
| *** jc_smith has quit IRC | 08:48 | |
| *** calavera has joined #openstack | 08:48 | |
| soren | mtaylor: wazzup? | 09:00 |
| mtaylor | soren: not much - I was just gonna bug you in to uploading a package for me... | 09:00 |
| soren | eday: the tarmac logs show the revision id. | 09:00 |
| soren | mtaylor: Oh, shite, that. | 09:00 |
| soren | mtaylor: What's the link again? | 09:00 |
| soren | mtaylor: Sorry, it's been a crazy couple of days. | 09:01 |
| mtaylor | soren: lp:~mordred/librabbitmq/debian | 09:01 |
| mtaylor | soren: and no worries ... I know that, you know, folks get busy and stuff :) | 09:02 |
| *** doude has joined #openstack | 09:18 | |
| *** arthurc has joined #openstack | 09:31 | |
| *** Apurv has quit IRC | 09:35 | |
| *** pws has left #openstack | 09:52 | |
| *** burris has quit IRC | 09:54 | |
| *** irahgel1 has joined #openstack | 09:57 | |
| *** burris has joined #openstack | 10:00 | |
| *** zykes- has joined #openstack | 10:04 | |
| *** miclorb_ has joined #openstack | 10:04 | |
| *** ivan has quit IRC | 10:04 | |
| *** miclorb_ has quit IRC | 10:17 | |
| *** almaisan-away is now known as al-maisan | 10:33 | |
| *** ar1 has quit IRC | 10:33 | |
| *** GasbaKid has quit IRC | 10:35 | |
| *** GasbaKid has joined #openstack | 10:37 | |
| *** miclorb_ has joined #openstack | 10:43 | |
| *** miclorb_ has quit IRC | 11:03 | |
| *** guigui has left #openstack | 11:05 | |
| *** al-maisan is now known as almaisan-away | 11:31 | |
| *** ArdRigh has quit IRC | 11:32 | |
| *** aryko has joined #openstack | 12:10 | |
| *** almaisan-away is now known as al-maisan | 12:12 | |
| *** shahid_ has quit IRC | 12:14 | |
| *** ramkrsna has quit IRC | 12:14 | |
| *** alekibango has quit IRC | 12:20 | |
| *** irahgel1 has quit IRC | 12:25 | |
| *** alekibango has joined #openstack | 12:27 | |
| *** ctennis has quit IRC | 12:30 | |
| *** ctennis has joined #openstack | 12:53 | |
| *** guigui1 has joined #openstack | 12:54 | |
| *** guigui1 has left #openstack | 12:54 | |
| *** gaveen has joined #openstack | 12:57 | |
| *** gaveen has joined #openstack | 12:57 | |
| *** jkakar has quit IRC | 12:57 | |
| *** jkakar has joined #openstack | 12:59 | |
| *** arthurc has quit IRC | 13:15 | |
| soren | eday: How familiar are you with Swift's data model? | 13:31 |
| sandywalsh | ah ha ... nova.sh does a 'drop/create database nova' every run. I thought I was doing something wrong that my subsequent runs were failing after a bad instance launch (when not using nova.sh) | 13:41 |
| sandywalsh | <checks to see if there is a bug report on cleaning up db properly after instance create failure> | 13:42 |
| *** pvo has joined #openstack | 13:49 | |
| *** ChanServ sets mode: +v pvo | 13:49 | |
| *** pvo has quit IRC | 13:50 | |
| *** gaveen has quit IRC | 13:52 | |
| *** jero has joined #openstack | 14:00 | |
| *** cbuben has joined #openstack | 14:00 | |
| *** zachbonham has joined #openstack | 14:00 | |
| *** calavera has quit IRC | 14:03 | |
| *** gaveen has joined #openstack | 14:04 | |
| *** pvo has joined #openstack | 14:05 | |
| *** ChanServ sets mode: +v pvo | 14:05 | |
| *** zachbonham has quit IRC | 14:07 | |
| uvirtbot | New bug: #677475 in nova "Database not cleaned up when run-instances fails" [Undecided,New] https://launchpad.net/bugs/677475 | 14:16 |
| *** Podilarius has joined #openstack | 14:19 | |
| sandywalsh | Tip: moving ./.nova-venv up a directory and symlink'ing to it from your bzr branches works with run_tests.sh | 14:28 |
| *** jimbaker has quit IRC | 14:32 | |
| *** alekibango has quit IRC | 14:33 | |
| *** alekibango has joined #openstack | 14:33 | |
| *** pvo- has joined #openstack | 14:39 | |
| cbuben | does nova have AWS-style metadata service, i.e. http://169.254.169.254/latest/meta-data/ ? | 14:45 |
| *** irahgel1 has joined #openstack | 14:46 | |
| *** dovetaildan has quit IRC | 14:46 | |
| cbuben | just started testing, and for example i notice pubkey handling (for example) is being handled by filesystem injection | 14:46 |
| *** sirp1 has joined #openstack | 14:49 | |
| *** al-maisan is now known as almaisan-away | 14:50 | |
| soren | Why is python-greenlet listed in pip-requires? | 14:54 |
| soren | We don't use it (directly). | 14:55 |
| soren | cbuben: Yes, we have the meta-data service. | 14:55 |
| *** almaisan-away is now known as al-maisan | 14:56 | |
| *** mdomsch has joined #openstack | 14:58 | |
| cbuben | soren: thanks, is it in trunk? I can't seem to find where it's bound to, testing with trunk + lp:~vishvananda/nova/fix-flatdhcp | 14:58 |
