| vishy | s1rp: ping | 00:00 |
|---|---|---|
| anotherjesse | mtaylor: private message ping | 00:13 |
| *** chipc has joined #openstack-dev | 00:18 | |
| *** heckj has quit IRC | 00:38 | |
| *** mgius has quit IRC | 00:44 | |
| *** mszilagyi has quit IRC | 00:45 | |
| *** rods has quit IRC | 00:49 | |
| *** alekibango has quit IRC | 00:53 | |
| *** soren has quit IRC | 00:53 | |
| *** alekibango has joined #openstack-dev | 00:53 | |
| *** soren has joined #openstack-dev | 00:53 | |
| *** ChanServ sets mode: +v soren | 00:53 | |
| *** mfer has joined #openstack-dev | 01:00 | |
| *** jakedahn has quit IRC | 01:13 | |
| *** anotherjesse has quit IRC | 01:15 | |
| *** Tushar has quit IRC | 01:40 | |
| *** anotherjesse has joined #openstack-dev | 01:44 | |
| *** anotherjesse has quit IRC | 01:44 | |
| *** mfer has quit IRC | 02:06 | |
| *** lorin1 has joined #openstack-dev | 02:13 | |
| *** lorin1 has left #openstack-dev | 02:13 | |
| openstackgerrit | Joe Heck proposed a change to openstack/openstack-ci-puppet: fixes lp827070 https://review.openstack.org/243 | 02:56 |
| *** nci has joined #openstack-dev | 03:05 | |
| *** zorzar has quit IRC | 03:30 | |
| *** jakedahn has joined #openstack-dev | 03:32 | |
| *** mwhooker has quit IRC | 03:33 | |
| *** zorzar has joined #openstack-dev | 03:39 | |
| *** pasik has quit IRC | 03:42 | |
| *** mwhooker has joined #openstack-dev | 03:45 | |
| *** jakedahn has quit IRC | 04:07 | |
| *** tsuzuki_ has joined #openstack-dev | 07:01 | |
| *** reidrac has joined #openstack-dev | 07:09 | |
| *** pasik has joined #openstack-dev | 07:37 | |
| *** alekibango has quit IRC | 08:02 | |
| *** negronjl has quit IRC | 08:13 | |
| *** alekibango has joined #openstack-dev | 08:38 | |
| *** negronjl has joined #openstack-dev | 08:39 | |
| openstackgerrit | Yuriy Taraday proposed a change to openstack/keystone: Moved run_test logic into abstract class. https://review.openstack.org/244 | 09:00 |
| *** negronjl has quit IRC | 09:07 | |
| *** darraghb has joined #openstack-dev | 09:07 | |
| *** rods has joined #openstack-dev | 09:12 | |
| *** phil121 has joined #openstack-dev | 09:17 | |
| *** tsuzuki_ has quit IRC | 09:20 | |
| *** negronjl has joined #openstack-dev | 09:35 | |
| *** mxant has joined #openstack-dev | 09:39 | |
| *** mxant has quit IRC | 09:53 | |
| *** lorin1 has joined #openstack-dev | 09:53 | |
| *** lorin1 has left #openstack-dev | 09:55 | |
| ttx | soren: anything to do on the nova side for bug 806911 ? If not, just mark it invalid | 10:22 |
| uvirtbot | Launchpad bug 806911 in nova "Split nova-compute package into nova-compute-{kvm,xen,etc.}" [Wishlist,Confirmed] https://launchpad.net/bugs/806911 | 10:22 |
| zykes- | why should one split it ? | 10:35 |
| reidrac | it's convenient if you want to use one hypervisor and you don't want to install rependencies related the ones you don't want to use (?) | 10:42 |
| zykes- | ah | 10:43 |
| *** jamespage has joined #openstack-dev | 11:09 | |
| *** markvoelker has joined #openstack-dev | 11:12 | |
| *** alekibango has quit IRC | 11:13 | |
| soren | ttx: Nope, I think it's all done. | 11:13 |
| *** lorin1 has joined #openstack-dev | 11:23 | |
| ttx | soren: ok -- marked the nova task invalid. | 11:36 |
| *** reidrac has left #openstack-dev | 11:39 | |
| *** reidrac has joined #openstack-dev | 11:39 | |
| *** alekibango has joined #openstack-dev | 11:44 | |
| *** mfer has joined #openstack-dev | 11:52 | |
| *** Adri2000 has quit IRC | 12:08 | |
| *** Adri2000 has joined #openstack-dev | 12:08 | |
| *** chipc has quit IRC | 12:13 | |
