| soren | mtaylor: It'll only sometimes work. | 00:00 |
|---|---|---|
| mtaylor | hrm | 00:00 |
| soren | anyjson prefers cjson over anything else. | 00:00 |
| soren | So if it finds cjson, it'll use it, even though simplejson is installed. | 00:00 |
| mtaylor | then why doens't it use cjson on jenkins? | 00:00 |
| soren | It has a Depends: cjson | simpjejson | 00:00 |
| mtaylor | can we conflict with cjson? | 00:00 |
| soren | mtaylor: Because it's not installed. | 00:00 |
| mtaylor | gotcha | 00:00 |
| soren | mtaylor: That's a good idea. | 00:00 |
| soren | mtaylor: Let's do that. | 00:00 |
| mtaylor | yay! | 00:00 |
| * mtaylor had a good idea | 00:01 | |
| * mtaylor will now consider it useful that he got out of bed :) | 00:01 | |
| creiht | soren: you may want to test jasonlib and jsonlib2 as well, as a quick look at anyjson seems to also prefer those | 00:03 |
| creiht | soren: or possibly can you use force_implementation in anyjson to force a specific json module? | 00:04 |
| creiht | I'm not sure how carrot is used, so not sure if that would be available | 00:05 |
| soren | creiht: Ubuntu has no jasonlib nor jsonlib2, so we're in the clear there. | 00:05 |
| soren | creiht: I could patch carrot. | 00:05 |
| creiht | soren: while that is true, that doesn't prevent someone from pip install jsonlib | 00:06 |
| mtaylor | we don't really test anything in a pip install environment | 00:06 |
| soren | They might also have pip installed carrot, so I can't fix it there for them. | 00:06 |
| creiht | I'm not saying continually test | 00:07 |
| vishy | more importantly, why is cjson broken? | 00:07 |
| vishy | should we patch that? | 00:07 |
| creiht | just test to see if it is compatible | 00:07 |
| mtaylor | probably | 00:07 |
| creiht | hah | 00:07 |
| mtaylor | I mean- just to be nice | 00:07 |
| creiht | cjson is broken in many ways | 00:07 |
| soren | vishy: I couldn't even find an upstream bugtracker. | 00:07 |
| mtaylor | well, if it's broken in many ways - perhaps we should get it removed from debian/ubuntu? | 00:07 |
| vishy | if cjson is broken why the hell does anyjson prefer it? | 00:07 |
| mtaylor | and patch anyjson to not prefer it? | 00:08 |
| mtaylor | vishy: people tend to like fast things even if they suck | 00:08 |
| soren | or maybe noone noticed. | 00:08 |
| mtaylor | probably the second thing | 00:08 |
| vishy | mtaylor: :) | 00:08 |
| creiht | cjson is very fast | 00:09 |
| creiht | and for common cases it is probably reasonable | 00:09 |
| creiht | but there are edge cases where most of the json implementations differ | 00:09 |
| creiht | especially in the area of unicode | 00:09 |
| soren | Having a dict where one of the keys has a '/' doesn't seem wildly unreasonable. | 00:09 |
| soren | Er... one of the values. | 00:09 |
| soren | It's probably the same for keys, but we're seeing it with values. | 00:09 |
| creiht | http://garybernhardt.blogspot.com/2007/07/when-json-isnt-json.html | 00:11 |
| creiht | old post, but still seems to be true | 00:11 |
| *** bcwaldon has quit IRC | 00:12 | |
| creiht | there was another post that did a similar comparison that focused on unicde a while back, but can't seem to find it | 00:12 |
| soren | Why the heck escape forward slashes anyway? | 00:13 |
| creiht | http://deron.meranda.us/python/comparing_json_modules/unicode | 00:13 |
| mtaylor | soren: looks like they just did re.escape('/dev/vdc') | 00:13 |
| * creiht shrugs | 00:13 | |
| creiht | I highly recommend settling on simplejson | 00:13 |
| mtaylor | which is probably someone not really being sure what they needed so deciding to 'escape' things to be 'safe' | 00:13 |
| creiht | and forcing things to just use that (if possible) | 00:13 |
