soren | jk0: Something like that. | 00:00 |
---|---|---|
soren | jk0: I don't think it offers bridging. | 00:00 |
*** larstobi has quit IRC | 00:00 | |
jk0 | soren: go to router.soren.com/mode.htm | 00:01 |
jk0 | there should be an option to switch from NAT to Bridge | 00:02 |
fcarsten | soren: Might be called "RFC 1482" | 00:02 |
fcarsten | 1483 that is | 00:02 |
jk0 | (make sure you jot down your auth creds first :)) | 00:02 |
soren | jk0: Uh. Yeah. | 00:02 |
soren | jk0: How'd you know? | 00:02 |
jk0 | know what? | 00:03 |
jk0 | oh and make note of the VCI/VPI settings in case those need to be set explicitly in bridge mode (shouldn't) | 00:04 |
masumotok | soren: sorry I send you merge request many times(since I found I forgot pep8 checking) | 00:04 |
masumotok | vishy: sorry I send you merge request many times(since I found I forgot pep8 checking) | 00:04 |
soren | jk0: How did you know about that mode.htm? | 00:05 |
jk0 | soren: oh, google :) | 00:05 |
soren | Ah. | 00:05 |
Xenith | Note that if you switch your router to bridge mode, you won't be able to switch it back without resetting it. | 00:05 |
jk0 | yeah, that's why jotting down those VCI/VPI settings is critical | 00:06 |
jk0 | that should be all you need, other than auth | 00:06 |
soren | This is looking like a project for a different day. | 00:06 |
soren | :) | 00:06 |
jk0 | or any static IP info if it's not DHCP | 00:06 |
soren | masumotok: np | 00:06 |
*** adiantum has quit IRC | 00:07 | |
soren | vishy: Approved. | 00:07 |
Xenith | Yea, I think your best bet for v6 right now without munging your network is the Sixxs ayiya tunnel. | 00:07 |
jk0 | last I heard there were problems with the sixxxs signup page | 00:08 |
Xenith | Dunno, I've never used them, of course :) | 00:08 |
jk0 | can't say I have either, this is what a friend told me a week or two ago | 00:09 |
jk0 | HE has been good to me for many years | 00:10 |
jk0 | since freebsd4.2 in fact ;) | 00:10 |
Xenith | Glad we can help. | 00:12 |
jk0 | are you with HE? | 00:12 |
*** adiantum has joined #openstack | 00:13 | |
Xenith | Yep | 00:13 |
jk0 | oh wow, nice, I had no idea | 00:13 |
*** AndChat| has joined #openstack | 00:14 | |
*** nati has quit IRC | 00:15 | |
Xenith | I'm looking forward to convincing my company to get involved in OpenStack :) | 00:15 |
jk0 | you should, that would be pretty cool :) | 00:15 |
soren | vishy: Are you going to review https://code.launchpad.net/~nttdata/nova/live-migration/+merge/44940 ? | 00:17 |
openstackhudson | Project nova build #402: SUCCESS in 1 min 25 sec: http://hudson.openstack.org/job/nova/402/ | 00:18 |
openstackhudson | Tarmac: This modifies libvirt to use CoW images instead of raw images. This is much more efficient and allows us to use the snapshotting capabilities available for qcow2 images. It also changes local storage to be a separate drive instead of a separate partition. | 00:18 |
openstackhudson | I'm proposing this branch for review to get feedback. I may have inadvertently broken a few things. Comments and possible issues: | 00:18 |
openstackhudson | 1. I haven't tested the other hypervisors. I may have broken libvirt xen support and uml support with this patch. | 00:18 |
openstackhudson | 2. Is it useful to have a use_cow_images param, or should it just be automatic for qemu/kvm and turned off for everything else. | 00:18 |
openstackhudson | 3. create_image is a large annoying method. I tried to clean it up a bit, but it could probably use a bit more refactoring. | 00:18 |
openstackhudson | 4. disk.py seems to be only used by the hypervisors, so perhaps it should move into virt dir. | 00:18 |
openstackhudson | 5. disk.py/partition() is unused now. Should we leave it in or throw it away? | 00:18 |
openstackhudson | Comments welcome | 00:18 |
Xenith | What are cow images? | 00:22 |
vishy | soren: just did, that took for ever | 00:22 |
vishy | woah, cow merged, nice! | 00:22 |
soren | vishy: Yup. | 00:22 |
vishy | Xenith: CopyOnWrite | 00:22 |
soren | vishy: I don't see your comments on the live-migration mp? | 00:23 |
soren | vishy: I've read through that patch 3 times already. It gets harder each time to stay focused. | 00:24 |
fcarsten | Xenith: http://en.wikipedia.org/wiki/Qcow | 00:25 |
vishy | i just sent | 00:25 |
vishy | t | 00:25 |
vishy | h | 00:25 |
vishy | soren: might take a minute to pop up. I made a ton of comments | 00:25 |
soren | You e-mailed it? | 00:26 |
vishy | soren: i wish i had time to review earlier, there are a lot of minor issues | 00:26 |
soren | Oh, I see it now. | 00:26 |
masumotok | vishy: I got your comments. thank you for reviewing! | 00:27 |
vishy | i really wish lp had line by line reviews a la reitveld | 00:27 |
vishy | msumotok: you're welcome. sorry it took so long, I've been super-busy | 00:27 |
masumotok | vishy: I know. | 00:28 |
vishy | masumotok: by the way, excellent work. Most of the issues my review are just for style and constincency | 00:29 |
vishy | s/constincency/consistency | 00:29 |
Xenith | So with cow images, one can do VPS-like functionality? | 00:30 |
Xenith | Actually, I'm probably just fairly confused on the whole writing to the running image thing. | 00:30 |
Xenith | Not sure if my understanding is at all correct :) | 00:30 |
soren | vishy: Quoting the patch in the review seems the best approach to me. It's the most compatible with e-mail reviewing, too. | 00:31 |
soren | vishy: The worst is when people refer to line numbers. They change every time someone pushes to the branch, so it's useless when looking back through the history. | 00:32 |
vishy | soren: yeah, it is just painful for long reviews. Reitveld would show you diffs on one file at a time and show you comments inline | 00:32 |
vishy | soren: it made it way easier | 00:32 |
soren | vishy: Did it allow you to review stuff by e-mail? | 00:33 |
soren | Otherwise, it's a fail :) | 00:33 |
vishy | soren: yes | 00:34 |
soren | So it'd extract your comments and show them in-line on the web? | 00:34 |
soren | That sounds awesome. | 00:35 |
fcarsten | Xenith: COW e.g. allows you to privision VM images very fast. You don't have to copy the complete image for every VM. You just create a "reference" to the one image and only keep track of changes (write operations) in a VM specific file | 00:35 |
* soren desperately needs sleep | 00:36 | |
fcarsten | Xenith: A COW image supports these "reference and keeping track of write changes". | 00:37 |
Xenith | Ah. | 00:37 |
* vishy thinks soren is a trooper | 00:38 | |
*** dfg_ has quit IRC | 00:39 | |
vishy | Xenith, fcarsten: + Qcow2 supports instance snapshotting and restore inside the image format | 00:39 |
masumotok | vishy: thank you! I'm struggling to fix now.. | 00:39 |
*** dirakx has quit IRC | 00:40 | |
Xenith | Am I correct in my understanding that inside an instance, its basically a ramdisk, so any writes you make are lost? Ignoring the volume and object storage for the moment. | 00:43 |
vishy | Xenith: no changes are saved on to disk | 00:43 |
vishy | if your host crashes, some stuff may be buffered and not make it | 00:43 |
vishy | but the instance on the host should get updates | 00:44 |
fcarsten | vishy: you might want to add a comma ;-) : no, changes are saved to disk. | 00:44 |
Xenith | Right. But changes made to the running instance are not saved if you restart the instance? | 00:44 |
vishy | if you do a virsh destroy <id> and virsh create /path/to/instance/dir/libvirt.xml you will | 00:44 |
vishy | have all of your changes | 00:45 |
vishy | in fact reboot instances basically does that | 00:45 |
sandywalsh | xtoddx, thx ... having a look | 00:45 |
vishy | fcarsten: yes, there was supposed to be a comma there. Typing fast and typing well are orthogonal for me apparently. | 00:45 |
vishy | Xenith: btw, the same is true for amazon as well, they just don't tell you. If you contact support i understand that they can recover crashed instances for you. The best use of cloud would be to not have to worry about recovery though. | 00:47 |
vishy | Xenith: so most people act as if the instances are ephemeral | 00:47 |
Xenith | Yea, just trying to think my way through this. I understand that generally you want to keep changing data on a volume or object storage or DB node | 00:48 |
Xenith | Trying to think if nova can be used in a more VPS-like fashion. | 00:48 |
vishy | Xenith: it can, in fact, the rackspace use-case is more vps | 00:49 |
fcarsten | vishy, Xenith: This may not always be true. E.g. OpenNebula explicitly cleans (ie.e deletes) VM instance specific files (unless especially marked not to) | 00:49 |
vishy | Xenith, fcarsten: yeah, eucalyptus was notorious for randomly deleting instance files | 00:50 |
vishy | on host reboot and such | 00:50 |
Xenith | Yea, we were originally looking at euca, but I kept pushing openstack :) | 00:50 |
vishy | Xenith: \o/ | 00:51 |
Xenith | vishy: One thing I'd love to do is replacing all our older hosting servers with VPSes. | 00:51 |
Xenith | Because it gets harder to find replacement hardware for them :) | 00:51 |
*** leted has joined #openstack | 00:53 | |
vishy | Xenith: no kidding | 00:54 |
*** dragondm has quit IRC | 00:56 | |
*** maplebed has quit IRC | 00:58 | |
*** Lcfseth has quit IRC | 00:59 | |
*** jfluhmann_ has joined #openstack | 01:00 | |
*** koji-iida has joined #openstack | 01:04 | |
*** trin_cz has quit IRC | 01:06 | |
sandywalsh | termie around? | 01:07 |
*** jdurgin has quit IRC | 01:13 | |
sandywalsh | any bzr experts in the room? | 01:18 |
*** sophiap has quit IRC | 01:19 | |
*** joearnold has quit IRC | 01:29 | |
*** dubsquared1 has quit IRC | 01:30 | |
*** koji-iida has quit IRC | 01:39 | |
*** koji-iida has joined #openstack | 01:44 | |
*** sophiap has joined #openstack | 01:48 | |
Tushar | does any one know how many approvals are needed to merge the branch to trunk lp:nova? | 01:54 |
Tushar | currently for ipv6 branch, only soren has given his approval. | 01:55 |
*** adiantum has quit IRC | 01:56 | |
*** adiantum has joined #openstack | 01:58 | |
Xenith | I believe it is two core approvals | 02:00 |
*** rlucio has quit IRC | 02:04 | |
Tushar | Xenith: Thank you. | 02:08 |
Tushar | Vishy: If you have time , can you please approve IPV6 branch if everything is OK for you. | 02:08 |
vishy | gl hope the merge goes through without changes to the hudson box :) | 02:11 |
Tushar | Vishy : Thank you very much for your approval, | 02:12 |
*** leted has quit IRC | 02:12 | |
*** adiantum has quit IRC | 02:12 | |
*** Tushar has quit IRC | 02:16 | |
*** adiantum has joined #openstack | 02:17 | |
creiht | mtaylor: when you get a moment, can you install netifaces package on hudson? | 02:22 |
termie | sandywalsh: around again | 02:23 |
termie | sandywalsh: i do get backlog if you prefer to just ask your questiosn | 02:23 |
termie | sandywalsh: i can answer them when i read them | 02:23 |
termie | sandywalsh: if you get backlog i already answered some possible questions of yours | 02:23 |
sandywalsh | that's ok, I think I got all my questions figured out | 02:23 |
sandywalsh | was just running into bzr problems | 02:23 |
sandywalsh | didn't get bin/stack on the pull / merge for some reason | 02:24 |
sandywalsh | was wondering how to re-sync the two | 02:24 |
sandywalsh | termie ^ | 02:25 |
*** dirakx has joined #openstack | 02:26 | |
sandywalsh | termie, whoops, I know what the problem is ... ignore. | 02:26 |
vishy | Tushar: looks like there is a merge conflict in nova.sh | 02:27 |
dabo | can someone merge https://code.launchpad.net/~ed-leafe/nova/xs-password-reset/+merge/45537 ? It's got two core approvals and everything's hunky-dory | 02:33 |
*** adiantum has quit IRC | 02:34 | |
koji-iida | vishy: Thank you for your approval of IPv6 branch. We are fixing the conflict problem of nova.sh now. | 02:34 |