| *** lts has joined #openstack-dev | 12:14 | |
| *** zykes- has quit IRC | 12:25 | |
| *** zykes- has joined #openstack-dev | 12:26 | |
| *** dprince has joined #openstack-dev | 12:32 | |
| *** sandywalsh has joined #openstack-dev | 12:36 | |
| dprince | Just killed the Jenkins tarmac job (it was hung for 14 hours). This should allow the approved merge props to land. | 12:39 |
| *** bsza has joined #openstack-dev | 12:47 | |
| *** chipc has joined #openstack-dev | 12:55 | |
| openstackjenkins | Project nova build #1,253: SUCCESS in 3 min 25 sec: https://jenkins.openstack.org/job/nova/1253/ | 13:06 |
| openstackjenkins | * Tarmac: Refactored the HostFilterScheduler and LeastCostScheduler classes so that they can be combined into a single class that can do both host filtering and host weighting, allowing subclasses to override those processes as needed. Also renamed the ZoneAwareScheduler to AbstractScheduler, for two reasons: one, the 'zone-aware' designation was necessary when the zone code was being developed; now that it is part of | 13:06 |
| openstackjenkins | * Tarmac: cosmetic change in test_extensions. | 13:06 |
| openstackjenkins | Avoids constant merge conflicts between proposals with new extensions. | 13:06 |
| openstackjenkins | * Tarmac: Validate the size of VHD files in OVF containers. | 13:06 |
| *** kbringard has joined #openstack-dev | 13:07 | |
| *** mancdaz has quit IRC | 13:18 | |
| *** nmistry has joined #openstack-dev | 13:27 | |
| *** amccabe has joined #openstack-dev | 13:31 | |
| dprince | jk0: Can you have a look at: https://bugs.launchpad.net/nova/+bug/827373 | 13:50 |
| uvirtbot | Launchpad bug 827373 in nova "Booting m1.tiny instance causes 'ImageTooLarge' exception" [Undecided,New] | 13:50 |
| jk0 | seems like having local_gb as 0 is the problem... math seems correct | 13:51 |
| jk0 | anyone know why that would be 0? | 13:51 |
| dprince | jk0: local_gb 0 is the default setting in our database. And lots of people use m1.tiny for testing etc. | 13:52 |
| jk0 | right but why would it be 0? | 13:52 |
| jk0 | that doesn't make any sense | 13:52 |
| jk0 | wonder if adding a migration and moving that to 1 would be the way to go | 13:53 |
| dprince | jk0: Yeah. That sounds reasonable. This is going to cause some pain otherwise. | 13:53 |
| jk0 | cool. having it at 0 is silly IMHO | 13:54 |
| dprince | jk0: Perhaps adding a check to the _check_image_size so that if 'local_gb' is zero we log that as well. | 13:55 |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: Added reponse handling for xsd static file rendering II https://review.openstack.org/245 | 13:55 |
| jk0 | yeah, good call | 13:56 |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: Ziad's changes from github.com/rackspace/openstack https://review.openstack.org/237 | 13:58 |
| dprince | jk0: you wanna take this one? | 13:58 |
| jk0 | I'll try and work it in today unless someone else grabs it before then | 13:58 |
| dprince | sure. Thanks. | 13:58 |
| *** nmistry has quit IRC | 14:19 | |
| *** zykes- has quit IRC | 14:21 | |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Added pylint message count as run_tests.sh -l https://review.openstack.org/246 | 14:29 |
| *** jaypipes has quit IRC | 14:39 | |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Moved run_test logic into abstract class. https://review.openstack.org/244 | 14:42 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Moved run_test logic into abstract class. https://review.openstack.org/244 | 14:44 |
| *** dolphm has joined #openstack-dev | 14:44 | |