| soren | It looks like a simple patch. | 00:14 |
| soren | *headdesk* | 00:18 |
| soren | There's even a unit test that verifies it does this correctly. | 00:18 |
| soren | It's commented out. | 00:18 |
| creiht | you will fix that one problem, but you will only run into many others | 00:18 |
| soren | The lesson is: Never try. | 00:19 |
| creiht | soren: yes that is exactly what I was saying | 00:19 |
| *** adiantum has quit IRC | 00:20 | |
| soren | Oh, it's not even an upstream bug. It's a patch that was added in Debian. | 00:20 |
| *** adiantum has joined #openstack-dev | 00:20 | |
| *** heckj has quit IRC | 00:21 | |
| *** mszilagyi has quit IRC | 00:22 | |
| soren | http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2009-4924 | 00:23 |
| uvirtbot | soren: Dan Pascu python-cjson 1.0.5 does not properly handle a ['/'] argument to cjson.encode, which makes it easier for remote attackers to conduct certain cross-site scripting (XSS) attacks involving Firefox and the end tag of a SCRIPT element. (http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2009-4924) | 00:23 |
| soren | I don't think 2:30 AM is the right time to figure this out. | 00:24 |
| soren | g'night. | 00:24 |
| *** Binbin has joined #openstack-dev | 00:43 | |
| *** cloudgroups has joined #openstack-dev | 00:43 | |
| *** adiantum has quit IRC | 00:49 | |
| *** RobertLaptop has quit IRC | 00:49 | |
| *** RobertLaptop has joined #openstack-dev | 00:53 | |
| *** cloudgroups has quit IRC | 00:55 | |
| *** markvoelker1 has joined #openstack-dev | 00:55 | |
| *** cloudgroups has joined #openstack-dev | 00:59 | |
| *** adiantum has joined #openstack-dev | 01:02 | |
| *** adiantum has quit IRC | 01:08 | |
| *** jdurgin has quit IRC | 01:14 | |
| *** jaypipes has quit IRC | 01:21 | |
| *** adiantum has joined #openstack-dev | 01:22 | |
| *** adiantum has quit IRC | 01:39 | |
| *** adiantum has joined #openstack-dev | 01:52 | |
| *** markvoelker1 has quit IRC | 01:52 | |
| *** nati has joined #openstack-dev | 01:59 | |
| *** mancdaz1203 has quit IRC | 02:00 | |
| *** mancdaz1203 has joined #openstack-dev | 02:02 | |
| *** jkoelker has quit IRC | 02:05 | |
| *** Binbin has quit IRC | 02:15 | |
| *** adiantum has quit IRC | 02:24 | |
| *** nati has quit IRC | 02:27 | |
| *** dragondm has quit IRC | 02:28 | |
| *** adiantum has joined #openstack-dev | 02:37 | |
| *** cloudgroups has left #openstack-dev | 02:49 | |
| *** adiantum has quit IRC | 02:53 | |
| *** adiantum has joined #openstack-dev | 02:59 | |
| *** adiantum has quit IRC | 03:09 | |
| *** adiantum has joined #openstack-dev | 03:14 | |
| *** libspeed has joined #openstack-dev | 03:27 | |
| *** adiantum has quit IRC | 03:37 | |
| *** adiantum has joined #openstack-dev | 03:51 | |
| *** bcwaldon has joined #openstack-dev | 03:59 | |
| *** bcwaldon has quit IRC | 04:12 | |
| *** adiantum has quit IRC | 04:21 | |
| *** Binbin has joined #openstack-dev | 04:27 | |
| *** adiantum has joined #openstack-dev | 04:34 | |
| *** adiantum has quit IRC | 05:08 | |
| *** adiantum has joined #openstack-dev | 05:14 | |
| *** zaitcev has quit IRC | 05:18 | |
| *** nati has joined #openstack-dev | 05:19 | |
| *** adiantum has quit IRC | 05:23 | |
| *** johnpur has quit IRC | 05:25 | |
| *** adiantum has joined #openstack-dev | 05:30 | |
| *** adiantum has quit IRC | 05:38 | |
| *** adiantum has joined #openstack-dev | 05:51 | |
| *** Binbin has quit IRC | 06:00 | |
| *** cloudgroups has joined #openstack-dev | 06:11 | |
| *** Binbin has joined #openstack-dev | 06:13 | |
| *** adiantum has quit IRC | 06:20 | |
| *** glenc_ has quit IRC | 06:25 | |
| *** glenc has joined #openstack-dev | 06:25 | |