*** flipzagging has quit IRC | 02:35 | |
*** adiantum has joined #openstack | 02:38 | |
*** reldan has quit IRC | 02:39 | |
masumotok | vishy: are you there? I just want to ask you about your comment. | 02:48 |
masumotok | vishy: in your comment, "is it really necessary to have a separate host table instead of adding host info to services table?" | 02:48 |
masumotok | vishy: actually, just adding host info to services table is good. | 02:48 |
masumotok | vishy: From other point of view, in nova.api.ec2.admin, there are some code "host_get" "host_get_all" | 02:49 |
masumotok | vishy: those code are expected separate host table. | 02:49 |
masumotok | vishy: so, I think if I just adding host info to services table, other people create host table and adding similar column. | 02:49 |
masumotok | vishy: Therefore, it may be better to choose keeping separate host table, and create relationship Service -> Host. | 02:49 |
masumotok | vishy: Or, if nova.api.ec2.admin is discussing phrase, I remove host table because we are not sure. | 02:49 |
vishy | masumotok: that is fine with me | 02:49 |
masumotok | OK, I keep separate host table. | 02:50 |
masumotok | vishy: thanks! | 02:50 |
vishy | masumotok: I don't see why we need a separate table yet, but we may at some point so I guess it is fine | 02:50 |
*** ivan has quit IRC | 02:50 | |
masumotok | vishy: I am not sure for future release, but I remove separate host table when it obviously unnecessary for anyone. | 02:53 |
*** jdurgin has joined #openstack | 02:55 | |
*** ivan has joined #openstack | 02:56 | |
Hisaharu | I resolved conflict of IPv6 branch. | 02:56 |
vishy | koji-iida: you don't need to repropose the merge after fixing stuff, you can just make a comment | 02:57 |
vishy | i'm away for a bit | 02:57 |
*** pvo_away is now known as pvo | 03:05 | |
koji-iida | vishy: oops, I'm sorry for the inconvenience | 03:05 |
*** fraggeln has quit IRC | 03:17 | |
JordanRinke | anyone have an idea why I would be getting 'sudo: no tty present and no askpass program specified\n' in the nova-compute log when trying to launch an instance? | 03:17 |
JordanRinke | interesting, looks like I am getting permission denied on /dev/kvm as well | 03:22 |
termie | annegentle: can i get the source for openstack.org or something? I would really like this bug with the sidebar to be fixed, makes the whole project look old and/or unprofessional | 03:33 |
*** hadrian has quit IRC | 03:33 | |
*** Ryan_Lane is now known as Ryan_Lane|away | 03:33 | |
*** omidhdl has joined #openstack | 03:34 | |
*** nati has joined #openstack | 03:35 | |
annegentle | termie: oh shoot I meant to ask Todd about that today. It's a Wordpress site, he's moving it to Silverstripe, and I don't have access (in a tiny nutshell). Sorry, I'll check with him again. | 03:35 |
annegentle | termie: ok, email sent to Todd, again I agree, the details matter. | 03:37 |
*** nati has quit IRC | 03:39 | |
*** nati has joined #openstack | 03:40 | |
*** nati has quit IRC | 03:45 | |
creiht | woot! | 03:46 |
creiht | s3 compatability layer is now in swift trunk | 03:46 |
uvirtbot | New bug: #702722 in glance "no tty present and no askpass program specified" [Undecided,Invalid] https://launchpad.net/bugs/702722 | 03:46 |
creiht | it is still a bit experimental, but there if anyone would like to try it out | 03:46 |
*** fraggeln has joined #openstack | 03:48 | |
uvirtbot | New bug: #702723 in nova "Launching instance fails: no tty present and no askpass program specified" [Undecided,New] https://launchpad.net/bugs/702723 | 03:51 |
*** sandywalsh has quit IRC | 03:54 | |
uvirtbot | New bug: #702725 in swift "Add s3 middleware support to swauth" [Low,Triaged] https://launchpad.net/bugs/702725 | 03:57 |
*** jdurgin has quit IRC | 04:00 | |
*** nati has joined #openstack | 04:01 | |
*** mdomsch has joined #openstack | 04:05 | |
*** nati has quit IRC | 04:06 | |
*** nati has joined #openstack | 04:15 | |
*** mdomsch has quit IRC | 04:16 | |
*** nati has quit IRC | 04:16 | |
*** mdomsch has joined #openstack | 04:16 | |
*** nati has joined #openstack | 04:25 | |
*** nati has quit IRC | 04:29 | |
*** adiantum has quit IRC | 04:30 | |
*** nati has joined #openstack | 04:30 | |
*** Facefoxdotcom has quit IRC | 04:34 | |
*** nati has quit IRC | 04:34 | |
*** adiantum has joined #openstack | 04:35 | |
*** nati has joined #openstack | 04:35 | |
nati | soren : Would you please merge ipv6 again? | 04:37 |
nati | soren:Sorry, I know you are superbusy, but I'm afraid of another conflicts. | 04:38 |
*** adiantum has quit IRC | 04:41 | |
*** nati has quit IRC | 04:42 | |
*** nati-1 has joined #openstack | 04:43 | |
openstackhudson | Project nova build #403: SUCCESS in 1 min 29 sec: http://hudson.openstack.org/job/nova/403/ | 04:43 |
openstackhudson | Tarmac: This change introduces support for Sheepdog (distributed block storage | 04:43 |
openstackhudson | system) which is proposed in | 04:43 |
openstackhudson | https://blueprints.launchpad.net/nova/+spec/sheepdog-support | 04:43 |
openstackhudson | Requirements: | 04:43 |
openstackhudson | - libvirt 0.8.7 or later | 04:43 |
openstackhudson | - qemu 0.13.0 or later | 04:43 |
openstackhudson | How to test: | 04:43 |
openstackhudson | 1. install Sheepdog | 04:43 |
openstackhudson | The software is available from SourceForge.net: | 04:43 |
openstackhudson | https://sourceforge.net/projects/sheepdog/files/ | 04:43 |
openstackhudson | See also: | 04:43 |
openstackhudson | http://wiki.qemu.org/Features/Sheepdog/Getting_Started#Install | 04:43 |
openstackhudson | 2. run the sheepdog daemon on each hosts | 04:43 |
openstackhudson | $ sheep /store_dir | 04:43 |
openstackhudson | /store_dir is a directory to store sheepdog objects. The directory | 04:43 |
openstackhudson | must be on the filesystem with an xattr support. | 04:43 |
openstackhudson | 3. format the sheepdog storage | 04:43 |
openstackhudson | $ collie cluster format --copies=3 | 04:43 |
openstackhudson | 4. run nova-volume | 04:43 |
openstackhudson | $ nova-volume --volume_driver=nova.volume.driver.SheepdogDriver | 04:43 |
nati-1 | Thanks again but anothor conflicts :( | 04:45 |
*** adiantum has joined #openstack | 04:47 | |
*** AndChat- has joined #openstack | 04:48 | |
*** nati-1 has quit IRC | 04:49 | |
*** nati-1 has joined #openstack | 04:51 | |
*** AndChat- has quit IRC | 04:52 | |
*** nati-1 has quit IRC | 04:55 | |
*** adiantum has quit IRC | 04:56 | |
*** howardroark has joined #openstack | 04:56 | |
*** sandywalsh has joined #openstack | 05:00 | |
*** nati-1 has joined #openstack | 05:00 | |
nati-1 | soren: sorry again. We fixed conflicts | 05:01 |
nati-1 | Would you please merge? | 05:02 |
koji-iida | soren: yes, I fixed the merge problem of IPv6 branch. | 05:02 |
*** AndChat- has joined #openstack | 05:05 | |
*** nati-1 has quit IRC | 05:06 | |
*** AndChat- has quit IRC | 05:06 | |
*** howardroark has quit IRC | 05:08 | |
*** jt_zg has quit IRC | 05:08 | |
*** adiantum has joined #openstack | 05:08 | |
*** jimbaker has quit IRC | 05:17 | |
*** nati has joined #openstack | 05:17 | |
uvirtbot | New bug: #702741 in nova "nbd device did not show up" [Undecided,New] https://launchpad.net/bugs/702741 | 05:31 |
*** adiantum has quit IRC | 05:33 | |
*** adiantum has joined #openstack | 05:34 | |
*** pvo is now known as pvo_away | 05:38 | |
*** f4m8_ is now known as f4m8 | 05:48 | |
*** f4m8 is now known as f4m8_ | 05:50 | |
nati | vish:Thank you for your approve agein :) | 05:53 |
*** fcarsten has quit IRC | 05:55 | |
*** mdomsch has quit IRC | 05:55 | |
nati | Ooops, Hudson says "ImportError: No module named netaddr " | 05:56 |
nati | We can not fix this problem. | 05:56 |
nati | vish: Sorry, Would you please check it? | 05:56 |
nati | Would you know who have hudson privileges online now? | 06:09 |
nati | mtaylor: I suppose you have hudson privileges. Would you please do apt-get install python-netaddr ? | 06:10 |
mtaylor | nati: I dunno. what're you gonna give me? | 06:11 |
nati | mtaylor: Sorry I misread log. | 06:11 |
mtaylor | nati: you did? did I not want to just install python-netaddr? | 06:12 |
nati | ipv6 branch uses python-netaddr package which not installed in Hudson. | 06:12 |
nati | https://code.launchpad.net/~ntt-pf-lab/nova/ipv6-support/+merge/46220 | 06:12 |
mtaylor | great. well, it's installed now | 06:12 |
nati | mtaylor: Thank you! | 06:12 |
mtaylor | we should probably add it to the deb package depends... | 06:13 |
nati | I added it on nova.sh and pep | 06:13 |
nati | on my branch | 06:13 |
mtaylor | soren: remind me or you do it - add python-netaddr to debian package depends for nova | 06:13 |
mtaylor | nati: we'll need to also add it to the debian packaging, but it's not a huge deal | 06:14 |
nati | mtaylor: I got it. How can I do that? | 06:14 |
nati | Or some privileges needed ? | 06:14 |
*** Ryan_Lane|away has quit IRC | 06:19 | |
*** omidhdl has quit IRC | 06:33 | |
*** miclorb_ has quit IRC | 06:38 | |
*** adiantum has quit IRC | 06:52 | |
*** f4m8_ is now known as f4m8 | 06:53 | |
*** adiantum has joined #openstack | 06:57 | |
anticw | what's the lp dev cycle to push a branch back and ask for review? | 07:02 |
*** adiantum has quit IRC | 07:05 | |
mtaylor | anticw: just push one up ... so : | 07:06 |
mtaylor | anticw: bzr push lp:~yournamehere/nova/branch-name | 07:06 |
*** ibarrera has joined #openstack | 07:07 | |
anticw | yeah, just found that | 07:07 |
anticw | that registers it as well? | 07:07 |
mtaylor | anticw: and then go to it online, such as https://code.edge.launchpad.net/~yournamehere/nova/branch-name ... and click "Propose for merging" | 07:07 |
fraggeln | is swift ready for production-use? | 07:07 |
mtaylor | fraggeln: yes - and is already in production use in fact | 07:08 |
anticw | we have clients with production use of it | 07:08 |
fraggeln | anticw: what type of clients? | 07:08 |
fraggeln | if you are allowed to talk about it that is :) | 07:09 |
anticw | fraggeln: umm... we (cloudscaling) do consulting... client details aren't public yet | 07:09 |
anticw | (maybe some are actually, i dont track this) | 07:09 |
*** guigui1 has joined #openstack | 07:09 | |
*** adiantum has joined #openstack | 07:10 | |
anticw | fraggeln: anyhow, to answer your question (mtalyor did already thouugh), yes it is | 07:10 |
mtaylor | fraggeln: rackspace cloud files is running swift on the backend, AIUI | 07:10 |
anticw | and some people other than rackspace have some fairly reasonable scale as well | 07:11 |
openstackhudson | Project nova build #404: SUCCESS in 1 min 27 sec: http://hudson.openstack.org/job/nova/404/ | 07:13 |
openstackhudson | Tarmac: OpenStack Compute (Nova) IPv4/IPv6 dual stack support | 07:13 |
openstackhudson | http://wiki.openstack.org/BexarIpv6supportReadme | 07:13 |
openstackhudson | Tested with | 07:13 |
openstackhudson | ?unit test | 07:13 |
openstackhudson | ?smoke test | 07:13 |
openstackhudson | No conflict with current branch r 562. | 07:13 |
openstackhudson | Fixed comment by Soren and Vish | 07:13 |
nati | Yappi! ipv6 Merged! Thank you Soren ,Vish, Devin and OpenStackers. | 07:14 |
anticw | mtaylor: what does bzr use for author details on a commit? | 07:16 |
fraggeln | anticw, mtaylor : thanks :) | 07:16 |
anticw | (sorry, i've avoided bzr like the plague with success thus far so am very ignorant) | 07:16 |
mtaylor | anticw: the value from bzr whoami | 07:16 |
anticw | mtaylor: is there a config to change that al la git? | 07:17 |
mtaylor | anticw: it's ok - I've been avoiding git for the same amount of time but just recently had to start using it for something, so I an empathize | 07:17 |
anticw | n, found it | 07:17 |
mtaylor | anticw: run "bzr whoami 'You Name <your@emnail.com>'" | 07:17 |
vishy | natti: you're welcome | 07:18 |
* vishy is very frustrated with bzr's ability to cherry-pick | 07:19 | |
vishy | or lack therof... | 07:19 |