| *** pikenAFK is now known as piken | 14:59 | |
| zul | soren: its not importing the greenlet module anywhere | 15:00 |
| zul | so maybe it should be dropped | 15:01 |
| *** maplebed has joined #openstack | 15:08 | |
| soren | cbuben: It's been in there since day 1. | 15:08 |
| soren | cbuben: It may require a bit of iptables magic to reach it. | 15:08 |
| *** rsampaio has joined #openstack | 15:09 | |
| cbuben | soren: yeah, i saw some iptables rules related to 169.254.169.254, i'll look further, thanks! | 15:09 |
| soren | cbuben: Sure. | 15:09 |
| *** dubsquared has joined #openstack | 15:11 | |
| *** burris has quit IRC | 15:12 | |
| *** dovetaildan has joined #openstack | 15:14 | |
| *** f4m8 is now known as f4m8_ | 15:15 | |
| sandywalsh | is nosetest supposed to run clean? I'm getting about 1/3 failures | 15:17 |
| *** abecc has joined #openstack | 15:18 | |
| *** abecc has joined #openstack | 15:18 | |
| *** spectorclan has joined #openstack | 15:19 | |
| *** damon__ has joined #openstack | 15:24 | |
| *** aryko has quit IRC | 15:26 | |
| soren | sandywalsh: Like what? | 15:26 |
| *** ivan has joined #openstack | 15:27 | |
| sandywalsh | I'll get a paste, sec | 15:27 |
| *** patri0t_ is now known as patri0t | 15:29 | |
| dabo | Hmmm... still getting the br100 errors, running on the dev labs machine: http://paste.openstack.org/show/168/ | 15:30 |
| dabo | Here's my ifconfig output: http://paste.openstack.org/show/169/ | 15:30 |
| *** eldarnugaev has joined #openstack | 15:33 | |
| soren | dabo: Your networks don't know to use br100. | 15:34 |
| soren | dabo: ...for some reason. | 15:34 |
| soren | dabo: Not sure why it happens, but you can update the db to make it work. | 15:34 |
| dabo | soren: well, it's that "some reason" that I'm trying to figure out | 15:34 |
| dubsquared | is br100 set as the netdevice in mysql for the network? | 15:35 |
| soren | dabo: Ah. | 15:35 |
| soren | dubsquared: Yes. | 15:35 |
| sandywalsh | soren, http://paste.openstack.org/show/170/ | 15:35 |
| soren | dubsquared: Or rather, it isn't. but should be. | 15:35 |
| soren | sandywalsh: How are you running it? | 15:35 |
| sandywalsh | just 'nosetests' | 15:36 |
| sandywalsh | virtualenv | 15:36 |
| soren | Ah. | 15:36 |
| soren | nosetests -w nova/tests/api | 15:36 |
| sandywalsh | but what about all the other failures? | 15:37 |
| soren | sandywalsh: False negatives. | 15:37 |
| sandywalsh | why? | 15:37 |
| soren | sandywalsh: They're the tests that are supposed to be run by run_tests. | 15:37 |
| dabo | dubsquared: Not sure, but does this answer your question? http://paste.openstack.org/show/171/ | 15:38 |
| soren | sandywalsh: The don't work with nose. Haven't bothered to check why. | 15:38 |
| sandywalsh | hmm, that's kinda tricky ... gotta remember what works where :) | 15:38 |
| soren | sandywalsh: Here's what hudson does: | 15:38 |
| soren | Darn it, where did that terminal go? | 15:39 |
| dubsquared | dab: yes; wanted to see what device you had set for 10.0.0.0/28 | 15:39 |
| soren | sandywalsh: pep8 --repeat --show-pep8 --show-source bin/* nova && python run_tests.py && nosetests -w nova/tests/api | 15:40 |
| dubsquared | dabo: can you paste you network interfaces config? | 15:40 |
| soren | dabo: Perhaps you're getting into trouble because both eth0 and br100 are in 10.x.x.x. | 15:40 |
| dubsquared | soren: yeah.. | 15:40 |
| sandywalsh | haha ... ok. So, which is the recommended way to write tests? for nose or testrunner? | 15:40 |
| soren | Testrunner. | 15:41 |
| sandywalsh | we're getting rid of nose? | 15:41 |
| dabo | dubsquared: you mean /etc/network/interfaces?? | 15:41 |
| soren | Don't know. | 15:41 |
| *** czajkowski has joined #openstack | 15:41 | |
| dubsquared | dabo: correct | 15:42 |
| soren | sandywalsh: Quoting Robert Collins: "Don't use nose." | 15:42 |
| sandywalsh | soren, because, if so, I could take on that refactoring first | 15:42 |
| dabo | dubsquared: http://paste.openstack.org/show/172/ | 15:42 |
| sandywalsh | get the openstack api tests to not require nose | 15:42 |
| soren | sandywalsh: When it comes to python and testing... I listen to Robert Collins. | 15:42 |
| dubsquared | dabo: it doesn't look like you have the bridge set up in there | 15:43 |
| sandywalsh | soren, that's fair. I'll add a ticket to remove nose from openstack api tests | 15:43 |
| dabo | dubsquared: I didn't touch that file. All I did was the 'brctl addbr br100' | 15:44 |
| dubsquared | dabo: you should have something similar to this -> http://paste.openstack.org/show/173/ | 15:44 |