| *** cloudgroups has left #openstack-dev | 06:35 | |
| *** reidrac has joined #openstack-dev | 07:02 | |
| *** nati has quit IRC | 07:10 | |
| *** gregm has joined #openstack-dev | 07:14 | |
| *** nati has joined #openstack-dev | 07:23 | |
| *** erik-s has quit IRC | 07:32 | |
| *** erik-s has joined #openstack-dev | 07:33 | |
| soren | ttx: PPA builds are working again. | 07:45 |
| *** nati has quit IRC | 08:00 | |
| *** gregm has quit IRC | 08:01 | |
| *** nati has joined #openstack-dev | 08:02 | |
| *** nati has quit IRC | 08:36 | |
| *** nsokolov has joined #openstack-dev | 09:02 | |
| ttx | kewl. | 09:04 |
| *** nsokolov has quit IRC | 09:15 | |
| *** nsokolov has joined #openstack-dev | 09:15 | |
| *** nsokolov is now known as chemikadze | 09:15 | |
| *** Binbin has quit IRC | 09:39 | |
| *** markvoelker has quit IRC | 11:02 | |
| *** markvoelker has joined #openstack-dev | 11:06 | |
| *** zul has quit IRC | 12:01 | |
| sandywalsh | finally ... a review day where I don't a deadline looming over me :) | 12:02 |
| *** dprince has joined #openstack-dev | 12:15 | |
| *** westmaas_ is now known as westmaas | 12:25 | |
| *** dprince has quit IRC | 12:25 | |
| *** dprince has joined #openstack-dev | 12:29 | |
| *** zul has joined #openstack-dev | 12:42 | |
| *** nati has joined #openstack-dev | 12:43 | |
| *** ameade has joined #openstack-dev | 12:56 | |
| *** agarwalla has joined #openstack-dev | 13:04 | |
| agarwalla | does the nova database have any key information to access the compute nodes ? | 13:05 |
| sandywalsh | agarwalla, can you elaborate? | 13:06 |
| agarwalla | sandywalsh: i need to access the compute nodes from the cloud controller node...i realised i can retrieve all the compute nodes from the "services" table but i need a way to get into the compute nodes...does nova create a default user/passwd for compute nodes or store keys to access the compute nodes | 13:08 |
| sandywalsh | agarwalla, you need to send an AMQP message from nova.compute.api. This will get picked up by a Compute node for processing. You can specify which compute node in the "topic" | 13:10 |
| agarwalla | sandywalsh: i would have chosen the rabbitmq channel to talk to the compte controller but i wish to run commands that the nova compute is not responsible | 13:10 |
| sandywalsh | agarwalla, not sure what you mean by "not responsible" | 13:11 |
| agarwalla | sandywalsh: ok i need to get information of the interfaces on the remote compute host...i cannot add a method on the nova compute side to do that right ? | 13:13 |
| sandywalsh | agarwalla, yes, if that info doesn't exist in the database and there is no existing Compute call for getting that info, you can add a new method to nova.compute.manager to support it (and in the Virt layer as well, of course) | 13:15 |
| agarwalla | afaik, the compute host interfaces info is not present in the nova databases today...why would i need any implementation in the virt layer ? (i need the compute host interface info) | 13:22 |
| sandywalsh | agarwalla, depends on if you're getting it from the hypervisor or just the Compute host. If it's just the Host then you don't need the Virt layer, correct. | 13:24 |
| agarwalla | sandywalsh: yes, true. i'll try to implement the AMQP way then..thanks a bunch for the response..appreciate it... also do i open a bug or something to track this | 13:28 |
| sandywalsh | agarwalla, np ... ideally a bug or blueprint to indicate what you're trying to accomplish. | 13:30 |
| *** throughn1thing has quit IRC | 13:31 | |
| sandywalsh | ttx, for a simple addition, is there something smaller than a blueprint? A bug almost doesn't seem appropriate. | 13:31 |