vishy | s/therof/thereof | 07:19 |
mtaylor | vishy: you can cherry pickj | 07:20 |
mtaylor | vishy: bzr merge -mrev1..rev2 other-branch | 07:20 |
mtaylor | vishy: you have to give both a starting and ending rev for it to treat it as a cherry-pick merge | 07:21 |
vishy | yes only it isn't really cherry picking | 07:21 |
anticw | mtaylor: can i diff the branch i just pushed against where i cloned it trivially? | 07:21 |
vishy | it doesn't keep revision information | 07:21 |
vishy | it basically just treats it like you generated a patch file | 07:21 |
mtaylor | vishy: that is true, and that also bugs me | 07:21 |
mtaylor | vishy: it's _slightly_ better than a patch file, as it does at least grok file moves, etc. | 07:22 |
anticw | im not sure about bzr, but some SCMs have to do that | 07:22 |
anticw | because the state is basically a directed graph of hashes, so you can't take things out of the middle robustly without taking ancestors | 07:22 |
mtaylor | anticw: yes | 07:22 |
*** omidhdl has joined #openstack | 07:23 | |
mtaylor | anticw: correct -that's the reason bzr doesn't really do cherry picking - bzr is all centered around the DAG impl | 07:23 |
mtaylor | anticw: but as for your diff question ... there's a couple of ways to go about it | 07:24 |
vishy | mtaylor: it really makes maintaining multiple branches difficult | 07:24 |
anticw | vishy: push back on who created the work and have them recreate patches you can merge directly | 07:25 |
mtaylor | anticw: but what you're looking for is bzr diff ancestor:path/to/branch | 07:25 |
anticw | mtaylor: oh, i mean on the web ui | 07:25 |
anticw | sorry | 07:25 |
mtaylor | anticw: oh. heh. | 07:25 |
mtaylor | anticw: the merge request you generate will generate a diff that shows what will be merged | 07:25 |
anticw | i guess if i do propse it will do that | 07:25 |
mtaylor | yup | 07:25 |
vishy | git is way better at that in particular. It also allows for nice clean branches due to rebasing. And it generally seems to be a little smarter about merges and replaced files. | 07:26 |
mtaylor | vishy: yeah - it's sort of about a workflow... I totally hear your complaint, but I find that the main key is to NEVER mix unrelated work into the same branch | 07:26 |
anticw | git cherry diffs then reapplies though internally | 07:26 |
vishy | mtaylor: that is great but we are basically maintaining three production branches and we need to pull fixes back and forth | 07:27 |
mtaylor | yup. it's certainly not perfect... and not great at back-and-forth | 07:27 |
* mtaylor doesn't particularly put much stock in the git approach to rebase/'clean branch' though | 07:28 | |
mtaylor | but would LOVE if bzr could support cherry picking patches in the workflow you're discussing | 07:28 |
mtaylor | you pretty much have to wind up making a branch that's a GCA of the production branches and then apply your patches to that and pull from that to the places it needs to go... sort of blows | 07:29 |
*** miclorb has joined #openstack | 07:30 | |
* ttx waves | 07:31 | |
mtaylor | hey ttx! | 07:31 |
ttx | let's see if we can get a few more branches in before freezing | 07:32 |
*** joearnold has joined #openstack | 07:35 | |
ttx | mtaylor: would you feel comfortable flipping the Approved switch on a few approved-by-core-people BMPs ? | 07:36 |
mtaylor | ttx: sure | 07:37 |
ttx | https://code.launchpad.net/~termie/nova/db_migration/+merge/46073 looks like a good candidate | 07:37 |
ttx | already went into the Hudson washing machine once | 07:38 |
ttx | and was fixed in consequence | 07:38 |
mtaylor | hehe | 07:38 |
* mtaylor loves hudson | 07:38 | |
ttx | Also https://code.launchpad.net/~ed-leafe/nova/xs-password-reset/+merge/45537 has 4 "accepted", which should be considered sufficient | 07:38 |
mtaylor | ttx: yup. first one looks good. | 07:39 |
ttx | looks like the others will need an small extension. | 07:39 |
ttx | mtaylor: press the button ! | 07:40 |
mtaylor | cool. well those two are flipped | 07:40 |
ttx | yay | 07:40 |
* mtaylor flexes his muscles | 07:40 | |
* mtaylor hurts himself | 07:40 | |
mtaylor | ttx: do we have anywhere online where we list when we're planning on dev meetings being? | 07:40 |
* mtaylor is working on calendar stuff | 07:41 | |
ttx | nice work team on cow and ipv6 ! | 07:41 |
ttx | hmm | 07:41 |
ttx | you mean design summits ? | 07:41 |
mtaylor | yes | 07:43 |
ttx | So the next one is in Santa Clara April 26 - 29, 2011 | 07:43 |
ttx | The one after that is not planned yet | 07:43 |
ttx | Asia, probably Japan | 07:43 |
Xenith | Man, wish I could go to that one. | 07:43 |
mtaylor | ttx: cool. thanks! I have to put these things on my calendar early so I can actually get to them | 07:44 |
ttx | mtaylor: for the date... probably early October | 07:44 |
mtaylor | cool. | 07:44 |
mtaylor | ttx: you don't happen to know the next UDS date do you? | 07:45 |
ttx | but that's still very much up in the air. | 07:45 |
ttx | mtaylor: you can abuse the tentative release schedules to have a relatively good idea. If Mark doesn't come up with a fun version number that requires moving everything two weeks | 07:45 |
ttx | http://wiki.openstack.org/EReleaseSchedule | 07:45 |
mtaylor | hehe | 07:45 |
ttx | https://wiki.ubuntu.com/PReleaseSchedule | 07:46 |
ttx | Places UDS the week of October 27th | 07:46 |
mtaylor | ttx: it was actually UDS-O I was wondering about | 07:47 |
* mtaylor never understands why they only announce them like, 3 months out | 07:47 | |
ttx | Oh ! That's Budapest, probably the week of May 12th | 07:47 |
mtaylor | some of us have to plan our lives a little more than that :) | 07:47 |
mtaylor | oh really? | 07:48 |
* mtaylor never knows anything | 07:48 | |
ttx | You don't know about Oscillating Ocelot, then ? | 07:48 |
openstackhudson | Project nova build #405: SUCCESS in 1 min 25 sec: http://hudson.openstack.org/job/nova/405/ | 07:48 |
openstackhudson | Tarmac: Implements the blueprint for enabling the setting of the root/admin password on an instance. | 07:48 |
openstackhudson | It uses a new xenapi plugin 'agent' that handles communication to/from the agent running on the guest. | 07:48 |
ttx | mtaylor: about the 3 month-notice and the location in the woods: the plan is to prevent people from coming. | 07:49 |
mtaylor | heh. that's funny | 07:49 |
mtaylor | ttx: I know about nothing - sabdfl never tells me anything | 07:50 |
ttx | mtaylor: if you need to know stuff, just bribe Jono, he is cheaper | 07:51 |
mtaylor | wow. I may have a 6 week stretch from late April to mid June where I'm essentially not home | 07:52 |
ttx | mtaylor: really ? | 07:52 |
mtaylor | yeah. I'm too busy | 07:52 |
ttx | mtaylor: maybe you should consider stopping slacking at conferences :P | 07:52 |
mtaylor | ttx: nah. slacking at conferences is too much fun | 07:53 |
ttx | rha, db_migration needs another rinse in the washing machine | 07:54 |
*** fabiand_ has joined #openstack | 07:55 | |
ttx | well, that'll be after freeze, then | 07:55 |
* mtaylor needs to be more involved in code over next cycle... | 07:55 | |
vishy | ttx: final merge of db_migration should be post_freeze, since the last few patches that made it in changed the schema | 07:56 |
ttx | vishy: sounds logical | 07:57 |
vishy | and db_migration should convert austin -> bexar (not bexar - patches) | 07:57 |
* vishy has been staring at his computer screen too long | 07:57 | |
*** brd_from_italy has joined #openstack | 07:58 | |
* mtaylor sticks his head out of vishy's computer screen and floats around | 07:58 | |
*** rcc has joined #openstack | 07:58 | |
* vishy highly recommends vagrant. I've managed to get chef-server working and an install of nova from packages. Working on a multinode chef recipe. I see automated integration testing in our future. | 07:58 | |
vishy | for all you packaging people: python-netaddr needs to be added to depends and qemu-nbd needs to be added to sudoers | 08:00 |
JordanRinke | yo | 08:01 |
vishy | if anyone feels like helping out with the chef stuff ... http://unchainyourbrain.com/openstack/12-testing-nova-openstack-compute-with-vagrant-and-chef | 08:01 |
JordanRinke | oh speaking of chef, have you tried devstructure.com? | 08:01 |
vishy | JordanRinke: any luck with that chardev issue? | 08:01 |
JordanRinke | sandboxes used to create blueprints used to create recipes | 08:01 |
JordanRinke | yeah, i just posted the "fix" | 08:01 |
JordanRinke | not sure where to go with it though | 08:02 |
vishy | JordanRinke: I think i know a guy who works there. Also, I used some of their stuff to build a maverick64 vagrant image. Seems pretty cool | 08:02 |
vishy | JordanRinke: yes i met Richard Crowley | 08:03 |
*** miclorb has quit IRC | 08:05 | |
*** adiantum has quit IRC | 08:05 | |
ttx | ok, Feature Freeze is on. I'll go through the proposals in the pipe and approve the ones that just need the final push. | 08:06 |
vishy | JordanRinke: how did you get raw images? | 08:06 |
*** allsystemsarego has joined #openstack | 08:07 | |
*** allsystemsarego has joined #openstack | 08:07 | |
ttx | soren: <vishy> for all you packaging people: python-netaddr needs to be added to depends and qemu-nbd needs to be added to sudoers | 08:07 |
fabiand_ | I read in a blog post, that objects are immutable in swift. so I wonder if they ever get deleted? | 08:08 |
JordanRinke | They are images from Anso, I rembered seeing raw conversion as part of the compute logs previously so I decided to try it | 08:08 |
JordanRinke | and the virsh command ran | 08:08 |
mtaylor | vishy: point me at anything you get in terms of multi-node testing - I REALLY want to get hudson/tarmac doing that sort of testing (pre-merge if possible) | 08:08 |
vishy | JordanRinke: the new code automatically converts the images to qcow | 08:09 |
mtaylor | ttx: I also pinged soren about that... although I suppose I could just get off my own ass and do it | 08:09 |
vishy | so perhaps kvm doesn't actually want you to specify | 08:09 |
ttx | mtaylor: I could just do it too | 08:09 |
ttx | mtaylor: but I'm kinda busy today | 08:09 |
*** adiantum has joined #openstack | 08:09 | |
vishy | JordanRinke: can you do a qemu-img info /path/to/instance_dir/disk | 08:10 |
*** elasticdog has quit IRC | 08:10 | |
ttx | mtaylor: can you ask for specific reviews to random BMPs in Nova ? | 08:10 |
JordanRinke | yeah, want it here or on launchpad? | 08:10 |
ttx | mtaylor: or do you need to be the proposer for that ? | 08:10 |
vishy | JordanRinke: just tell me if it says that it is a qcow2 image | 08:10 |
mtaylor | ttx: I don't think you need to be the proposer | 08:10 |
mtaylor | ttx: lemme check | 08:10 |
JordanRinke | indeed it does | 08:10 |
vishy | yeah so kvm is R-tarded | 08:11 |
ttx | mtaylor: I thought maybe members of the review team can | 08:11 |
vishy | it actually booted if you changed it to raw? | 08:11 |
mtaylor | ttx: it appears I can request review for branches I did not propose | 08:11 |
ttx | oh, cool. You have 5 minutes ? | 08:11 |
mtaylor | sure | 08:11 |
JordanRinke | root@00000r100020001:/var/lib/nova/instances/instance-00000017# virsh -d 5 create libvirt.xml | 08:11 |
JordanRinke | command: "create libvirt.xml " | 08:11 |
ttx | Go through https://code.launchpad.net/nova/+activereviews and make sure I'm specifically requested on all of them | 08:11 |
JordanRinke | create: file(DATA): libvirt.xml | 08:11 |
ttx | mtaylor: ^ | 08:12 |
JordanRinke | Domain instance-00000017 created from libvirt.xml | 08:12 |
vishy | and can you log in to it? | 08:12 |
ttx | that way I can preapprove without requiring people to file exceptions | 08:12 |
*** drico has quit IRC | 08:12 | |
vishy | that is ssh to its ip? | 08:12 |
*** calavera has joined #openstack | 08:12 | |
ttx | mtaylor: You can add "FFE" to review type | 08:13 |
mtaylor | ttx: k. are you ttx on launchpad? | 08:13 |