| *** BK_man has joined #openstack | 15:45 | |
| dubsquared | im not 100% familiar with the brctl commands…i just edit that file directly and update mysql accordingly | 15:45 |
| dubsquared | (/etc/network/interfaces) | 15:45 |
| *** rchavik has quit IRC | 15:46 | |
| soren | Again, "bridge_ports eth0" doesn't sound like a good idea to me at all. | 15:46 |
| dubsquared | well, depending on the amount of interfaces on the server, how you have traffic routed etc | 15:46 |
| dubsquared | you can make that br100 use the appropriate adpater | 15:47 |
| soren | Yes. | 15:47 |
| dubsquared | that was just for illustrative purposes | 15:47 |
| *** jaypipes has joined #openstack | 15:47 | |
| dubsquared | illustrative* | 15:48 |
| *** daleolds has joined #openstack | 15:50 | |
| *** maplebed has quit IRC | 15:50 | |
| *** aryko has joined #openstack | 15:50 | |
| _cerberus_ | soren: sandywalsh: I believe jaypipes and gundlach were the ones to decide on using nose for said tests. We probably want to figure out why first | 15:52 |
| *** eldarnugaev has quit IRC | 15:53 | |
| dabo | dubsquared: ok, I used your interfaces file, restarted networking, cleaned out the mysql instance info, and re-launched nova-network. Same result. | 15:53 |
| sandywalsh | soren, is this the Richard Collins quote you cited: http://www.advogato.org/person/robertc/diary.html?start=126 | 15:54 |
| *** kashyapc has quit IRC | 15:54 | |
| *** eldarnugaev has joined #openstack | 16:04 | |
| *** rnirmal has joined #openstack | 16:05 | |
| vishy | cbuben: flatdhcp doesn't set up forwarding rules | 16:11 |
| *** dragondm has joined #openstack | 16:13 | |
| dabo | vishy: question on your flat networking file. Should I just do a wholesale replacement of my nova directory instead of grabbing it from LP? | 16:13 |
| vishy | cbubuen, it is probably a pretty simple addition | 16:13 |
| vishy | dabo: i don't understand... | 16:13 |
| dabo | I have the fix-flatdhcp directory. What do I do with it? | 16:14 |
| dabo | it looks like the regular nova stuff | 16:15 |
| vishy | tis | 16:15 |
| vishy | you can change nova.sh to run that dir instead | 16:15 |
| dubsquared | vishy just added that | 16:15 |
| dubsquared | er there he is :D | 16:15 |
| dubsquared | dabo: sorry went afk for a bit | 16:15 |
| dabo | ah - so just change NOVA_DIR to point to that in nova.sh? | 16:16 |
| vishy | yup | 16:16 |
| dabo | kewl - I'll let you know how that works. | 16:16 |
| JordanRinke | morning. | 16:19 |
| soren | sandywalsh: It was face-to-face, but he did mention a blog post on the subject. | 16:21 |
| cbuben | vishy: thanks, i'll take a crack at making it work | 16:24 |
| eday | soren: re: swift's data model, a fair amount. mostly conceptual, havn't looked at much code | 16:26 |
| soren | eday: I'm just thinking about whether we could apply some of the same magic to our data model. | 16:27 |
| soren | eday: They use consistent hashing to figure out where blobs go, and use a similar mechanism to store lists of things.. | 16:29 |
| eday | soren: well, we have natural sharding to our data, so it's a different problem. Doing the ring thing and moving things around wouldn't make as much sense | 16:29 |
| eday | for example, we know compute worker X should be the master for it's shard, and we don't want to move that. We can also use that context to do better routing to get requests there rather than trying to use a hash ring | 16:30 |
| *** ivan has quit IRC | 16:30 | |
| soren | eday: Certainly. | 16:30 |
| pvo | dabo: when installing from http://wiki.openstack.org/InstallFromSource, did you need to add the lvm2 package from apt? | 16:31 |
| soren | eday: I'm not suggesting we adopt the ring concept. | 16:31 |
| soren | eday: Erm.. Can you poing me at the current datamodel proposal? | 16:31 |
| dabo | pvo: I didn't add that package | 16:31 |
| soren | eday: Having a common reference would probably help me explain myself. | 16:31 |
| eday | soren: https://blueprints.launchpad.net/nova/+spec/bexar-distributed-scheduler see spec and linked etherpad from that page | 16:31 |
| soren | pvo: You will need to for nova-volume. | 16:31 |
| dabo | soren: ah - I'm not using nova-volume | 16:32 |
| soren | eday: Etherpad? Not wiki page? | 16:32 |
| *** jkakar has quit IRC | 16:32 | |
| *** aliguori has joined #openstack | 16:33 | |
| pvo | dabo: I updated the doc to include lvm2 | 16:35 |
| soren | eday: Ah, i see the etherpad now. | 16:36 |
| vishy | good catch, added lvm2 to nova.sh | 16:37 |