| ttx | sandywalsh: bug with "Wishlist" importance sounds like the solution here. | 13:32 |
| sandywalsh | ttx thanks ... agarwalla ^^ | 13:32 |
| agarwalla | great...thanks sandywalsh and ttx | 13:33 |
| sandywalsh | np | 13:33 |
| *** throughnothing has joined #openstack-dev | 13:35 | |
| *** throughnothing has quit IRC | 13:36 | |
| *** throughnothing has joined #openstack-dev | 13:37 | |
| *** throughnothing has quit IRC | 13:39 | |
| *** throughnothing has joined #openstack-dev | 13:41 | |
| *** throughnothing has quit IRC | 13:41 | |
| *** bcwaldon has joined #openstack-dev | 13:44 | |
| *** throughnothing has joined #openstack-dev | 13:50 | |
| *** throughnothing has quit IRC | 13:51 | |
| *** jkoelker has joined #openstack-dev | 13:55 | |
| *** exlt_ has joined #openstack-dev | 13:58 | |
| *** throughnothing has joined #openstack-dev | 13:59 | |
| *** throughnothing has quit IRC | 13:59 | |
| *** exlt has quit IRC | 14:00 | |
| *** throughnothing has joined #openstack-dev | 14:01 | |
| *** throughnothing has quit IRC | 14:01 | |
| *** throughnothing has joined #openstack-dev | 14:01 | |
| *** throughnothing has quit IRC | 14:01 | |
| *** nati has quit IRC | 14:02 | |
| *** throughnothing has joined #openstack-dev | 14:04 | |
| *** throughnothing has quit IRC | 14:04 | |
| *** throughnothing has joined #openstack-dev | 14:05 | |
| *** nati has joined #openstack-dev | 14:19 | |
| *** markvoelker has quit IRC | 14:25 | |
| *** deshantm has quit IRC | 14:40 | |
| *** deshantm has joined #openstack-dev | 14:45 | |
| *** heckj has joined #openstack-dev | 15:20 | |
| *** nati has quit IRC | 15:24 | |
| *** jaypipes has joined #openstack-dev | 15:37 | |
| *** reidrac has quit IRC | 15:42 | |
| *** mattray has joined #openstack-dev | 15:45 | |
| *** jkoelker has quit IRC | 15:58 | |
| *** jkoelker has joined #openstack-dev | 16:11 | |
| *** johnpur has joined #openstack-dev | 16:11 | |
| *** ChanServ sets mode: +v johnpur | 16:11 | |
| *** jkoelker has quit IRC | 16:15 | |
| *** jkoelker has joined #openstack-dev | 16:16 | |
| *** chemikadze has quit IRC | 16:16 | |
| *** rnirmal has joined #openstack-dev | 16:20 | |
| *** jkoelker has quit IRC | 16:23 | |
| *** johnpur has quit IRC | 16:28 | |
| *** ohnoimdead has joined #openstack-dev | 16:29 | |
| *** bsa has joined #openstack-dev | 16:41 | |
| *** markvoelker has joined #openstack-dev | 16:43 | |
| *** mszilagyi has joined #openstack-dev | 16:55 | |
| *** jdurgin has joined #openstack-dev | 16:57 | |
| jaypipes | ohnoimdead: nice nick. :) | 17:20 |
| ohnoimdead | jaypipes: :D | 17:21 |
| *** pyhole has joined #openstack-dev | 17:37 | |
| *** jkoelker has joined #openstack-dev | 17:41 | |
| *** cp16net has joined #openstack-dev | 17:50 | |
| vishy | jaypipes: ping | 17:53 |
| jaypipes | vishy: pong | 18:07 |
| vishy | jay: do you recall what the deal is with compute images api | 18:07 |
| vishy | are we getting rid of it? | 18:07 |
| vishy | markwash, blamar, bwaldon: ping(s) | 18:07 |
| markwash_ | o/ | 18:08 |
| *** blamar_ has joined #openstack-dev | 18:08 | |
| jaypipes | vishy: hmm, yes, I think I remember that /images will be removed from the compute API. | 18:09 |
| jaypipes | vishy: just not sure whether the plan was for 1.1 or 2.0 to have the removal... | 18:09 |
| vishy | ok | 18:10 |
| westmaas | right now /images on nova is not a straight proxy for glance, so probably not 1.1? | 18:10 |
| vishy | I noticed an issue in the compute images api and i want to verify and figure out the right fix | 18:10 |
| *** dragondm has joined #openstack-dev | 18:10 | |
| *** dragondm has quit IRC | 18:11 | |
| vishy | the href for an image that is returned in links is for the compute api | 18:11 |