ttx | yes | 08:13 |
JordanRinke | no, but I could have my networking screwed up right now | 08:13 |
JordanRinke | I am at home, can you link me nova-debug again? I will see if it is actually booting | 08:14 |
vishy | JordanRinke: i think it isn't booting | 08:14 |
vishy | can you change the line to read <driver name="qemu" type="qcow2"/> | 08:15 |
vishy | and see if it still barfs | 08:15 |
JordanRinke | hmm lsmod for kvm shows a count now, so I think it may have actually booted | 08:15 |
JordanRinke | it was qcow2, i changed it to make it work, I can change it back and confirm though, hold. | 08:16 |
JordanRinke | virsh # list Id Name State | 08:17 |
JordanRinke | ---------------------------------- 16 instance-00000017 running | 08:17 |
mtaylor | ttx: k. done | 08:17 |
ttx | cool, many thanks | 08:17 |
JordanRinke | err, that was while still raw | 08:17 |
JordanRinke | and ahh, i see the diff there with the qemu, hang on | 08:18 |
vishy | JordanRinke: in my testing if i gave it a qcow2 and called it raw it would basically say it booted but not load the drive. You'll probably see a failed to mount root filesystem if you cat console.log with your raw change. | 08:19 |
JordanRinke | failed with qemu/qcow2 | 08:20 |
vishy | bleh | 08:21 |
vishy | JordanRinke: same error? | 08:22 |
JordanRinke | error: operation failed: failed to retrieve chardev info in qemu with 'info chardev' | 08:22 |
JordanRinke | you want a login to this box to look around? | 08:24 |
vishy | yes pls | 08:24 |
mtaylor | ttx: ok. I made a branch adding that depend. I'm not in openstack-ubuntu-packagers, so I submitted a merge req | 08:25 |
mtaylor | soren ^^ | 08:25 |
ttx | mtaylor: will review | 08:25 |
ttx | if soren doesn't beat me to ity | 08:25 |
mtaylor | it doesn't look like you guys are using bzr-builddeb there, so I'm _way_ out of practice with how I'd test :) | 08:26 |
mtaylor | but that's ok | 08:26 |
redbo | mtaylor: can we get python-netifaces on the hudson box when you get a chance? | 08:30 |
mtaylor | redbo: sure. done | 08:31 |
redbo | thanks | 08:31 |
mtaylor | ok. ... the bed is a calling | 08:33 |
*** adiantum has quit IRC | 08:34 | |
*** adiantum has joined #openstack | 08:39 | |
*** taihen has quit IRC | 08:45 | |
*** miclorb has joined #openstack | 08:46 | |
*** fabiand_ has quit IRC | 08:47 | |
*** jfluhmann__ has joined #openstack | 08:55 | |
*** jfluhmann_ has quit IRC | 08:59 | |
*** adiantum has quit IRC | 09:03 | |
uvirtbot | New bug: #702778 in nova "Cannot log into instance when network_manager=FlatManager" [Undecided,New] https://launchpad.net/bugs/702778 | 09:06 |
*** perestre1ka has joined #openstack | 09:06 | |
*** perestrelka has quit IRC | 09:09 | |
*** miclorb has quit IRC | 09:15 | |
*** adiantum has joined #openstack | 09:15 | |
soren | mtaylor: We are using bzr-builddeb there? | 09:19 |
soren | mtaylor: What makes you say we don't? | 09:19 |
*** MarkAtwood has quit IRC | 09:20 | |
soren | mtaylor: Oh, and you are in openstack-ubuntu-packagers. | 09:20 |
soren | mtaylor: nova-core, ubuntu-virt and you. | 09:20 |
soren | mtaylor: At least I thought that was the case. | 09:20 |
soren | mtaylor: Hm... guess not. That's unintentional. | 09:21 |
*** joearnold has quit IRC | 09:22 | |
*** rcc has quit IRC | 09:23 | |
soren | mtaylor: Anyways, branch merged, added it as a runtime dep, too, and building new packages now. | 09:24 |
soren | ttx: Happy feature freeze! | 09:25 |
soren | ttx: Or is it merry feature freeze? I always forget. | 09:26 |
ttx | so far it's not happy or merry, it's just busy. | 09:26 |
soren | Tell me about it. I was up till after 2 last night. | 09:26 |
vishy | soren: | 09:31 |
ttx | vishy is trying to beat you but just fell asleep. | 09:32 |
vishy | so apparently kvm fails to boot with two qcow2 drives | 09:32 |
vishy | :) | 09:32 |
vishy | soren: any idea on that one? it is failing with a permissions error | 09:32 |
vishy | qemu: could not open disk image /var/lib/nova/instances/instance-00000015/local: Permission denied | 09:34 |
vishy | oddly: i can't replicate the error locally | 09:35 |
soren | vishy: How are the two systems different? | 09:37 |
vishy | woah hold up that was weird | 09:37 |
soren | vishy: (i.e. you local system and the one where you see this problem) | 09:37 |
vishy | i tried reversing the order of the two images and it still failed on local | 09:38 |
vishy | mine is running in qemu mode in a vm as opposed to hardware | 09:39 |
soren | Oh. | 09:39 |
soren | Are they otherwise identical? | 09:39 |
vishy | but i tried switching the hardware one to qemu mode and it still fails | 09:39 |
soren | vishy: Anything in dmesg? | 09:40 |
soren | apparmour sort of stuff. | 09:40 |
vishy | audit(1294997948.075:234): apparmor="STATUS" operation="profile_remove" name="libvirt-e956df6e-2621-a872-1d45-679a1e6f36e0" pid=32351 comm="apparmor_parser" | 09:41 |
*** miclorb has joined #openstack | 09:41 | |
soren | Yeah, that's not it. | 09:41 |
vishy | that shows up right after the permission denied in syslog | 09:41 |
soren | The qemu process quits, right? | 09:42 |
vishy | there we go | 09:42 |
vishy | Jan 14 03:38:35 00000r100020001 kernel: [29180.002881] type=1400 audit(1294997915.445:230): apparmor="DENIED" operation="open" parent=28913 profile="/usr/lib/libvirt/virt-aa-helper" name="/var/lib/nova/instances/instance-00000015/local" pid=32144 comm="virt-aa-helper" requested_mask="r" denied_mask="r" fsuid=0 ouid=105 | 09:43 |
soren | Which version of Ubuntu is this? | 09:43 |
soren | Rather, which version of libvirt? | 09:43 |
vishy | why is apparmor being evil | 09:43 |
vishy | it was the default maverick | 09:43 |
vishy | but i upgraded to the ppa version | 09:43 |
ttx | vishy: it's for your own good ! | 09:43 |
soren | vishy: So which one is itnow? | 09:43 |
soren | vishy: 0.8.3-1ubuntu14.1~ppamaverick1 ? | 09:44 |
vishy | libvirt-bin 0.8.3-1ubuntu14.1~ppamaverick1 | 09:44 |
soren | Ok. | 09:44 |
vishy | btw it loads the first image just fine | 09:44 |
vishy | if i comment out local | 09:44 |
vishy | and just load disk it boots | 09:44 |
*** adiantum has quit IRC | 09:45 | |
soren | what the... | 09:45 |
soren | lol | 09:45 |
vishy | apparmor="DENIED" operation="open" parent=1 profile="libvirt-e956df6e-2621-a872-1d45-679a1e6f36e0" name="/var/lib/nova/instances/_base/ami-maverick" | 09:46 |
soren | Ok, give me a minute to sort this out. I know at least why it used to work. | 09:46 |
vishy | looks like it was denied opening the backing file | 09:46 |
soren | Yup. | 09:46 |
*** rcc has joined #openstack | 09:46 | |
*** koji-iida has quit IRC | 09:46 | |
* vishy gives soren a minute | 09:46 | |
soren | another minute, please. | 09:48 |
*** adiantum has joined #openstack | 09:50 | |
soren | bug 696318 | 09:51 |
uvirtbot | Launchpad bug 696318 in libvirt "Starting VMs on qcow2 format with base images in level three fails" [Medium,Confirmed] https://launchpad.net/bugs/696318 | 09:51 |
soren | vishy: Easy workaround: Disable apparmor. | 09:54 |
soren | /etc/libvirt/qemu.conf (look for security_driver) | 09:55 |
vishy | kind of odd that it only fails with two images though | 09:56 |
vishy | testing the disable | 09:56 |
soren | That's not why it fails. | 09:57 |
vishy | oh? | 09:57 |
soren | let me dig a little bit more. | 09:59 |
soren | It's got to do with the name of the file, as it turns out. | 10:01 |
vishy | disabling selinux did work btw | 10:01 |
vishy | o rly? | 10:01 |
soren | apparmor. | 10:01 |
vishy | the name of the backing file or the front file? | 10:01 |
soren | The front. | 10:01 |
soren | /etc/apparmor.d/usr.lib.libvirt.virt-aa-helper | 10:01 |
vishy | rofl, so because i used the name local? | 10:01 |
soren | Well, because you didn't name it disk or something.qcow2 or whatnot. | 10:01 |
soren | Or put it in /var/lib/eucalyptus... :) | 10:02 |
*** MarkAtwood has joined #openstack | 10:02 | |
vishy | lmao | 10:02 |
vishy | strange that disk works | 10:02 |
soren | No. | 10:02 |
soren | /**/disk{,.*} r, | 10:02 |
vishy | so if i name the front file *.qcow2 it will work | 10:02 |
vishy | lmao | 10:03 |
vishy | hilarious | 10:03 |
soren | That's my hypothesis. | 10:03 |
vishy | i think you are right | 10:03 |
vishy | b | 10:03 |
vishy | because local failed both ways | 10:03 |
vishy | now why didn't it crash on my local machine ... | 10:03 |
vishy | because it was in my home dir? | 10:04 |
vishy | @{HOME}/** r | 10:05 |
larissa | vishy: Error: "{HOME}/**" is not a valid command. | 10:05 |
vishy | that is really hilarious | 10:05 |
vishy | ok so i'm going to change local to local.qcow2 | 10:05 |
vishy | sow that it doesn't go boom | 10:05 |
*** fabiand has joined #openstack | 10:10 | |
fabiand | hey. is it correct that objects never get deleted in swift? | 10:10 |
soren | That or we add an exception for nova to the apparmor profile. | 10:11 |
colinnich | fabiand: no, that's not correct - objects are deleted as soon as you command it | 10:11 |
vishy | although that is an annoying name if use_qcow_images is off | 10:11 |
fabiand | colinnich: are they really erased from the disk? | 10:11 |
colinnich | fabiand: yes | 10:11 |
fabiand | colinnich: Because I read about that .tombstone thingy and wondered when those files get deleted .. | 10:11 |
*** miclorb has quit IRC | 10:12 | |
fabiand | colinnich: okay, i will re-read that part then :) | 10:12 |
vishy | will disk.local will match that regex right? | 10:12 |
soren | vishy: Yup. | 10:13 |
vishy | k doing that then | 10:13 |
soren | vishy: It's an extended glob of sorts, not a regex, fwiw. | 10:13 |
vishy | yeah i realized that after i said it but i was hoping you were going to let it slip by | 10:13 |
vishy | :p | 10:13 |
*** adiantum has quit IRC | 10:14 | |
soren | Heh :) | 10:15 |
vishy | works like a charm | 10:15 |
* vishy throttles apparmor | 10:16 | |
*** adiantum has joined #openstack | 10:20 | |
soren | One thing I miss about university is having gigabit Internet connectivity. | 10:21 |
soren | Uploading 80 GB worth of backups takes quite a while on DSL. | 10:22 |
*** elasticdog has joined #openstack | 10:28 | |
*** fabiand_ has joined #openstack | 10:28 | |
vishy | soren: https://code.launchpad.net/~vishvananda/nova/lp702741/+merge/46245 | 10:29 |
vishy | fixed there so we don't have to patch libvirts app-armor profile | 10:29 |
vishy | :) | 10:29 |
vishy | ok time for me to sleep now | 10:30 |
*** Dweezahr has quit IRC | 10:30 | |
*** fabiand has left #openstack | 10:30 | |
soren | vishy: Sleep tight. | 10:30 |
*** dirakx has quit IRC | 10:30 | |
*** Dweezahr has joined #openstack | 10:31 | |
fabiand_ | can I somehow tell swift that rsyncd is running on a different port? | 10:31 |
colinnich | soren: you should install your own swift cluster :-) | 10:35 |
*** allsystemsarego has quit IRC | 10:38 | |
fabiand_ | mh okay, does not seem so ... | 10:38 |
*** hazmat has quit IRC | 10:39 | |
*** allsystemsarego has joined #openstack | 10:41 | |
soren | colinnich: I should do that, too, yes. But I still want off-site backups :) | 10:51 |
colinnich | fabiand_: no, I don't think so. Had a quick look at the code and there doesn't appear to be any configuration there | 10:56 |
*** adiantum has quit IRC | 10:57 | |
*** adiantum has joined #openstack | 10:59 | |
fabiand_ | colinnich: It would be a nice thing in my eyes, so rsyncd could be run from userspace. | 11:03 |
fabiand_ | I thought about ading a key to the replicator config to specify rsyncd's port | 11:03 |
*** allsystemsarego has quit IRC | 11:16 | |
*** MarkAtwood has quit IRC | 11:17 | |
soren | creiht, gholt, letterj: I don't want to step on anyone's toes, so I just want to check with you guys if it's ok if I set up per-commit builds for swift in an identical manner to what we have for nova? | 11:26 |
* soren goes to lunch | 11:27 | |