| *** jkakar has joined #openstack | 16:37 | |
| *** ivan has joined #openstack | 16:37 | |
| dabo | pvo: cool | 16:38 |
| soren | eday: Man, this is hard without a white board. | 16:40 |
| pvo | http://www.skrbl.com/154093269 <- shared whiteboard? | 16:41 |
| *** damon__ has quit IRC | 16:41 | |
| *** damon__ has joined #openstack | 16:41 | |
| soren | pvo: If i draw on my laptop's screen with a sharpie, will you be able to see it? | 16:43 |
| pvo | probably | 16:43 |
| soren | You see it? | 16:43 |
| soren | It's one of my fingers. guess which one. | 16:43 |
| pvo | its a goat, right? | 16:43 |
| soren | :D | 16:43 |
| soren | Actually, the page doesn't even load for me. | 16:44 |
| soren | What was that thing josh used at the summit? | 16:44 |
| soren | doodle! | 16:44 |
| spectorclan | soren: I can open a GoToMeeting session if you like??? | 16:44 |
| soren | Um.. no. | 16:44 |
| soren | spectorclan: Er. | 16:44 |
| soren | spectorclan: I don't know what that is. | 16:45 |
| spectorclan | soren: it allows people to share desktops with each other and has voip built in | 16:45 |
| ttx | spectorclan: sounds scary | 16:45 |
| spectorclan | ttx: works really well | 16:45 |
| spectorclan | just an offer if you need to see each others screens to show something | 16:46 |
| pvo | spectorclan: does it work with linux? | 16:46 |
| spectorclan | pvo: aha, there is a patch but it is 50/50 for it working. Sorry, forgot about that | 16:47 |
| soren | spectorclan: If I understand techonology correctly I don't think my drawing with a sharpie on my laptop screen even shows up in a screenshot. | 16:47 |
| spectorclan | soren: that is true; however I have never tried to Sharpie on the laptop screen :) | 16:47 |
| soren | I suck at drawing with a trackpoint. | 16:47 |
| soren | Or a mouse. | 16:48 |
| soren | Or a sharpie, for that matter. | 16:48 |
| spectorclan | soren: that is what Todd Morey is for | 16:48 |
| soren | That's a "Todd Morey"? :) | 16:48 |
| spectorclan | we should name all graphic issues a ToddMorey, like it | 16:49 |
| pvo | its like a Tom Collins | 16:49 |
| *** ivan has quit IRC | 16:49 | |
| soren | eday: I think what I'm suggesting is coming up with a scheme that lets us know, just by looking at an instance id, which scheduler to ask about it. | 16:49 |
| soren | eday: And similar for other things we need to look up. | 16:51 |
| eday | soren: yeah, that's basically it | 16:51 |
| *** ivan has joined #openstack | 16:51 | |
| soren | eday: Is that what that long spec is all about? | 16:51 |
| soren | eday: Man...I didn't get that out of it at all :) | 16:51 |
| eday | soren: the schedulers would keep a copy of the data (or a subset if/when it gets really large) to do that type of routing | 16:51 |
| vishy | soren: popplet | 16:52 |
| soren | eday: No, no, that's not it. | 16:52 |
| soren | eday: Not based on lookups. Some sort of hashing. | 16:52 |
| vishy | soren: (was the thing that josh used at the summit) | 16:52 |
| eday | soren: why do you need a hash when you can just look up the record in an index? | 16:53 |
| soren | vishy: Great, that's what I was looking for. | 16:53 |
| soren | eday: Because that way, you can get rid of the index. :) | 16:53 |
| soren | eday: My problem with these trains of thought is always that I've stumbled when I've tried to work out how to list things. | 16:53 |
| vishy | soren, eday: I'm out for an hour, but i'm very interested in seeing where this goes | 16:53 |
| soren | vishy: I'm leaving in a few minutes, too, so you won't miss much. | 16:54 |
| eday | soren: a copy of all aggregated host data in one place to answer those questions (see a list, find out which host to route to, ...) | 16:54 |
| soren | eday: Swift solves that elegantly. | 16:54 |
| vishy | if things go downhill, can i suggest a fight to the death with rusty usb drives? | 16:54 |
| soren | vishy: Plastic rusts? | 16:54 |
| soren | vishy: Not on this planet. | 16:54 |
| vishy | soren: the usb connector does | 16:54 |
| soren | vishy: Good point. | 16:54 |
| vishy | which is what you would stab with i assume | 16:55 |
| vishy | :) | 16:55 |
| alekibango | lol | 16:55 |
| soren | vishy: I've never been good at stabbing. | 16:55 |
| vishy | :) | 16:55 |
| *** vishy is now known as vishy-afk | 16:55 | |
| eday | soren: I just see a hash to find that info as being less direct rather than just looking up the instance ID in a indexed table with all other info. | 16:56 |
| eday | soren: we could use a hash-based index? :) | 16:56 |