| *** nmistry has joined #openstack-dev | 14:45 | |
| *** martine has joined #openstack-dev | 14:45 | |
| openstackgerrit | A change was merged to openstack/keystone: Moved run_test logic into abstract class. https://review.openstack.org/244 | 14:48 |
| *** jaypipes has joined #openstack-dev | 14:53 | |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: Added reponse handling for xsd static file rendering III https://review.openstack.org/237 | 14:53 |
| *** dprince has quit IRC | 14:57 | |
| *** rnirmal has joined #openstack-dev | 15:01 | |
| *** dragondm has joined #openstack-dev | 15:02 | |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Creating an artificial whitespace merge conflict https://review.openstack.org/247 | 15:04 |
| *** zykes- has joined #openstack-dev | 15:05 | |
| *** reidrac has quit IRC | 15:06 | |
| openstackgerrit | A change was merged to openstack/keystone: Creating an artificial whitespace merge conflict https://review.openstack.org/247 | 15:07 |
| *** mancdaz has joined #openstack-dev | 15:08 | |
| *** yogirackspace has joined #openstack-dev | 15:10 | |
| openstackjenkins | Project nova build #1,254: SUCCESS in 3 min 56 sec: https://jenkins.openstack.org/job/nova/1254/ | 15:10 |
| openstackjenkins | Tarmac: Updated the EC2 metadata controller so that it returns the correct value for instance-type metadata. | 15:10 |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: Added reponse handling for xsd static file rendering III Extra extension tests (for RS-KEY) https://review.openstack.org/237 | 15:12 |
| *** zns has joined #openstack-dev | 15:14 | |
| *** kbringard has quit IRC | 15:15 | |
| *** mattray has joined #openstack-dev | 15:15 | |
| openstackgerrit | A change was merged to openstack/keystone: Added reponse handling for xsd static file rendering III Extra extension tests (for RS-KEY) https://review.openstack.org/237 | 15:16 |
| *** nmistry has quit IRC | 15:19 | |
| *** heckj has joined #openstack-dev | 15:20 | |
| *** nmistry has joined #openstack-dev | 15:20 | |
| *** cp16net has joined #openstack-dev | 15:21 | |
| *** heckj has quit IRC | 15:23 | |
| *** heckj has joined #openstack-dev | 15:23 | |
| *** cp16net has quit IRC | 15:31 | |
| *** cp16net has joined #openstack-dev | 15:31 | |
| *** dprince has joined #openstack-dev | 15:37 | |
| ttx | hmm; what's the correct way of running run_tests.sh with just one test ? | 15:44 |
| *** Guest77784 has quit IRC | 15:45 | |
| *** Guest77784 has joined #openstack-dev | 15:46 | |
| comstud | i end up just running (from root of a branch): nosetests -s <path/to/file>:Class.method | 15:46 |
| jaypipes | ttx: ./run_tests.sh -V nova.tests.test_utils | 15:46 |
| comstud | ttx: ^^ | 15:46 |
| ttx | thx guys ! | 15:46 |
| jaypipes | comstud: ./run_tests.sh takes all the nose args, too.. | 15:47 |
| comstud | cools | 15:47 |
| *** nmistry has quit IRC | 15:47 | |
| *** cp16net has quit IRC | 15:58 | |
| *** zns has quit IRC | 15:58 | |
| *** cp16net has joined #openstack-dev | 16:00 | |
| *** zns has joined #openstack-dev | 16:04 | |
| *** p1ra has joined #openstack-dev | 16:17 | |
| *** p1ra has left #openstack-dev | 16:17 | |
| openstackgerrit | Monty Taylor proposed a change to openstack/openstack-ci-puppet: Added python-webtest to quantum slave. https://review.openstack.org/248 | 16:27 |
| openstackgerrit | Monty Taylor proposed a change to openstack/openstack-ci-puppet: Added memcached to the keystone slave. https://review.openstack.org/202 | 16:27 |
| *** troytoman-away is now known as troytoman | 16:29 | |