*** adiantum has quit IRC | 11:28 | |
*** omidhdl has quit IRC | 11:28 | |
*** nati has quit IRC | 11:30 | |
*** adiantum has joined #openstack | 11:34 | |
*** adiantum has quit IRC | 11:42 | |
*** adiantum has joined #openstack | 11:48 | |
soren | Mmm... Lunch. Easily one of my favourite meals of the day. | 11:56 |
*** adiantum has quit IRC | 11:58 | |
*** adiantum has joined #openstack | 12:02 | |
*** rcc has quit IRC | 12:09 | |
*** skrusty has quit IRC | 12:13 | |
*** reldan has joined #openstack | 12:13 | |
*** trin_cz has joined #openstack | 12:14 | |
*** thatsdone has joined #openstack | 12:16 | |
*** ctennis has quit IRC | 12:19 | |
*** adiantum has quit IRC | 12:21 | |
trin_cz | soren: thanks for yesterday's answer. | 12:30 |
*** ctennis has joined #openstack | 12:33 | |
*** reldan has quit IRC | 12:36 | |
*** reldan has joined #openstack | 12:37 | |
*** DubLo7 has quit IRC | 12:38 | |
trin_cz | soren: so I need later than Austin ... which means bzr. Let's go the other way around ... do you know some source of images which include ramdisk? | 12:40 |
ttx | sandywalsh: please check https://code.launchpad.net/~anso/nova/wsgirouter/+merge/45330 : if you're good with it, we'll be able to merge the two paste-deploy branches | 12:41 |
*** reldan has quit IRC | 12:45 | |
soren | trin_cz: You don't need bzr. | 12:45 |
soren | trin_cz: We publish tarballs for every single commit to bzr. | 12:45 |
soren | trin_cz: ..as well as ubuntu packages. | 12:45 |
soren | For Lucid, Maverick, and Natty. | 12:45 |
trin_cz | soren: ok, thanks | 12:47 |
*** omidhdl has joined #openstack | 12:55 | |
*** gaveen has joined #openstack | 12:57 | |
*** gaveen has joined #openstack | 12:57 | |
*** arthurc has joined #openstack | 12:58 | |
sandywalsh | ttx, will do | 13:04 |
trin_cz | soren: I used the ttylinux for now, so I have the ramdisk, but I still cannot run it. euca-run-instances says ECONNREFUSED . Any idea where to start? | 13:04 |
*** drico has joined #openstack | 13:07 | |
soren | Sounds like nova-api isn't running. | 13:08 |
trin_cz | yeah, looks like some recent egg update broke the scripts | 13:08 |
soren | "egg update"? | 13:09 |
trin_cz | I'm still fighting with the correct versions of python eggs ... | 13:15 |
trin_cz | soren: where do I find the bzr snapshot tarballs ... I cannot find them on launchpad | 13:16 |
trin_cz | soren: also, is there a list of python egg versions which a particular nova version expect? | 13:18 |
*** DubLo7 has joined #openstack | 13:18 | |
*** taihen has joined #openstack | 13:20 | |
soren | Which OS are you running? | 13:21 |
trin_cz | openSUSE - I'm trying to package nova, so that it works out of the box. | 13:23 |
*** reldan has joined #openstack | 13:24 | |
trin_cz | soren: I know a bout the package version list at "Getting Started with Nova". But for example euca2ools 1.3 reqiure python-boto >=2.0 , so I have to stick with euca2ools 1.2 | 13:25 |
*** fabiand_ has quit IRC | 13:27 | |
soren | Oh, there's a new euca2ools out? | 13:30 |
soren | That's recent. | 13:30 |
sandywalsh | xtoddx around? | 13:31 |
*** fabiand_ has joined #openstack | 13:32 | |
sandywalsh | xtoddx, nm ... got it :) | 13:32 |
*** DubLo7 has left #openstack | 13:35 | |
*** westmaas has joined #openstack | 13:37 | |
*** jfluhmann__ has quit IRC | 13:39 | |
sandywalsh | ttx, anything else need reviewing? | 13:42 |
ttx | sandywalsh: https://code.launchpad.net/~citrix-openstack/nova/xenapi-glance-2/+merge/45977 could use some XenAPI-aware analysis | 13:43 |
sandywalsh | ttx on it | 13:43 |
ttx | soren: mind flipping the approved switch on wsgirouter ? | 13:44 |
ttx | (or any other nova-core dude ^^) | 13:46 |
trin_cz | soren: what about the nova snapshot tarballs? Packaging will be easier this way | 13:46 |
*** adiantum has joined #openstack | 13:48 | |
*** thatsdone has quit IRC | 13:49 | |
*** pvo_away is now known as pvo | 13:52 | |
creiht | fabiand_, colinnich: re deletes, the file is deleted as soon as the DELETE call is made. a 0-byte tombstone is put in it's place to ensure that an older version of the file doesn't get replicated back (due to failure scenarios) | 13:54 |
creiht | The tombstone will be deleted after reclaimation age has passed, which is 7 days | 13:54 |
fabiand_ | creiht, thanks for this note :) It came up to my mind that the file might get truncated to free space. | 13:55 |
creiht | fabiand_: yeah we don't currently support running rsyncd on a different port, but it probably wouldn't be difficult to add that as an option | 13:55 |
*** aliguori has joined #openstack | 13:55 | |
creiht | soren: sounds good to me :) | 13:55 |
creiht | fabiand_: As an example, say you delete an object, but you had a node down when that delete was issued. When the failed node comes back online, it still has the old copy of the object. Replication could try to push that old copy out to the other nodes, but the newer tombstones on the other nodes will prevent that | 13:58 |
*** arthurc has quit IRC | 13:58 | |
creiht | eventually the other nodes will also replicate the tombstone to the node that had failed, and delete the old object | 13:58 |
*** hadrian has joined #openstack | 13:59 | |
*** calavera has quit IRC | 13:59 | |
*** arthurc has joined #openstack | 13:59 | |
fabiand_ | creith, very nice solution ... in the beginning i was sceptical about the usage of rsync ... being able to specify the rsyncd port might enable the usage of swift without root | 14:00 |
creiht | fabiand_: we keep thinking that we have to replace rsync, but we keep finding ways to make it work effectively :) | 14:01 |
creiht | As to root usage, the services start as root, but then quickly drop priveledges to whatever user you have defined | 14:02 |
creiht | root is needed to bind to port 80 or 443 | 14:02 |
creiht | We also set a couple of rlimits in the kernel (like max file descriptors) on startup | 14:04 |
*** thatsdone has joined #openstack | 14:05 | |
*** jguiditta has joined #openstack | 14:06 | |
*** reldan has quit IRC | 14:06 | |
annegentle | ttx: around? | 14:07 |
sandywalsh | ewan or salvatore ever on here? | 14:07 |
ttx | annegentle: yes | 14:08 |
*** adiantum has quit IRC | 14:10 | |
*** Ryan_Lane has joined #openstack | 14:11 | |
annegentle | ttx: I read with interest the suggestion to move automation scripts to a separate project. Can you help me set one up for documentation? | 14:12 |
*** thatsdon_ has joined #openstack | 14:12 | |
annegentle | ttx: I need a place to check doc files in and out that's "openstack" rather than doc files for each project (nova/swift/glance) | 14:12 |
*** thatsdone has quit IRC | 14:13 | |
ttx | annegentle: we already have a few projects at the top level, like openstack-devel or openstack-common... maybe it can fit there ? what is the documentation about exactly ? | 14:15 |
annegentle | ttx: openstack-common would be perfect. It's a collection of documentation from all three projects as well as scenarios for them working together. | 14:16 |
annegentle | ttx: it's the documentation that will live on docs.openstack.org | 14:16 |
ttx | annegentle: hm, openstack-common is the common library to all projects, so maybe it's not the perfect fit | 14:17 |
annegentle | ttx: I'm updating http://wiki.openstack.org/WebContentRequirements now to add more info about the steps for completion by Bexar | 14:17 |
ttx | annegentle: could you check with jaypipes first ? | 14:18 |
annegentle | ttx: sure, what am I asking him exactly? | 14:18 |
annegentle | ttx: glance has glance.openstack.org set up with RST-based docs | 14:18 |
ttx | annegentle: ask if he had other plans for openstack-common that make posting a doc branch there inappropriate | 14:19 |
sandywalsh | jaypipes around? | 14:19 |
annegentle | ttx: got it. | 14:19 |
soren | trin_cz: What do you mean by "packaging"? | 14:19 |
soren | trin_cz: Oh, sorry, I miseed the bit about opensuse. | 14:19 |
soren | trin_cz: The tarballs are at http://nova.openstack.org/tarballs/ | 14:20 |
*** ppetraki has joined #openstack | 14:21 | |
trin_cz | soren: perfect! thanks | 14:23 |
fabiand_ | may I ask what the status of glance is? | 14:26 |
*** ctennis has quit IRC | 14:31 | |
*** pvo is now known as pvo_away | 14:34 | |
*** Ryan_Lane has quit IRC | 14:39 | |
*** johnpur has quit IRC | 14:39 | |
*** f4m8 is now known as f4m8_ | 14:45 | |
*** gondoi has joined #openstack | 14:47 | |
*** thatsdon_ has quit IRC | 14:48 | |
*** dirakx has joined #openstack | 14:56 | |
*** omidhdl has quit IRC | 14:59 | |
*** zul has joined #openstack | 15:00 | |
*** rcc has joined #openstack | 15:01 | |
annegentle | fabiand_: sure, though jaypipes is the one to ask. You can't install it from packages yet (at least that was the case yesterday), but there are docs at http://glance.openstack.org if you want to read more. | 15:02 |
soren | annegentle: I'm working on packages now. | 15:04 |
soren | Along with roughly 27 other things, so it'll probably be a couple of days. | 15:04 |
soren | But soon! | 15:04 |
*** koji-iida has joined #openstack | 15:05 | |
*** dendrobates is now known as dendro-afk | 15:07 | |
*** rcc has quit IRC | 15:08 | |
sandywalsh | any docs around on getting started with Glance? | 15:11 |
*** trin_cz has quit IRC | 15:15 | |
*** koji-iida has quit IRC | 15:16 | |
annegentle | sandywalsh: me, I'm waiting on the packages since installing seems the right first step. | 15:17 |
sandywalsh | annegentle, thanks. I'm going through the learning process now ... I'll be sure to keep notes for you. | 15:18 |
annegentle | sandywalsh: wonderful, thank you. I'll take anything you write up. | 15:19 |
*** zul has quit IRC | 15:19 | |
*** reldan has joined #openstack | 15:20 | |
*** adiantum has joined #openstack | 15:20 | |
*** dendro-afk is now known as dendrobates | 15:21 | |
*** jfluhmann_ has joined #openstack | 15:34 | |
*** hadrian has quit IRC | 15:38 | |
mtaylor | soren: cool - I didn't see a .bzr-builddeb dir, so I wasn't sure how it worked in that branch | 15:41 |
*** pvo_away is now known as pvo | 15:44 | |
*** reldan has quit IRC | 15:44 | |
*** lvaughn_ has quit IRC | 15:45 | |
*** lvaughn has joined #openstack | 15:45 | |
*** hadrian has joined #openstack | 15:46 | |
*** zul has joined #openstack | 15:48 | |
soren | mtaylor: It doesn't need it. The defaults are fine. | 15:51 |
soren | mtaylor: "bzr bd -S" realises its missing the orig, so it uses get-orig-tarball to grab it, which in turn uses uscan. | 15:51 |
annegentle | mtaylor: I created a man page for nova-manage, but it doesn't look like it's connected in when I run man nova-manage. What's the next step for enabling the man page? | 15:52 |
soren | mtaylor: In short, it just works. | 15:52 |
annegentle | mtaylor: oh and sorry for the "interruption" - was still typing when you started typing. spooky! | 15:52 |
*** hggdh has joined #openstack | 15:55 | |
*** DigitalFlux has joined #openstack | 15:56 | |
*** lvaughn_ has joined #openstack | 15:56 | |
*** dragondm has joined #openstack | 15:57 | |
*** lvaughn has quit IRC | 15:58 | |
jaypipes | fabiand_: working on packages and bug fixes right now...aiming to have a functional beta for Bexar... | 15:59 |
jaypipes | fabiand_: also working on documentation for starting up nodes, etc | 15:59 |
*** MarkAtwood has joined #openstack | 16:02 | |
jaypipes | ttx, annegentle: I'm not following you on the openstack-common thing... | 16:05 |
ttx | jaypipes: anne is looking for a project under which to upload some openstack-wide doc branch | 16:05 |
ttx | jaypipes: should she create a new project or piggyback on one of the existing "common" projects N | 16:06 |
ttx | ? | 16:06 |
ttx | openstack-common or openstack-devel... | 16:06 |
*** jimbaker has joined #openstack | 16:06 | |
jaypipes | ttx: hmm, good question. :) I think openstack-dev may be more appropriate, as the future of openstack-common is not clear at this moment. | 16:08 |