| alekibango | i would keep it at hand, rotating the disk, like david against goliath | 16:56 |
| uvirtbot | New bug: #677537 in nova "Nova-api requires nova-compute templates in order to run." [Undecided,New] https://launchpad.net/bugs/677537 | 16:56 |
| soren | eday: It just seems that an even better scalable solution to the "avoid a central datastore" problem is to avoid a datastore as much as possible, not just distribute it all over the place. | 16:56 |
| soren | eday: So, the way swift can list things without keeping a central data store is really elegant. | 16:57 |
| *** maplebed has joined #openstack | 16:57 | |
| eday | soren: well, we want those copies at the other layer for more than just routing, for example, 'list my vm's is a lot quicker if you don't need to even hit the message queue | 16:58 |
| notmyname | technically, there are lots of "central data stores" instead of one for everybody. there is one (replicated) data store for each container | 16:58 |
| soren | eday: If you want a list of containes for a given user, the user's name or id or something is looked up in the same fashion as the blobs are. It's hashed, and based on the result of this, it can deduce the X servers that has this list. | 16:58 |
| eday | soren: especially when you don't need to fan out to every compute worker to answer it | 16:58 |
| soren | eday: We could adopt the same strategy. | 16:59 |
| eday | soren: but when we want to list all VMs with their metadata, and we know they are on 100 compute nodes (large customer), we don't want a distributed map/reduce job to 100 compute nodes to answer that | 17:00 |
| eday | soren: if all that info is in a local data store in the scheduler/api, we can answer it right there without touching the network | 17:00 |
| soren | eday: True. | 17:00 |
| eday | soren: so having each compute/volume/network node being the write masters, and bubbling that up through a data aggregation layer lets us solve the HA, security, and scalability problems somewhat elegantly, IMO | 17:02 |
| soren | eday: I still haven't completely understood the updating mechanism, though. | 17:02 |
| eday | soren: the higher up layers see a instance record (ID/host/metadata) and just writes it to the local table for later lookup/routing | 17:03 |
| *** jimbaker has joined #openstack | 17:03 | |
| eday | soren: and modifications are initiated by the relevant compute node | 17:03 |
| eday | soren: and that change is pushed up via the message queue (possibly multiple hops if there are >1 scheduling layers_ | 17:04 |
| soren | eday: ..and this update also results in a message being broadcast about the change? | 17:04 |
| soren | eday: Ok. | 17:04 |
| soren | eday: ...and what if one gets lost? | 17:04 |
| *** aliguori has quit IRC | 17:04 | |
| soren | eday: The upward messages are fine. They'll be handled by the persistent mq, but they have to be broadcast down in the tree again, right? | 17:05 |
| eday | soren: we use a durable message queue, so it shouldn't, but if there is a spectacular failure, we have a record out of date and could initiate a resync, which would say something like "push all events again after time X", or maybe even just "push all events" | 17:05 |
| eday | soren: this is something that will happen when a scheduler restarts/boots up too | 17:05 |
| soren | eday: "events since time x" sounds hard. | 17:06 |
| soren | eday: We'd have to not only keep a record of the current dataset, but of all updates within X hours. | 17:06 |
| eday | soren: for x in 'select * from instances where updated > X': rpc.cast(x) :) | 17:06 |
| soren | For a million workers, that sounds like a lot of information. | 17:06 |
| *** al-maisan is now known as almaisan-away | 17:06 | |
| soren | Well,I have to go to dinner now. I may stop by later. | 17:07 |
| eday | soren: you only need the lastest record, not all events | 17:07 |
| *** aliguori has joined #openstack | 17:08 | |
| *** burris has joined #openstack | 17:09 | |
| *** doude has quit IRC | 17:12 | |
| dabo | vishy-afk: Tried the flat-networking fix, and am getting a nova-compute error related to the br100 bridge: http://paste.openstack.org/show/174/ | 17:17 |
| *** ibarrera has quit IRC | 17:43 | |
| *** sirp1 has quit IRC | 17:45 | |
| *** sirp1 has joined #openstack | 17:46 | |
| *** dendrobates is now known as dendro-afk | 17:46 | |
| *** dendro-afk is now known as dendrobates | 17:47 | |
| *** rlucio has joined #openstack | 17:51 | |
| *** joearnold has joined #openstack | 17:55 | |
| *** hazmat has joined #openstack | 17:56 | |
| *** eldarnugaev has quit IRC | 18:01 | |