| openstackgerrit | Monty Taylor proposed a change to openstack/openstack-ci-puppet: Added python-webtest to quantum slave. https://review.openstack.org/248 | 16:32 |
| openstackgerrit | Verification of a change to openstack/keystone failed: Added pylint message count as run_tests.sh -l https://review.openstack.org/246 | 16:34 |
| *** negronjl has quit IRC | 16:38 | |
| *** negronjl has joined #openstack-dev | 16:39 | |
| openstackjenkins | Project nova build #1,255: SUCCESS in 4 min 5 sec: https://jenkins.openstack.org/job/nova/1255/ | 16:46 |
| openstackjenkins | Tarmac: This branch implements a nova api extension which allows you to manage and update tenant/project quotas. | 16:46 |
| heckj | mtaylor: did I do something wrote with https://review.openstack.org/243 that you created https://review.openstack.org/248 | 16:48 |
| heckj | mtaylor: er, /wrote/wrong | 16:49 |
| mtaylor | heckj: nope! I just somehow missed yours. oops :) | 16:54 |
| heckj | mtaylor: heh, I wondered if you were just hammering out patches this morning and ran over it. Tried to save you some work. | 16:54 |
| openstackgerrit | A change was merged to openstack/openstack-ci-puppet: fixes lp827070 https://review.openstack.org/243 | 16:54 |
| heckj | mtaylor: is "bug/######" the preferred format to link to bugs in LP? | 16:55 |
| heckj | mtaylor: I kept with the lp###### style. | 16:55 |
| mtaylor | heckj: you can mention bugs in the commit message either way | 16:55 |
| mtaylor | heckj: bug#### bug:#### lp:##### lp #####, etc ... | 16:55 |
| mtaylor | heckj: ([Bb]ug|[Ll][Pp])\s*[#:]?\s*(\d+) | 16:56 |
| heckj | mtaylor: cool - nice! Don't have to worry about format so much then | 16:56 |
| mtaylor | heckj: actually is the regex | 16:56 |
| mtaylor | heckj: and then we're working on more integration stuff with that at the moment | 16:56 |
| *** Tushar has joined #openstack-dev | 16:57 | |
| *** cp16net has quit IRC | 16:59 | |
| *** ohnoimdead has joined #openstack-dev | 17:03 | |
| *** YorikSar has joined #openstack-dev | 17:21 | |
| *** jpipes has joined #openstack-dev | 17:21 | |
| *** jpipes has quit IRC | 17:21 | |
| *** jaypipes has quit IRC | 17:21 | |
| *** jaypipes has joined #openstack-dev | 17:23 | |
| *** jaypipes has quit IRC | 17:24 | |
| *** jaypipes has joined #openstack-dev | 17:24 | |
| YorikSar | Hello. Any Keystone or testing fans here? | 17:24 |
| jaypipes | YorikSar: yes, fan of both :) | 17:25 |
| YorikSar | I've sent to review the first part of my suggestion. Teh easy one. | 17:25 |
| jaypipes | YorikSar: link? | 17:26 |
| YorikSar | But now I'm thinking about shell scripts used for testing and fixturing. Is it really nesessary? | 17:26 |
| jaypipes | YorikSar: not sure what you mean... are you asking if run_tests.sh and run_tests.py are necessary? | 17:27 |
| YorikSar | jaypipes: Oh, as I see now, that request is already merged. | 17:27 |
| jaypipes | YorikSar: :) | 17:28 |
| YorikSar | jaypipes: I mean, why do we have both | 17:28 |
| jaypipes | YorikSar: the shell script is a little wrapper that handles the virtualenv stuff cleanly... | 17:28 |
| jaypipes | YorikSar: I suppose one could refactor it out. | 17:28 |
| YorikSar | jaypipes: Shouldn't it be better to write all this logic in one Python script | 17:28 |
| jaypipes | YorikSar: yep. It's just legacy, that's all. Feel free to consolidate them if you wish. | 17:29 |
| YorikSar | Is venv being used in run_test.py? | 17:29 |
| jaypipes | YorikSar: venv is activated by the run_tests.sh script (it prepends tools/with_venv.sh to the call to run_tests.py | 17:30 |