ttx | annegentle: so openstack-devel sounds like a good bet | 16:08 |
jaypipes | ttx: plus, docs that may (eventually) go into openstack-common will likely only be for any shared util libs... | 16:08 |
ttx | right | 16:09 |
annegentle | jaypipes: thanks for thinking it through. | 16:09 |
*** dirakx has quit IRC | 16:09 | |
jaypipes | holy translations, batman. Iida-san was busy last night :) | 16:09 |
annegentle | jaypipes, ttx: one consideration is that these docs may not be Apache-licensed, they may be Creative Commons. Does that consideration change any part of the decision? The licensing is NOT final but something to think about. | 16:10 |
jaypipes | annegentle: shouldn't be an issue... it's just the copyrighter headers in the rst files and in the output templates... | 16:11 |
* jaypipes notes that when creiht talks about tombstones, he just sounds creepy :) | 16:12 | |
*** dirakx has joined #openstack | 16:12 | |
annegentle | jaypipes: ok, yup. thanks | 16:12 |
annegentle | mtaylor: any thoughts on the state of the man page for nova-manage? | 16:15 |
*** rcc has joined #openstack | 16:17 | |
*** guigui1 has quit IRC | 16:18 | |
*** adiantum has quit IRC | 16:21 | |
*** johnpur has joined #openstack | 16:24 | |
*** ChanServ sets mode: +v johnpur | 16:24 | |
*** DigitalFlux has quit IRC | 16:28 | |
*** hggdh has quit IRC | 16:30 | |
*** hggdh has joined #openstack | 16:30 | |
*** DigitalFlux has joined #openstack | 16:31 | |
*** adiantum has joined #openstack | 16:35 | |
ttx | jaypipes: any reason to hold up https://code.launchpad.net/~chris-slicehost/glance/add-s3-backend_updated/+merge/45217 ? | 16:37 |
jaypipes | ttx: one sec...sorry, didn't get around to that one yesterday... | 16:38 |
*** dfg_ has joined #openstack | 16:38 | |
ttx | ok, preapproving ffe just in case you can marge it | 16:38 |
ttx | merge* | 16:38 |
jaypipes | ttx: looks to have merge conflicts again... | 16:39 |
*** reldan has joined #openstack | 16:39 | |
jaypipes | ttx: lemme investigate. | 16:39 |
*** Ryan_Lane has joined #openstack | 16:50 | |
*** DigitalFlux has quit IRC | 16:50 | |
soren | creiht: What is the upcoming version of swift? | 16:58 |
creiht | 1.2 | 16:59 |
*** maplebed has joined #openstack | 16:59 | |
*** brd_from_italy has quit IRC | 16:59 | |
soren | creiht: Alright. When do you plan on bumping the version in the code? | 17:06 |
soren | creiht: In Nova, we did it when Austin was released. | 17:06 |
creiht | yeah we should have done that sooner | 17:06 |
creiht | I will do that soon | 17:06 |
soren | Can you ping me before you do it? | 17:07 |
creiht | sure | 17:07 |
* ttx is about to call it a full week. | 17:07 | |
soren | I need to adjust something in the packaging before you do. | 17:07 |
*** dendrobates is now known as dendro-afk | 17:09 | |
*** ccustine has joined #openstack | 17:11 | |
*** Ryan_Lane_ has joined #openstack | 17:15 | |
*** Ryan_Lane has quit IRC | 17:17 | |
*** Ryan_Lane_ is now known as Ryan_Lane | 17:17 | |
*** dendro-afk is now known as dendrobates | 17:17 | |
*** rlucio has joined #openstack | 17:21 | |
vishy | trin_cz: http://images.ansolabs.com/maverick.tgz | 17:27 |
mtaylor | soren: oh, ok. cool. I'm just so used to import-dsc based setups, the bare debian dir just looked really weird to me | 17:27 |
jk0 | can I get a couple of reviews for https://code.launchpad.net/~jk0/nova/lp702965/+merge/46291 please? | 17:27 |
mtaylor | annegentle: did you add it to the manpage section of config.py? | 17:27 |
annegentle | mtaylor: ah, no. Let me look at that. | 17:28 |
mtaylor | annegentle: I mean conf.py | 17:28 |
mtaylor | annegentle: search for "man_page" | 17:28 |
annegentle | mtaylor: here's the line: http://paste.openstack.org/show/472/ | 17:30 |
uvirtbot | New bug: #702965 in nova "Diagnostics broken in trunk" [Undecided,In progress] https://launchpad.net/bugs/702965 | 17:32 |
mtaylor | annegentle: so, what's the problem? | 17:32 |
*** joearnold has joined #openstack | 17:32 | |
fabiand_ | jaypipes, thanks. | 17:32 |
annegentle | mtaylor: on a box where I have installed nova, I type man nova-manage, and get "No manual entry for nova-manage" in return | 17:35 |
annegentle | mtaylor: I can run nova-manage commands no problem | 17:35 |
fabiand_ | jaypipes, are there thoughts about using some space by using some kind of "cow" technique when pushing images which are based on an already present image? | 17:35 |
mtaylor | annegentle: installed via deb? or via python setup.py install? | 17:35 |
jaypipes | fabiand_: yep, that is already done. The xs-snapshot branch in Nova does that (registers images in Glance as snapshots of XS instances" | 17:37 |
annegentle | installed via add-apt-repository ppa:nova-core/trunk (is that via deb?) | 17:37 |
mtaylor | yes | 17:37 |
mtaylor | ok, that just means that the manpage isn't in the deb packaging | 17:38 |
mtaylor | soren: ^^^ you want me to fix? | 17:38 |
jk0 | ttx: bugfixes can be merged to trunk, right? | 17:38 |
annegentle | mtaylor: ah, ok. Is that a last-minute thing anyway? I do have one edit to make to the man page (remove nova-manage.conf to refer to nova.conf) | 17:38 |
annegentle | mtaylor: gahhh gotta run to lunch before all the good Indian buffet food is gone :) thanks for the help | 17:39 |
mtaylor | annegentle: well, trunk is already creating the manpage, right? | 17:39 |
jk0 | thanks vishy | 17:39 |
mtaylor | annegentle: sure thing | 17:39 |
vishy | aye | 17:39 |
soren | mtaylor: That would be great. I need to go make some dinner. | 17:39 |
mtaylor | soosfarm_: k | 17:39 |
mtaylor | soren: k | 17:39 |
mtaylor | (bad irc client tab completion!) | 17:40 |
vishy | wow there are some bugs in compute api | 17:40 |
vishy | + looks like some extra stuff in cloud | 17:40 |
* vishy goes to clean | 17:40 | |
uvirtbot | New bug: #702973 in nova "Network injection creates an invalid /etc/network/interfaces file if no DNS is specified" [Undecided,New] https://launchpad.net/bugs/702973 | 17:41 |
fabiand_ | jaypipes, very cool :) .. i browsed through the branch ... but could not nail it down ... | 17:45 |
jaypipes | fabiand_: :) | 17:46 |
fabiand_ | i just noted that it is not yet supported for libvirt ;) | 17:47 |
jaypipes | fabiand_: sirp- wrote the code. he'd be a good resource to talk to about it (as would ewanmellor if he's ever around in IRC ;) | 17:47 |
jaypipes | fabiand_: yep, hopefully soon, though :) | 17:47 |
*** jdurgin has joined #openstack | 17:48 | |
openstackhudson | Project nova build #406: SUCCESS in 1 min 29 sec: http://hudson.openstack.org/job/nova/406/ | 17:48 |
openstackhudson | Tarmac: Create and use a generic handler for RPC calls to compute. | 17:48 |
*** Dweezahr has quit IRC | 17:49 | |
*** Dweezahr has joined #openstack | 17:49 | |
*** ctennis has joined #openstack | 17:50 | |
fabiand_ | Is there any compyn from europe involved in openstack? | 17:51 |
*** reldan has quit IRC | 17:52 | |
colinnich | fabiand_: If you include the UK in that, then yes | 17:53 |
creiht | fabiand_: And if you also include Rackspace UK :) | 17:54 |
fabiand_ | colinnich, yep - i meant EMEA .. so even good old britain :) | 17:54 |
colinnich | creiht: yeah yeah :-) | 17:54 |
fabiand_ | creiht, hehe | 17:54 |
creiht | lol | 17:54 |
colinnich | fabiand_: we are in Glasgow, Scotland | 17:54 |
fabiand_ | colinnich: may I ask what company it is? :) | 17:54 |
colinnich | fabiand_: http://www.iomart.com/ | 17:55 |
fabiand_ | colinnich: thanks | 17:55 |
colinnich | fabiand_: our logo's even on http://www.openstack.org/community/ :-) | 17:56 |
fabiand_ | colinnich: yep - I've seen it, I just did not reallize that quick that it is in scotland ... looked for EMEA tlds | 17:57 |
creiht | fabiand_: There were some people from French telecom at the last dev summit as well | 17:58 |
fabiand_ | creiht, ah nice. | 18:00 |
*** Ryan_Lane has quit IRC | 18:00 | |
*** adiantum has quit IRC | 18:01 | |
pvo | fabiand_: Citrix, in Cambridge | 18:05 |
*** dirakx has quit IRC | 18:11 | |
fabiand_ | pvo: ah that company, yes :) | 18:11 |
fabiand_ | thanks everybody. and good evening. :) | 18:12 |
*** maple_bed has joined #openstack | 18:13 | |
*** maplebed has quit IRC | 18:13 | |
*** fabiand_ has quit IRC | 18:13 | |
*** maple_bed has quit IRC | 18:14 | |
*** maplebed has joined #openstack | 18:14 | |
*** dendrobates is now known as dendro-afk | 18:19 | |
*** dendro-afk is now known as dendrobates | 18:20 | |
*** leted has joined #openstack | 18:27 | |
*** joearnold has quit IRC | 18:28 | |
uvirtbot | New bug: #702998 in nova "attach_volume fails due to missing import." [High,In progress] https://launchpad.net/bugs/702998 | 18:31 |
*** jfluhmann__ has joined #openstack | 18:33 | |
*** arthurc has quit IRC | 18:34 | |
*** jfluhmann_ has quit IRC | 18:37 | |
*** glenc has left #openstack | 18:41 | |
*** glenc has joined #openstack | 18:42 | |
*** dendrobates is now known as dendro-afk | 18:47 | |
*** maple_bed has joined #openstack | 18:54 | |
*** maplebed has quit IRC | 18:54 | |
*** nelson__ has quit IRC | 18:57 | |
*** nelson__ has joined #openstack | 18:58 | |
*** maple_bed is now known as maplebed | 19:00 | |
*** hggdh has quit IRC | 19:01 | |
*** ctennis has quit IRC | 19:02 | |
*** ctennis has joined #openstack | 19:06 | |
*** ctennis has joined #openstack | 19:06 | |
*** kpepple has joined #openstack | 19:09 | |
uvirtbot | New bug: #703012 in nova "TrialTestCase breaks running individual tests" [Medium,In progress] https://launchpad.net/bugs/703012 | 19:11 |
uvirtbot | New bug: #703014 in nova "get_console_output should be in compute.api" [Undecided,New] https://launchpad.net/bugs/703014 | 19:11 |
*** dendro-afk is now known as dendrobates | 19:12 | |
eday | vishy: ahh, you beat me to the get_console_output move. should be the last of the rpc module usage in there too now | 19:15 |
vishy | eday: yeah there are crazy bugs i'm dealing with right now | 19:16 |
eday | vishy: what compute.api bugs did you see? just broken imports for attach_volume? | 19:16 |
vishy | running the unit tests individually fail | 19:16 |
eday | ahh | 19:16 |
*** mark_ is now known as mark | 19:18 | |
vishy | ha my bad | 19:18 |
*** dirakx has joined #openstack | 19:20 | |
*** leted has quit IRC | 19:34 | |
ttx | jk0: yes, as many as you can find and fix :) | 19:38 |
jk0 | thanks | 19:38 |
jaypipes | _0x44, sirp-, ttx, dubs: pls see https://code.launchpad.net/~jaypipes/glance/s3-backend/+merge/46317. It is Chris' s3 branch all merged up with trunk and with conflicts resolved...review at will and sirp-, please feel free to set it to Approved when ready. | 19:43 |
sirp- | jaypipes: will do | 19:43 |
jaypipes | cheers :) | 19:43 |
ttx | jaypipes: merge at will ! | 19:44 |
jaypipes | ttx: k, thx :) | 19:46 |
jaypipes | dabo: https://bugs.launchpad.net/nova/+bug/703041 | 19:48 |
uvirtbot | Launchpad bug 703041 in nova "i18n strings with >1 placeholder should always use dictionary, not tuple, replacement" [Low,Confirmed] | 19:48 |
dabo | thx jaypipes | 19:48 |
dabo | I was about to put something like that together after I finished my current review | 19:49 |
jaypipes | dabo: no worries... I'll assign it to myself during cactus some time. | 19:49 |
*** rcc has quit IRC | 19:50 | |
uvirtbot | New bug: #703037 in nova "Instances fail to spawn - Ubuntu 10.10 package install" [Undecided,New] https://launchpad.net/bugs/703037 | 19:52 |
openstackhudson | Project nova build #407: SUCCESS in 1 min 27 sec: http://hudson.openstack.org/job/nova/407/ | 19:53 |
openstackhudson | Tarmac: Re-removes TrialTestCase. It was accidentally added in by some merges and causing issues with running tests individually. | 19:53 |
*** hggdh has joined #openstack | 19:54 | |
uvirtbot | New bug: #703041 in nova "i18n strings with >1 placeholder should always use dictionary, not tuple, replacement" [Low,Confirmed] https://launchpad.net/bugs/703041 | 19:56 |