| *** dragondm has joined #openstack-dev | 18:11 | |
| vishy | but if you try to launch an instance using that href it blows up because imageRef is expecting a glance href | 18:11 |
| vishy | so do we modify compute images api to return a glance href, or do we modify the instance create to allow compute hrefs as well? | 18:12 |
| * markwash_ slowly backs away | 18:12 | |
| markwash_ | I guess I'm leaning towards allowing compute hrefs | 18:13 |
| markwash_ | but maybe just to disagree with westmass | 18:14 |
| vishy | i think that makes more sense as well | 18:14 |
| westmaas | sounds good. disagree with that markwash_! | 18:15 |
| markwash_ | the difficulty there is that right now all the drivers talk directly to glance for images | 18:16 |
| markwash_ | :-( | 18:16 |
| vishy | that shouldn't be a problem | 18:16 |
| vishy | it is just allowing an additional option for imageRef | 18:17 |
| vishy | right now we support 3 and http://glance:port/image/3 | 18:17 |
| markwash_ | so we could just translate http://nova:port/images/3 into http://glance:port/image/3 | 18:17 |
| vishy | so we just add support for http://compute:port/v.1.1/image/3 | 18:17 |
| vishy | right, basically we need a check to see if the server port matches the current flags and if so just turn it into a 3 and use the default glance server | 18:18 |
| vishy | as if just '3' was passed in | 18:18 |
| markwash_ | sounds good to me | 18:18 |
| jaypipes | vishy: and to throw another wrench in the works, we'll likely be switching to UUID ids in Glance soon. | 18:20 |
| vishy | dammit jaypipes | 18:21 |
| vishy | :) | 18:21 |
| vishy | jaypipes: also, didn't we discuss is_public at some point? | 18:21 |
| vishy | jaypipes: I thought that at some point the images details was returning all images and not just the public ones | 18:21 |
| jaypipes | vishy: well, with keystone integration, we have the concept of an owner of an image, so we can support finer-grained things like shared image groups. But... that said, Keystone integration will not be required in Glance, and if the auth middleware if not installed, we will continue the same behaviour of is_public that exists now. | 18:23 |
| jaypipes | vishy: of course, all subject to change if there is sufficient demand for it.. | 18:24 |
| vishy | ok I guess i just imagined it | 18:24 |
| vishy | currently making an image private is broken in nova | 18:24 |
| jaypipes | vishy: perhaps you are remembering the bug (fixed now) that was not limiting the returned images in /images and /images/detail to only active status ones? | 18:24 |
| jaypipes | vishy: pls explain. what's broken about it? bug link? | 18:25 |
| jaypipes | vishy: is this a break on the Glance side or the Nova side? | 18:27 |
| vishy | well it is just a disconnect between is_public | 18:33 |
| vishy | nova sets is_public false | 18:33 |
| vishy | then it can't see the images to display them to the user anymore | 18:33 |
| vishy | :) | 18:33 |
| jaypipes | vishy: ah! | 18:38 |
| jaypipes | vishy: but /images/<ID> works... | 18:38 |
| jaypipes | vishy: you're referring to listing images available in the Nova /images endpoint, I assume? | 18:38 |
| jaypipes | markwash_: got a sec? | 18:39 |
| vishy | jaypipes: the listing in euca-describe-images actually | 18:42 |
| vishy | but same for images endpoint | 18:42 |
| vishy | jaypipes: nova doesn't maintain a separate list of images so it doesn't know to request the id specifically | 18:43 |
| jaypipes | vishy: gotcha. | 18:43 |
| openstackjenkins | Project swift build #283: SUCCESS in 27 sec: http://jenkins.openstack.org/job/swift/283/ | 18:47 |