| YorikSar | jaypipes: It can be activated by the script itself as well | 17:30 |
| jaypipes | YorikSar: so you would need to roll that functionality into the run_tests.py, which may be a bit difficult, since the shell script sets the venv environment then executes the tests within that venv... | 17:30 |
| dolphm | YorikSar: jaypipes: i added a pylint option to run_tests.sh, but it hasn' | 17:31 |
| dolphm | ...hasn't been merged yet | 17:31 |
| YorikSar | jaypipes: I had a bit of experience with some complex import-reimport and sys.path hacking stuff | 17:31 |
| YorikSar | dolphm: All that pylints and pep8 can be run from the Python script - that's my point. | 17:32 |
| YorikSar | So as I understand, if I send a request that eliminates both .sh files tomorrow, it has chance to be merged? | 17:33 |
| dolphm | YorikSar: i find ./run_tests.sh and python run_tests.py to be way too many characters to type, so i actually have a fabric script in my repo so i can just run "fab test", "fab review" etc | 17:34 |
| jaypipes | dolphm: hmm ./run<TAB> is all I need ;) | 17:34 |
| zykes- | clea | 17:34 |
| YorikSar | dolphm: ./r<tab><tab><tab> is way shorter | 17:35 |
| dolphm | ./ is too awkward to constantly type | 17:35 |
| jaypipes | YorikSar: :) | 17:35 |
| *** lorin1 has quit IRC | 17:35 | |
| * jaypipes buys dolphm typing lessons :P | 17:35 | |
| * dolphm can type faster than you | 17:35 | |
| YorikSar | dolphm: I even type "git fetch origin" all the time. | 17:35 |
| *** jamespage has left #openstack-dev | 17:36 | |
| YorikSar | dolphm: It gives some time to think about the action before pressing enter | 17:36 |
| * dolphm wouldn't mind taking a typing class for a dvorak layout | 17:36 | |
| dolphm | jenkins is currently dependent on being able to use ./run_tests.sh, right? | 17:37 |
| YorikSar | dolphm, jaypipes: So, one more question. Is bin/ dir really nessesary? | 17:37 |
| *** joesavak has joined #openstack-dev | 17:37 | |
| dolphm | YorikSar: a lot of the code in it should 110% be refactored into keystone/ | 17:38 |
| jaypipes | YorikSar: uhm, yeah? | 17:38 |
| dolphm | bin/ should contain just a couple lines of actual python code | 17:38 |
| jaypipes | YorikSar: bin/ is for executables installed | 17:38 |
| YorikSar | dolphm, jaypipes: I mean, there is smth like a convention never to write a scrips but to use a distutils script installers instead | 17:39 |
| dprince | cd ~/projects/nova/trunk/ | 17:39 |
| dprince | ls | 17:39 |
| dprince | oops | 17:39 |
| dolphm | :) | 17:39 |
| openstackjenkins | Project nova build #1,256: SUCCESS in 3 min 32 sec: https://jenkins.openstack.org/job/nova/1256/ | 17:40 |
| openstackjenkins | Tarmac: * Added search instance by metadata. | 17:40 |
| openstackjenkins | * instance_get_all_by_filters should filter deleted | 17:40 |
| jaypipes | YorikSar: could you show an example of that for WSGI servers? | 17:40 |
| YorikSar | Well, the bin dir is there, but shouldn't it be removed just like the shell scripts? | 17:41 |
| jaypipes | YorikSar: I don't think so... | 17:41 |
| *** cp16net has joined #openstack-dev | 17:42 | |
| jaypipes | YorikSar: where would you put executables? In the keystone/ directory along with regular python modules? | 17:43 |
| dolphm | jaypipes: i think his point is that they're completely unnecessary | 17:43 |
| YorikSar | jaypipes: For example, SimpleHTTPServer in stblib :) | 17:43 |
| jaypipes | dolphm: how so? | 17:43 |
| jaypipes | YorikSar: you're not serious, are you? | 17:44 |
| *** anotherjesse has joined #openstack-dev | 17:44 | |