| *** rlucio_ has joined #openstack | 18:21 | |
| *** jfluhmann has quit IRC | 18:22 | |
| *** rlucio has quit IRC | 18:23 | |
| *** rlucio_ is now known as rlucio | 18:23 | |
| *** mdomsch has quit IRC | 18:26 | |
| *** spectorclan has quit IRC | 18:26 | |
| vishy-afk | it didn't create br100 automatically? | 18:30 |
| *** aryko has quit IRC | 18:35 | |
| *** vishy-afk is now known as vishy | 18:36 | |
| *** jfluhmann has joined #openstack | 18:36 | |
| *** mmalgeri has joined #openstack | 18:38 | |
| vishy | dabo: it seems like you must be using FlatManager instead of FlatDHCPManager... | 18:38 |
| vishy | or else it would be trying to create the bridge | 18:38 |
| *** spectorclan has joined #openstack | 18:40 | |
| dabo | vishy: that's correct. Didn't know that there was a FlatDHCPManager... | 18:42 |
| *** anotherjesse has joined #openstack | 18:45 | |
| vishy | hehe, yeah that is how you use flatdhcp | 18:46 |
| vishy | :) | 18:46 |
| vishy | flat network does no bridge config for you | 18:47 |
| *** eldarnugaev has joined #openstack | 18:47 | |
| *** jc_smith has joined #openstack | 18:49 | |
| *** anotherjesse has quit IRC | 18:55 | |
| dubsquared | anyone wish to help diagnose an image creation/spin up issue? | 18:56 |
| vishy | dubsquared: details | 18:57 |
| dubsquared | http://paste.openstack.org/show/175/ | 18:58 |
| vishy | 401 | 18:59 |
| vishy | actually probably a 404 | 19:00 |
| vishy | i'm guessing the image doesn't exist | 19:00 |
| vishy | probably decrypting failed or some such | 19:01 |
| vishy | euca-describe-images, does it show as available? | 19:01 |
| *** aliguori has quit IRC | 19:03 | |
| *** anotherjesse has joined #openstack | 19:03 | |
| *** aliguori has joined #openstack | 19:04 | |
| dubsquared | Yeah, it does. | 19:06 |
| dubsquared | root@ubuntu:/opt# euca-describe-images | grep b0h | 19:06 |
| dubsquared | IMAGE ami-b0h0bdqq mybucket/machine.manifest.xml demo available private x86_64 machine ami-8m5jst16 ami-yvfc9pbh | 19:06 |
| piken | ok, I think you guys might laugh. Out side of system deps like libvirt for example. I just used disttools and built a working binary install for nova-api to start which works and has all python deps dist'ed with it using cxfreeze. | 19:08 |
| piken | next step setup.py bdist_rpm :) | 19:09 |
| piken | then getting all components in the setup. | 19:09 |
| piken | lol | 19:09 |
| piken | just made installing nova-api on 3 diff boxes in diff environments easy. lol | 19:09 |
| dabo | vishy: ok, running the FlatDHCPManager helped :) Now when I try to ssh into the instance, I get: http://paste.openstack.org/show/176/ | 19:10 |
| vishy | dabo: do you get a leased ip message in nova-network? | 19:11 |
| *** schisamo has joined #openstack | 19:12 | |
| dabo | vishy: yep: 2010-11-19 19:11:29+0000 [-] (root): DEBUG Leasing IP 10.0.0.2 | 19:12 |
| *** joshuamckenty has joined #openstack | 19:17 | |
| *** timrc_ has joined #openstack | 19:18 | |
| *** timrc_ is now known as timrc | 19:20 | |
| *** mmalgeri has quit IRC | 19:29 | |
| annegentle | ok, with ttx's help I started a page on the wiki summarizing the decisions and themes from the Summit - feel free to add additional notes. http://wiki.openstack.org/BexarDeveloperTopics/Summary | 19:37 |
| *** jonkelly has joined #openstack | 19:44 | |
| *** sirp1 has quit IRC | 19:46 | |
| *** rsampaio has quit IRC | 19:47 | |
| *** aliguori has quit IRC | 19:52 | |
| *** joshuamckenty has quit IRC | 19:56 | |
| *** joshuamckenty has joined #openstack | 19:56 | |
| vishy | dabo: can you paste the output from ifconfig? | 19:58 |
| *** sirp1 has joined #openstack | 19:58 | |
| vishy | dabo: you may need to do a euca-authorize -P tcp -p 22 default | 20:00 |
| dabo | vishy: 'k - lemme try... | 20:01 |
| dabo | aha! That was it! | 20:01 |
| vishy | cool | 20:02 |
| JordanRinke | we need to post some kind of common gotchas or something | 20:02 |
| vishy | dubsquared: i would check your images/ami-b0h0bdqq/ dir for a flat file called image and make sure it is there | 20:02 |
| dabo | I'll add that step to the wiki page | 20:02 |
| JordanRinke | IP injection with AMI tiny | 20:02 |
| *** anotherjesse has quit IRC | 20:02 | |
| JordanRinke | enable icmp and 22 with euca-auth etc | 20:03 |
| *** ctennis has quit IRC | 20:05 | |
| *** aliguori has joined #openstack | 20:18 | |
| *** Ryan_Lane is now known as Ryan_Lane|away | 20:24 | |
| *** gondoi has joined #openstack | 20:24 | |
| *** ctennis has joined #openstack | 20:27 | |
| *** ctennis has joined #openstack | 20:27 | |