| openstackjenkins | Tarmac: This fixes the bug that drop_buffer_cache() doesn't work on systems where off_t isn't 64 bits. | 18:47 |
| vishy | jaypipes: i think the short term solution is to not use is_public but have a property for it properties['ec2_public'] or something | 18:48 |
| jaypipes | vishy: that would certainly work. | 18:49 |
| vishy | until flitering works | 18:49 |
| vishy | i actually did it that way initially, but we switched back to using is_public and I don't know why | 18:49 |
| *** CatKiller has joined #openstack-dev | 18:59 | |
| CatKiller | Hi there, | 18:59 |
| ameade | is there a good reason instance types have a flavorid column? | 19:06 |
| vishy | because they are flavorids? | 19:08 |
| vishy | :) | 19:08 |
| ameade | couldn't the id column just be used wherever that is used? | 19:09 |
| vishy | ameade: flavor id is used by the openstack api to refer to an instance type | 19:09 |
| markwash_ | jaypipes: I do now! | 19:10 |
| vishy | i think the reason for a second column is id is autoinc | 19:10 |
| vishy | so it can't be set by an administrator | 19:10 |
| ameade | vishy: so we just want admins to have the ability to set the id | 19:11 |
| ameade | vishy: something they cant do with autoinc | 19:11 |
| vishy | yes, | 19:11 |
| vishy | that is my assumption anyway :) | 19:11 |
| jaypipes | markwash_: hey :) was wondering if you'd be able to give your input on this MP: https://code.launchpad.net/~jshepher/glance/functional_tests_using_httplib2/+merge/65078. Specifically, Justin's last comment... | 19:11 |
| markwash_ | onit | 19:12 |
| *** dprince has quit IRC | 19:20 | |
| *** agarwalla_ has joined #openstack-dev | 19:36 | |
| *** agarwalla has quit IRC | 19:37 | |
| *** ohnoimdead has quit IRC | 20:11 | |
| *** jaypipes is now known as jaypipes-afk | 20:31 | |
| vishy | jaypipes: perhaps we could change image_get_all_public to image_get_all and put is_public in as a default filter? | 20:46 |
| bcwaldon | vishy: I would be all for that. I think we have to wait until we have authz to solve the 'public' problem correctly | 20:59 |
| vishy | right but we can fake it using filters for now | 21:00 |
| vishy | :) | 21:00 |
| bcwaldon | vishy: sure thing | 21:00 |
| *** ameade has quit IRC | 21:21 | |
| *** exlt_ is now known as exlt | 21:37 | |
| *** zaitcev has joined #openstack-dev | 21:47 | |
| *** adiantum has joined #openstack-dev | 22:04 | |
| *** adiantum has quit IRC | 22:10 | |
| *** adiantum has joined #openstack-dev | 22:14 | |
| *** cp16net has quit IRC | 22:15 | |
| *** markvoelker has quit IRC | 22:26 | |
| openstackjenkins | Project swift build #284: SUCCESS in 27 sec: http://jenkins.openstack.org/job/swift/284/ | 22:32 |
| openstackjenkins | Tarmac: removed stats/logging code from swift | 22:32 |
| *** adiantum has quit IRC | 22:34 | |
| *** bcwaldon has quit IRC | 22:35 | |
| *** bsa has quit IRC | 22:39 | |
| *** mattray has quit IRC | 22:47 | |
| *** adiantum has joined #openstack-dev | 22:47 | |
| openstackjenkins | Project nova build #1,028: SUCCESS in 2 min 56 sec: http://jenkins.openstack.org/job/nova/1028/ | 22:49 |
| openstackjenkins | Tarmac: Fixes issues with displaying exceptions regarding flavors in nova-manage | 22:49 |
| *** adiantum has quit IRC | 22:54 | |
| *** rnirmal has quit IRC | 23:05 | |
| *** jaypipes-afk is now known as jaypipes | 23:06 | |
| *** adiantum has joined #openstack-dev | 23:06 | |
| *** adiantum has quit IRC | 23:18 | |
| *** adiantum has joined #openstack-dev | 23:31 | |
| *** adiantum has quit IRC | 23:40 | |
| *** adiantum has joined #openstack-dev | 23:45 | |
| *** adiantum has quit IRC | 23:52 | |
| *** adiantum has joined #openstack-dev | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!