| YorikSar | jaypipes: nowhere. There should be functionality somewhere in keystone package that can be called from outside either by "python -m" or by a distutils-like script. | 17:45 |
| dolphm | jaypipes: 'python -m keystone' <-- instead of './bin/keystone' | 17:45 |
| dolphm | 'python -m keystone --admin-only' <-- instead of './bin/keystone-admin' | 17:46 |
| dolphm | 'python -m keystone --service-only' <-- instead of './bin/keystone-service' | 17:46 |
| YorikSar | jaypipes: I'm serious a bit. You can run 'pythom -m SimpleHTTPServer' and get a cute small server for static files. | 17:46 |
| dolphm | YorikSar: the fact that we're serving static files through keystone makes me cry | 17:46 |
| jaypipes | dolphm: And that's how you want people to call keystone? | 17:46 |
| YorikSar | dolphm: may be 'python -m keystone.server.admin' and 'python -m keystone.server.service' should be better. | 17:47 |
| jaypipes | dolphm: well, I agree with you on the static files thing... | 17:47 |
| dolphm | YorikSar: that works! | 17:47 |
| dolphm | jaypipes: sure, why not? it's python, treat it like python and you'll have a cleaner project | 17:47 |
| jaypipes | YorikSar, dolphm: Would you guys think starting Apache (if Apache were a Python executable...) like "python -m httpd" would be appropriate? :( | 17:48 |
| dolphm | jaypipes: we're serving a billion more static files today than we were yesterday... | 17:48 |
| Vek | jaypipes: you have a moment? Trying to puzzle out how best to approach the functional tests... | 17:48 |
| YorikSar | jaypipes: That's how they can be called by developers, but for human beings distutils will generate a script. | 17:48 |
| jaypipes | YorikSar: well, I'm certainly open to seeing a demo of it. :) | 17:48 |
| YorikSar | jaypipes: For example, try "cat `which pep8`" | 17:48 |
| jaypipes | YorikSar: easy_install is not how someone would install on a production system. That's not package-worthy... | 17:50 |
| dolphm | YorikSar: jaypipes: if we refactor the python code out of bin/, making 'python -m keystone', 'python -m keystone.server', 'python -m keystone.server.admin', 'python -m keystone.server.service' all work would be trivial | 17:50 |
| YorikSar | jaypipes: About httpd - certainly no. But nothing prevents you from keeping all that apache code in some apache package, and just create an entry point in setup.py. | 17:50 |
| YorikSar | jaypipes: Well, lots of debs actually just wrap this process... | 17:51 |
| dolphm | jaypipes: it also bugs me that keystone is NOT an executable, it's just a freaking web service, so why is it not running under a real web server (like apache/nginx) anyway? i don't see any advantage to reinventing the wheel here | 17:52 |
| dolphm | /vent | 17:52 |
| *** lorin1 has joined #openstack-dev | 17:52 | |
| jaypipes | YorikSar: like I said, I welcome the code :) | 17:52 |
| jaypipes | dolphm: you might want to ask creiht that question.... | 17:53 |
| YorikSar | dolphm: Good point. Currently we force users to use a Paste, but why not cute things like uwsgi or popular httpd? | 17:53 |
| *** darraghb has quit IRC | 17:55 | |
| *** mxant has joined #openstack-dev | 17:56 | |
| *** mxant has quit IRC | 17:57 | |
| *** jsavak has joined #openstack-dev | 18:00 | |
| *** joesavak has quit IRC | 18:00 | |
| *** jsavak has quit IRC | 18:00 | |
| YorikSar | jaypipes: Well, I've checked some projects and all of them actually have bin/ dir with scripts almost identical to the distutils ones. | 18:02 |
| zykes- | nginx :p | 18:02 |