*** hggdh has quit IRC | 19:59 | |
*** trin_cz has joined #openstack | 19:59 | |
*** jaypipes has quit IRC | 20:02 | |
*** jeffrey_taylor has joined #openstack | 20:03 | |
*** nelson__ has quit IRC | 20:05 | |
*** nelson has joined #openstack | 20:05 | |
*** kpepple has left #openstack | 20:06 | |
devcamcar | vim ../trunk/setup.py | 20:09 |
devcamcar | haw | 20:09 |
*** pvo is now known as pvo_away | 20:11 | |
*** MarkAtwood has quit IRC | 20:12 | |
*** opengeard has quit IRC | 20:12 | |
openstackhudson | Project nova build #408: SUCCESS in 1 min 25 sec: http://hudson.openstack.org/job/nova/408/ | 20:13 |
openstackhudson | Tarmac: This branch fixes two outstanding bugs in compute. It also fixes a bad method signature in network and removes an unused method in cloud. | 20:13 |
*** opengeard has joined #openstack | 20:25 | |
*** whaley is now known as al-- | 20:25 | |
*** al-- is now known as whaley | 20:27 | |
*** MarkAtwood has joined #openstack | 20:35 | |
*** EdwinGrubbs has quit IRC | 20:37 | |
*** fabiand_ has joined #openstack | 20:38 | |
*** leted has joined #openstack | 20:42 | |
*** fabiand_ has quit IRC | 20:50 | |
*** hadrian has quit IRC | 20:55 | |
*** hggdh has joined #openstack | 20:57 | |
jeffrey_taylor | Does it make more sense to have the load-balancers in front of the firewalls and then the proxies behind the firewalls for a Swift deployment across several data-centres? OR, should the proxies sit between the firewalls and load-balancers? | 21:03 |
creiht | jeffrey_taylor: that sounds like a question for your network engineer :) | 21:05 |
*** hadrian has joined #openstack | 21:05 | |
jeffrey_taylor | creiht, that's me :S | 21:05 |
creiht | hehe | 21:05 |
jeremyb | the storage nodes talk to each other not the proxies right? | 21:05 |
jeffrey_taylor | right | 21:05 |
jeremyb | for replication, etc. | 21:05 |
creiht | yeah | 21:05 |
jeffrey_taylor | creiht, I've got a lot on my plate :P | 21:05 |
jeremyb | and other services like container | 21:05 |
creiht | jeffrey_taylor: we really don't have cross-datacenter replication yet in swift | 21:05 |
jeffrey_taylor | creiht, I was going to go the vlan route | 21:06 |
creiht | though it is planned | 21:06 |
jeremyb | creiht: but if you have 5 DCs you could do a zone each | 21:06 |
jeffrey_taylor | creiht, what about in the scenario of one DC. should the proxies be behind the firewalls or behind? | 21:06 |
creiht | so the storage nodes need to be able to talk with each other | 21:06 |
jeremyb | so, i was thinking on the train last night... | 21:06 |
creiht | jeremyb: as long as you have the connectivity, sure | 21:06 |
creiht | and can handle the latency | 21:06 |
creiht | but we haven't done any optimizations for that scenario yet | 21:07 |
jeremyb | if you have multi-region set up, the different regions will be different rings | 21:07 |
creiht | jeffrey_taylor: short answer is that you probably want the firewalls in front of the proxies | 21:07 |
jeffrey_taylor | creiht, thanks. Just wanted to confirm that train of thought :) | 21:08 |
creiht | weather or not that is behind or in front of the lb is more up to you | 21:08 |
jeffrey_taylor | creiht, Ya, I'm really conflicted on that. LB's in the front of the firewalls are fine to a point, but I think we'll need LB's for the proxies once we grow | 21:08 |
creiht | though it seems more logical for the firewall to be outside the lb | 21:09 |
creiht | but I'm not a network expert | 21:09 |
jeremyb | so, if you know you're going to hit more spindles than you ever imagined you could move one spindle at a time to a new ring (in a new region) and eventually it's all moved (i guess you'd have to move the containers or objects too not just spindles. but at least you don't need to double your spindle count overnight | 21:09 |
jeremyb | ) | 21:09 |
jeffrey_taylor | creiht, you make a good point. Those FWs should be getting hammered and even if they are, they should be able to handle it. Back to the drawing board! | 21:10 |
creiht | :) | 21:10 |
*** flipzagging has joined #openstack | 21:10 | |
jeremyb | flipzagging! | 21:10 |
creiht | jeremyb: There are a couple of ways that have been discussed | 21:10 |
jeffrey_taylor | shouldn't* | 21:10 |
flipzagging | jeremyb: hi | 21:10 |
jeremyb | creiht: is multi region cactus? | 21:11 |
creiht | one of the first questions to ask is do you want it cluster wide, or optional on a per container basis | 21:11 |
jeremyb | last i checked the bp said bexar | 21:11 |
creiht | jeremyb: yeah we are going to have to move it to cactus | 21:11 |
jeremyb | :/ | 21:11 |
creiht | It is a hard problem :) | 21:11 |
creiht | And things are busy | 21:11 |
creiht | I think we are currently leaning towards it being optional on a per-container basis | 21:12 |
creiht | so say you have cluster a, and cluster b | 21:12 |
creiht | you could say some container in cluster a, should also have n replicas in cluster b | 21:12 |
jeremyb | what about the rax keeps one region of a private cluster? still no idea if that would be available? | 21:12 |
creiht | I don't know much about private clouds at rax | 21:13 |
creiht | ok | 21:13 |
creiht | wait | 21:13 |
creiht | I see what you are asking now | 21:13 |
creiht | so that is part of the idea | 21:14 |
jeremyb | i mean i run on my own hardware and racks and then rax does one of the regions | 21:14 |
creiht | in theory you have your private cluster, but could say that one container should also have some replicas in rax cloudfiles | 21:14 |
jeremyb | right | 21:14 |
creiht | that has been discussed | 21:14 |
creiht | but no promises :) | 21:14 |
creiht | and I think that would be ideal | 21:15 |
jeremyb | btw, i've wondering and just now remembered... | 21:16 |
jeremyb | are objects stored on disk raw? i.e. can i just less/cat/grep/etc. directly on the xfs path? | 21:16 |
creiht | jeremyb: yes | 21:16 |
jeremyb | iirc that is one of nelson's requirements | 21:16 |
*** pvo_away is now known as pvo | 21:17 | |
jeremyb | creiht: cool. same as i have now except i sometimes have to pipe to `gzip -d` | 21:17 |
creiht | jeremyb: and you can use the tool swift-get-nodes to determine on which nodes and where on the filesystem a given db or file is supposed to be located | 21:17 |
creiht | hehe | 21:17 |
*** westmaas has quit IRC | 21:19 | |
creiht | and swift-object-info will tell you all the information you need about a specific file (where it should be on other nodes, metadata, etc.) | 21:19 |
jeremyb | so it reads xattrs? | 21:20 |
jeremyb | and are xattrs part of the checksum? | 21:21 |
creiht | the xattrs are not part of the checksum | 21:22 |
creiht | and yeah it reads the xattrs | 21:22 |
jeremyb | hrmmmm | 21:23 |
jeremyb | so there's no integrity checking on the xattrs then | 21:24 |
*** EdwinGrubbs has joined #openstack | 21:24 | |
jeremyb | and they're not raided | 21:24 |
creiht | the xattrs are stored on each copy of the object | 21:24 |
jeremyb | right but if the read succeeds you don't know if it's good data or not | 21:25 |
creiht | hrm | 21:25 |
jeremyb | was there much discussion (or even a conscious decision) about not checksumming them? | 21:26 |
jeremyb | also, what's the checksum algo again? | 21:26 |
creiht | the checksum is the md5 hash of the object | 21:26 |
jeremyb | ok, weak but short | 21:26 |
creiht | Do you need something stronger there? | 21:27 |
jeremyb | not atm but maybe at some point | 21:27 |
creiht | It wouldn't be hard to change, if you are paranoid | 21:27 |
creiht | S3 uses it as well | 21:27 |
*** EdwinGrubbs has quit IRC | 21:28 | |
*** EdwinGrubbs has joined #openstack | 21:28 | |
jeremyb | well one scenario i've brought up here before (which I'm not actually implementing right now but may do in the future) is a ring of trusted nodes and a ring of untrusted storage nodes | 21:28 |
jeremyb | so, trusted nodes would do all the services and sign stuff and also some storage | 21:28 |
jeremyb | but there would be a pool of servers that are just storage | 21:29 |
creiht | What problem are you worried about? | 21:29 |
*** ctennis has quit IRC | 21:30 | |
creiht | In general users don't have access to storage nodes | 21:30 |
creiht | but maybe, I don't understand the issue that you are wanting to solve | 21:30 |
*** hhoover has joined #openstack | 21:31 | |
jeremyb | my idea (with no immediate plans) was to have people stick a $300 box with 4 sata bays in their houses or other not entirely secure places. that could be used as one region for swift and then have the other 2 regions in DC racks | 21:32 |
creiht | jeremyb: but back to the metadata, what type of corruption are you worried about in the metadata | 21:32 |
jeremyb | right, metadata | 21:32 |
jeremyb | so, if content can go bad why not metadata? unless maybe xfs provides some protection for the metadata? | 21:33 |
creiht | The main reason for the object checksum is to ensure that the data the user PUTs/GETs is the whole data | 21:33 |
jeremyb | oh, not corruption? | 21:33 |
creiht | It is used secondarily for that as well | 21:33 |
creiht | we have auditors that double check that | 21:33 |
jeremyb | right | 21:34 |
jeremyb | i was thinking checksum the data and make that an xattr and then checksum all xattrs (including the object checksum) and store that in an xattr | 21:34 |
creiht | It also wouldn't be hard to add to the auditors to double check the metadata | 21:34 |
jeremyb | and then just have auditors compare the second checksum | 21:34 |
*** hhoover has left #openstack | 21:35 | |
jeremyb | (was double check the metadata meaning compare to another copy? because aiui they don't even talk to more than one copy at a time now) | 21:36 |
creiht | jeremyb: and back to the public region thing, that is an entirely different problem :) | 21:36 |
creiht | You might find something like tahoefs interesting | 21:36 |
jeremyb | right | 21:36 |
jeremyb | yeah, i looked briefly | 21:37 |
jeremyb | are they still active? | 21:37 |
creiht | jeremyb: yeah I was ensuing that we could add something like that, or have the auditors add the checksum | 21:37 |
creiht | they seem to be active still, but I haven't followed it very closely | 21:38 |
jeremyb | seems they have an update in sep 2010 | 21:38 |
creiht | http://tahoe-lafs.org/trac/tahoe-lafs | 21:38 |
creiht | 1.8.1 was realeased 11-28-10 | 21:39 |
jeremyb | oh, i was at http://tahoe-lafs.org/hacktahoelafs/ | 21:39 |
*** jfluhmann__ has quit IRC | 21:39 | |
jeremyb | also, their correction coding is nice | 21:40 |
*** ctennis has joined #openstack | 21:42 | |
*** ctennis has joined #openstack | 21:42 | |
creiht | We've talked about using reasure-coding in the backend of swift | 21:43 |
jeremyb | i'm planning to have containers sharded on date so a container will only have a short read-write lifespan and then turns r/o when it's "full". then do a final "golden" backup of the r/o container and reduce replica count or even apply correction coding | 21:43 |
creiht | It would be nice to have the same amount of durability, while using less space | 21:43 |
creiht | but it also complicates things quite a bit | 21:43 |
jeremyb | have you seen the hdfs erasure coding? | 21:43 |
creiht | nope | 21:43 |
jeremyb | http://wiki.apache.org/hadoop/HDFS-RAID | 21:44 |
redbo | have you seen hdfs hold more than a couple billion files? me either. | 21:44 |
redbo | :) | 21:44 |
creiht | lol | 21:44 |
jeremyb | no | 21:44 |
jeremyb | of course not! | 21:44 |
jeremyb | redbo: btw, is anything you wrote for the compression tests available for me to use? | 21:45 |
creiht | jeremyb: storage is cheap :) | 21:45 |
jeremyb | can middleware change the container or object name for read and write requests? (ideally before auth checks) | 21:46 |
creiht | jeremyb: if we used erasure coding, you wouldn't be able to go the filesystem and cat out a file :) | 21:46 |