| *** ctennis has quit IRC | 20:31 | |
| *** jimbaker has quit IRC | 20:33 | |
| *** dendrobates is now known as dendro-afk | 20:33 | |
| *** dendro-afk is now known as dendrobates | 20:34 | |
| *** anotherjesse has joined #openstack | 20:46 | |
| * hazmat wonders who committed rev 404. " By root <root@ubuntu> on 2010-11-18" | 20:47 | |
| *** adrian_otto has joined #openstack | 20:47 | |
| adrian_otto | I have a developer who wants to join the swift+nova mailing lists, but the "Join Team" button/link seems to be missing from the LP project page. Is that intentional? | 20:48 |
| *** smoser has quit IRC | 20:49 | |
| spectorclan | adrian_otto: looking at it now | 20:51 |
| *** joshuamckenty has quit IRC | 20:52 | |
| *** joshuamckenty has joined #openstack | 20:52 | |
| adrian_otto | tx | 20:53 |
| spectorclan | adrian_otto: I See it at https://launchpad.net/~openstack | 20:53 |
| spectorclan | adrian_otto: I believe you need to be a registered launchpad user | 20:54 |
| *** dendrobates is now known as dendro-afk | 20:56 | |
| * joshuamckenty blinks | 20:59 | |
| spectorclan | adrian_otto: if you have problems send me an email to stephen.spector@openstack.org and I will look at this over the weekend, I am signing off | 21:02 |
| *** spectorclan has quit IRC | 21:02 | |
| *** schisamo has quit IRC | 21:10 | |
| *** aliguori has quit IRC | 21:16 | |
| notmyname | exlt: https://twitter.com/#!/AlTobey/status/5730396792684544 | 21:17 |
| notmyname | exlt: ^ swift works for me, but nova hangs | 21:18 |
| piken | Well all, have a good couple weeks and holiday, I am off on vaca. | 21:22 |
| *** piken has quit IRC | 21:23 | |
| *** schisamo has joined #openstack | 21:29 | |
| *** localhost2 has quit IRC | 21:34 | |
| *** localhost2 has joined #openstack | 21:37 | |
| exlt | notmyname: yes, I'm aware of those problems.. it is due to bzr allowing multiple authors.. I've worked on resolving it, and don't have an elegant solution - I may rip/replace and rewrite history.. https://github.com/termie/git-bzr-ng/issues#issue/10 | 21:43 |
| *** dendro-afk is now known as dendrobates | 21:43 | |
| notmyname | exlt: hmmm...we should all switch to git then ;-) | 21:44 |
| exlt | \o/ | 21:44 |
| vishy | o// | 21:44 |
| *** anotherjesse has quit IRC | 21:44 | |
| vishy | \\o | 21:44 |
| *** anotherjesse has joined #openstack | 21:44 | |
| anotherjesse | is launchpad thinking of integrating with any other systems like git? | 21:45 |
| notmyname | EdwinGrubbs: ^^ ? | 21:45 |
| exlt | highly doubtful.. what would happen to bzr, then? | 21:45 |
| anotherjesse | competition ;) | 21:46 |
| dendrobates | I have asked that in the past, and have been told bzr is too tightly integrated with LP to easily support git | 21:49 |
| *** allsystemsarego has quit IRC | 21:56 | |
| zaitcev | I only use bzr as a transport mechanism. Do cd swift && bzr pull, then immediately tar|tar it elsewhere and diff, patch. Probably somewhat a pain when you are a heavy contributor, but completely unproblematic for a casual reader. | 21:57 |
| EdwinGrubbs | notmyname: no, Launchpad just makes it easy to mirror a git branch in Launchpad, but it is unlikely that it will start hosting git branches. | 21:59 |
| *** localhost2 has quit IRC | 22:00 | |
| *** jdarcy has quit IRC | 22:00 | |
| EdwinGrubbs | anotherjesse: ^^^ | 22:02 |
| anotherjesse | EdwinGrubbs: are the mirrored git branches "first class citizens" -- eg is there anything that stops us from using git for our hacking? | 22:03 |
| jk0 | what's the difference between 'Fix Committed' and 'Fix Released' statuses on LP? | 22:03 |
| EdwinGrubbs | jk0: it depends on the project. Normally, Fix Committed means it's in trunk, whereas as Fix Released means it is released with a version number. | 22:04 |
| jk0 | in trunk as in, after a successful review/merge? | 22:05 |
| openstackhudson | Project nova build #250: SUCCESS in 1 min 26 sec: http://173.203.104.204:8080/job/nova/250/ | 22:05 |
| *** dfg_ has quit IRC | 22:08 | |
| EdwinGrubbs | anotherjesse: A mirrored git branch is actually converted into a bzr branch on Launchpad. It's actually intended more for mirroring the trunk of projects. If you are using git for local development and just want to convert it to bzr to submit a bzr merge proposal, you might be better of with the plugin that allows bzr to import git branches. http://doc.bazaar.canonical.com/plugins/en/git-plugin.html | 22:11 |
| *** gaveen has quit IRC | 22:13 | |
| *** atobey has joined #openstack | 22:18 | |
| *** gaveen has joined #openstack | 22:18 | |