redbo | the compression testing was never anything fancy. https://github.com/redbo/test-storage/blob/master/test-compress | 21:46 |
creiht | jeremyb: middleware can do just about anything :) | 21:46 |
creiht | jeremyb: and actually that is pretty much how the s3 compatibility layer works | 21:47 |
jeremyb | creiht: i figured that would change the cattability | 21:47 |
jeremyb | ahh, cool | 21:47 |
vishy | mtaylor, soren: either of you here? | 21:47 |
creiht | if you do something like erasure coding, then you are dealing with blocks, not files | 21:47 |
vishy | have a question on how hudson does bzr bd | 21:47 |
jeremyb | redbo: ohh, i was thinking you made some middleware or something | 21:49 |
jeremyb | redbo: anyway, that may help, thanks | 21:49 |
creiht | jeremyb: he was using that to try to determine if compression would even buy us anything | 21:49 |
redbo | well, that was just to see how compressable files were | 21:49 |
jeremyb | right, i remember. i just didn't know how you did it | 21:50 |
redbo | It took all night to run on a drive | 21:51 |
jeremyb | just took a spindle from production or what? | 21:52 |
*** pvo is now known as pvo_away | 21:52 | |
*** jguiditta has quit IRC | 21:52 | |
redbo | from your previous conversation, I'm not too worried about metadata being invisibly corrupted. The nice thing about the auditor is making sure there aren't any sectors that die and are never read to find that out. | 21:53 |
redbo | yeah | 21:53 |
mtaylor | vishy: yup | 21:54 |
jeremyb | redbo: so the auditor reads all attrs for everything it checks? | 21:55 |
redbo | yeah, it has to read all the xattrs to unpickle them and get the hash | 21:56 |
jeremyb | haha, it's a pickle?! | 21:56 |
redbo | I can't remember if it's pickle or json | 21:56 |
jeremyb | maybe pickle does some verification for you? | 21:56 |
vishy | mtaylor: so I'm assuming it does the following: cd /trunk && bzr pull && cd /packaging && bzr merge lp:packaging && dch (some options here) && bzr commit && bzr bd --merge=True --export-upstream=/trunk | 21:57 |
vishy | is that about right? | 21:57 |
jeremyb | ok, i'm off for now | 21:57 |
jeremyb | thanks redbo and creiht | 21:57 |
creiht | np | 21:57 |
mtaylor | vishy: pretty much - except the opposite way | 21:57 |
mtaylor | wait -- actually - | 21:58 |
mtaylor | sorry, soren changed things. | 21:58 |
vishy | can you pastie the shell script? | 21:58 |
mtaylor | vishy: it just branches packaging and does bzr bd - the packaging branch only has the packaging - so it grabs the tarball using uscan | 21:58 |
vishy | mtaylor: ah ok, but it must have some magic for finding the current version number to update teh changelog? | 21:59 |
mtaylor | vishy: I'm not sure how soren has it set up, at this point, tbh - the way I usually set this up is closer to what you were writing earlier | 22:00 |
vishy | mtaylor: and I would need to generate the tarball as well so maybe i'll use your method | 22:00 |
tr3buchet | i'm considering the modeling for multi-nic. So far I think adding a virtual_interfaces table to hold the macs for each instance and a public boolean to networks should be all that needs to change. thoughts? | 22:00 |
* mtaylor likes my method better | 22:01 | |
vishy | mtaylor: is there some magic in dch to make a changelog entry from the revision number of the other branch? | 22:01 |
vishy | i suppose I could do it myself in a script, but if there is magic... | 22:01 |
rlucio | hey guys, it looks like the deb build is consistently failing | 22:03 |
rlucio | dh_install: python-nova missing files (usr/lib/python*/dist-packages/*), aborting | 22:03 |
rlucio | for the trunk ppa | 22:03 |
rlucio | here's a link to the last build for lucid: https://launchpad.net/~nova-core/+archive/trunk/+build/2153060 | 22:04 |
mtaylor | vishy: no - I do it in a script -lemme paste you something | 22:05 |
mtaylor | vishy: oh - no, none of my scripts do that specifically | 22:07 |
sirp- | jaypipes: did you see my note RE: image-service-uses-glance? | 22:11 |
vishy | mtaylor: wth, it wants the patchfile backwards? | 22:12 |
vishy | so weird | 22:12 |
mtaylor | vishy: sorry - I have to run out real quick - I can give more of an actual hand with this as soon as I get back | 22:13 |
vishy | mtaylor: np i think i've got this | 22:13 |
uvirtbot | New bug: #703095 in swift "catch_errors middleware needs more verbose logging" [Medium,Triaged] https://launchpad.net/bugs/703095 | 22:16 |
*** pvo_away is now known as pvo | 22:16 | |
soren | mtaylor: I'm here now. | 22:17 |
soren | mtaylor: What's up? | 22:17 |
soren | vishy: Nope, sorry, that's not how we do packaging at all. | 22:19 |
*** troytoman has joined #openstack | 22:19 | |
soren | vishy: Two completely separate branches get stitched together and I add a changelog entry, build a source package, and upload. | 22:20 |
vishy | soren: yeah mtaylor said that you didn't do it that way, do you have dch magic to get the bzr revnumber or some kind of shell script? | 22:20 |
soren | vishy: Ok, so this is how it works: | 22:21 |
soren | vishy: Whenever a new commit lands on trunk, hudson notices. | 22:21 |
soren | vishy: It then builds the nova project. | 22:21 |
soren | vishy: ..which triggers the nova-tarball project... | 22:22 |
soren | vishy: ..which builds a tarball and triggers the nova-ppa project. | 22:22 |
soren | Hang on, I'm going to look it up to make sure I don't lie. | 22:23 |
soren | Ok. | 22:23 |
*** pvo is now known as pvo_away | 22:24 | |
soren | vishy: The nova-ppa project does the following: | 22:24 |
soren | vishy: It grabs the most recent tarball from Hudson, deduces the version from its name. | 22:24 |
soren | vishy: Unpacks the tarball.. | 22:24 |
soren | vishy: ..checks out the packaging branch and slaps it on top. | 22:24 |
soren | vishy: Adds a changelog entry corresponding to the bzr revision and target distribution series. | 22:25 |
soren | vishy: Builds the source package. | 22:25 |
soren | vishy: Uploads. | 22:25 |
soren | vishy: http://paste.openstack.org/show/474/ | 22:26 |
soren | vishy: Those are the gory details. With me so far? | 22:26 |
vishy | ah gets it from the tarball | 22:26 |
vishy | yup | 22:27 |
soren | Yeah. | 22:27 |
soren | Well, you could just do "bzr revno". | 22:27 |
soren | ..but I wanted to actually match what the tarball says. | 22:27 |
vishy | sure | 22:27 |
soren | a) to avoid a race where the something new gets added to trunk in between nova-tarball and nova-ppa run. | 22:27 |
soren | b) to keep the version string generation in one place. | 22:27 |
vishy | what do you pass in to dch? | 22:28 |
soren | dch --force-distribution --v "${pkgversion}" "Automated PPA build." -D $series | 22:28 |
vishy | or do you generate the changelog manually? | 22:28 |
soren | It's all in that pastebin. | 22:28 |
vishy | oh cool | 22:28 |
vishy | t | 22:28 |
vishy | hs | 22:28 |
vishy | x | 22:28 |
soren | There's a couple of reasons why I do it this way. | 22:28 |
vishy | gah, thx that is. enter key got stuck | 22:28 |
* vishy is going to do the simple version to start | 22:29 | |
soren | So, we (OpenStack) may want to consume nova trunk at a different pace than Ubuntu. | 22:29 |
soren | ..but we want to collaborate on the packaging. | 22:29 |
soren | That's easier if they're kept completely separate. | 22:30 |
soren | The reason I do the tarball thing is to make the process of building these snapshots be identical to the process of building packages of final releases. | 22:31 |
soren | building packages of final releases is always based on tarballs. | 22:31 |
soren | I'd hate to be building awesome packages all along, but have the final packages such because some difference in the process caused a problem. | 22:31 |
soren | s/such/suck/ | 22:31 |
*** jdarcy has quit IRC | 22:32 | |
*** ppetraki has quit IRC | 22:34 | |
vishy | soren: so there is no way I can easily match package numbers | 22:37 |
*** Ryan_Lane has joined #openstack | 22:42 | |
*** dendrobates is now known as dendro-afk | 22:43 | |
*** dendro-afk is now known as dendrobates | 22:44 | |
soren | vishy: Whuh? | 22:46 |
vishy | soren: we're building our own packages | 22:47 |
vishy | soren: so our rev numbers will be totally different | 22:47 |
soren | vishy: Ok. | 22:48 |
soren | vishy: Sorry, I'm not sure if you're asking a question or what? | 22:49 |
vishy | soren: no statement | 22:49 |
vishy | soren: i think i just have to live with the fact that our packages won't match :) | 22:50 |
soren | vishy: Do you want them to? | 22:50 |
soren | I may very well be lacking some context here. | 22:50 |
vishy | soren: i guess it doesn't need to | 22:51 |
*** gondoi has quit IRC | 22:53 | |
*** adiantum has joined #openstack | 23:14 | |
*** EdwinGrubbs has quit IRC | 23:17 | |
*** nelson has quit IRC | 23:23 | |
*** adiantum has quit IRC | 23:23 | |
*** nelson has joined #openstack | 23:23 | |
*** EdwinGrubbs has joined #openstack | 23:25 | |
*** adiantum has joined #openstack | 23:28 | |
soren | redbo, letterj, gholt, creiht: Why is it again that all the swift daemons are disabled by default in the packages? | 23:29 |
soren | (pardon if I'm flogging a dead horse) | 23:29 |
*** adiantum has quit IRC | 23:36 | |
*** pvo_away is now known as pvo | 23:41 | |
*** johnpur has quit IRC | 23:42 | |
creiht | soren: because that's how our ops guys wanted it | 23:45 |
soren | creiht: Ah, so it's an internal Rackspace thing? | 23:47 |
creiht | soren: that, and it makes sense for swift :) | 23:47 |
soren | How so? | 23:47 |
*** hggdh has quit IRC | 23:47 | |
* creiht sighs | 23:47 | |
creiht | do we have to go through this *again*? :) | 23:47 |
letterj | soren: I believe that our guys have tried it both way and found that they wanted more control of the services. | 23:48 |
*** adiantum has joined #openstack | 23:49 | |
creiht | a.) There is a lot of configuration that needs to be done before starting services makes any sense on a new install | 23:49 |
*** zul has quit IRC | 23:49 | |
creiht | b.) We talked about having a default swift-one install that could install a working system on one machine, but that hasn't had any priority | 23:49 |
letterj | No one has more operational experience running swift clusters than our ops guys. | 23:50 |
soren | I'm just trying to understand here. Almost every single other daemon in Ubuntu and Debian is started by default. So when I see on that doesn't, it looks odd. | 23:51 |
creiht | you said the same thing last time :) | 23:51 |
soren | I'm sure I did. | 23:51 |
soren | :) | 23:51 |
soren | I'm glad I'm consistent :) | 23:52 |
creiht | There may be a point some time in the future that it will make sense to do that, it just doesn't right now | 23:52 |
soren | General style in Debian packaging is to enable stuff by default if there are good, safe defaults. "If you don't want the service running, dont' install it" sort of stuff. | 23:53 |
soren | creiht: Fair enough. | 23:53 |
soren | I'll make a note of that. | 23:53 |
soren | This will probably be clearer once I try to run it more. | 23:53 |
soren | So a "# We don't yet have a default configuration, so we can't start the deamons by default" sort of comment next to the dh_installinit --no-start stuff would be accurate? | 23:55 |
creiht | good enough | 23:55 |
*** ccustine has quit IRC | 23:56 | |
eday | I'm not quite sure how you would do it, given the filesystem requirements | 23:57 |
*** dragondm has quit IRC | 23:57 | |
redbo | yeah, a "default configuration" doesn't make any sense | 23:58 |
creiht | well there may be a "I just want to install a single instance to try out" package that would have a default package at some point | 23:58 |
redbo | yeah, that's a different animal | 23:58 |
eday | soren: do you know of any other packages that create loopback devices or partitions as part of the install? | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!