| *** gaveen has quit IRC | 22:20 | |
| *** gaveen has joined #openstack | 22:20 | |
| *** Ryan_Lane|away is now known as Ryan_Lane | 22:21 | |
| *** rlucio has quit IRC | 22:25 | |
| *** ArdRigh has joined #openstack | 22:25 | |
| *** jc_smith has quit IRC | 22:26 | |
| zul | is there a plan to use openvswitch down the road at some point? | 22:26 |
| *** rlucio has joined #openstack | 22:26 | |
| anotherjesse | zul: there was a session about it at the design summit | 22:27 |
| zul | anotherjesse: do you remember what it was called? | 22:27 |
| pvo | zul: we're likely going to use it | 22:28 |
| zul | k | 22:29 |
| anotherjesse | zul: and we are looking at it heavily - vishy - do you remember the session name? | 22:29 |
| *** jfluhmann has quit IRC | 22:30 | |
| *** irahgel1 has left #openstack | 22:30 | |
| pvo | zul: anotherjesse: http://etherpad.openstack.org/0bp9S20wZR | 22:30 |
| zul | pvo: sweet | 22:31 |
| *** jc_smith has joined #openstack | 22:31 | |
| vishy | L2 / L3 Networking | 22:32 |
| *** dendrobates is now known as dendro-afk | 22:38 | |
| *** localhost2 has joined #openstack | 22:54 | |
| hazmat | is there any way in etherpad.openstack.org to browse the summit pads? | 23:00 |
| soren | hazmat: I don't believe there is, no. | 23:01 |
| hazmat | bummer | 23:03 |
| *** damon__ has quit IRC | 23:03 | |
| *** dendro-afk is now known as dendrobates | 23:08 | |
| *** dendrobates is now known as dendro-afk | 23:08 | |
| *** gondoi has quit IRC | 23:09 | |
| *** gaveen has quit IRC | 23:11 | |
| *** pvo has quit IRC | 23:12 | |
| *** pvo- is now known as pvo | 23:12 | |
| rlucio | hazmat: you should still be able to browse them via http://summit.openstack.org/ods-b/ | 23:13 |
| rlucio | sometimes the etherpads are linked directly otherwise they are in each talks associated discussion link | 23:14 |
| hazmat | rlucio, thanks, thats very helpful | 23:16 |
| rlucio | np | 23:16 |
| *** rlucio_ has joined #openstack | 23:21 | |
| *** rnirmal has quit IRC | 23:25 | |
| *** rlucio has quit IRC | 23:25 | |
| *** rlucio_ is now known as rlucio | 23:25 | |
| *** gaveen has joined #openstack | 23:27 | |
| *** Podilarius has left #openstack | 23:28 | |
| *** ArdRigh has quit IRC | 23:32 | |
| tr3buchet | anyone out there? | 23:33 |
| tr3buchet | having this issue trying to start network | 23:33 |
| tr3buchet | http://pastie.org/1312044 | 23:33 |
| anotherjesse | which network model? | 23:33 |
| tr3buchet | default | 23:33 |
| *** aimon has quit IRC | 23:34 | |
| *** aimon has joined #openstack | 23:34 | |
| vishy | hmm | 23:34 |
| vishy | looks like nova-dhcpbridge isn't getting the right flagfile | 23:35 |
| tr3buchet | also, this worked a bit ago.. | 23:35 |
| tr3buchet | i started troubleshooting a permissions error with libvirt, and suddenly this stops working | 23:36 |
| vishy | looks like nova-dhcpbridge is using the wrong db | 23:36 |
| vishy | oh | 23:36 |
| vishy | do you have an old instance running from a previous run? | 23:37 |
| tr3buchet | i dropped the database a while back and reran the nova-manage stuff to get nova.zip | 23:37 |
| vishy | hmm no it looks like the network is in the database | 23:37 |
| *** ArdRigh has joined #openstack | 23:38 | |
| vishy | yeah you have a corrupted db looks like | 23:38 |
| tr3buchet | alright i'll drop it and try again | 23:38 |
| vishy | try virsh list | 23:38 |
| vishy | well hold on | 23:39 |
| vishy | do you have any instances running? | 23:39 |
| vishy | that might be trying to get ips? | 23:39 |
| tr3buchet | no | 23:39 |
| tr3buchet | virsh list is empty | 23:39 |
| vishy | so why is dhcpbridge trying to get network? | 23:39 |
| vishy | hmm | 23:39 |
| tr3buchet | euca-describe-instances shows two in "scheduling" | 23:40 |
| *** dubsquared has quit IRC | 23:40 | |
| tr3buchet | i dropped the DB earlier because I was getting the integrity error on a euca run instances call | 23:40 |
| vishy | somehow you have a network in the db with a host set and no bridge | 23:42 |
| vishy | not quite sure how that happened | 23:43 |
| tr3buchet | beats me | 23:44 |
| vishy | select vlan, bridge, host from networks; | 23:44 |
| vishy | be interesting to see the output of that | 23:44 |
| vishy | perhaps you created networks in flat mode? | 23:45 |
| vishy | that is the only way i can think of that happening | 23:45 |
| *** ctennis has joined #openstack | 23:49 | |
| *** ctennis has joined #openstack | 23:49 | |
| *** anotherjesse has quit IRC | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!