| zns | Yoriksar: many of the responses from Keystone are static (ex: WADL, version contracts, etc.). I agree we don't want it to be a static file server, and anyone can configure it to point to a different endpoint for the contracts, but at least installing it on a laptop and having access to all the contracts is OK imo. | 18:03 |
| dolphm | zykes-: nginx is awesome, but doesn't it have some kind of issue running python wsgi? | 18:03 |
| zykes- | dunno dolphm | 18:03 |
| dolphm | zykes-: i've only used it as a static file server & reverse proxy | 18:03 |
| zns | And anyone should fee free to cache the static files… | 18:03 |
| zns | feel | 18:04 |
| *** joesavak has joined #openstack-dev | 18:04 | |
| zykes- | varnish :) | 18:04 |
| YorikSar | dolphm, zykes-: It cannot run wsgi, but it has a cool uwsgi protocol implementation. | 18:04 |
| zns | Squid | 18:04 |
| jaypipes | YorikSar: aha :) | 18:04 |
| zykes- | zns: varnish is nicer i think :p | 18:04 |
| YorikSar | jaypipes: But I still don't get the reason. | 18:04 |
| jaypipes | Squarnish FTW! | 18:05 |
| dolphm | zns: if that's the acknowledged intention, it static files simply shouldn't be served by keystone in a production deployment ... the threat vectors on the static files controller are pretty scary | 18:05 |
| YorikSar | jaypipes: Why don't use that distutils to generate the same, but version-aware script | 18:06 |
| jaypipes | YorikSar: lack of knowledge on our part? :) Like I said, I'd love to see code demonstrating that | 18:06 |
| zns | dolphm: yes. Is there a better way to serve static files on a one-machine, one port deployment? | 18:08 |
| dolphm | zns: that's exactly what i've used nginx for | 18:09 |
| zykes- | what's bad about keystone serving static ? | 18:09 |
| zns | dolphm: but I don't have to install/config nginx to run keystone on my machine now… we could offer an alternative config (like we support sqlite and MySQL for different scale deployments). | 18:11 |
| jaypipes | zykes-: just consumes resources that can be better served through a simple web server like nginx or lighttpd... | 18:11 |
| dolphm | zns nginx listens on your public port (say, :5000), checks for requests to a specific path... '/static/', if there's a match, serve the static file, if there's no match, internally forward (reverse proxy) the request to the underlying keystone server (on some arbitrary private port) | 18:11 |
| zykes- | that's easy to do | 18:11 |
| dolphm | zykes-: security is our responsibility, code overhead, limited end user configuration | 18:11 |
| dolphm | zns: for example, django serves your static files for you in a development environment (with zero overhead on your part), but refuses to do so in a production environment (citing security) | 18:13 |
| dolphm | and i believe rails follows the same pattern (?) | 18:14 |
| YorikSar | jaypipes: I can explain how to use it, but of course I cannot provide a full solution for Keystone right now. | 18:17 |
| YorikSar | jaypipes: Just add an entry_points parameter in which you provide names of the generated scripts and module:method pairs for them. | 18:18 |
| dolphm | YorikSar: .. why 'of course'? lol | 18:18 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 18:19 |
| YorikSar | jaypipes: So after setup.py install or bdist the will be generated and put into apropriate dir under either build/ directory or somewhere in /usr/local/bin. | 18:19 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 18:21 |
| YorikSar | dolphm: Because I started clearing out the keystone-manage script and found there some bad force to fight with. | 18:21 |
| dolphm | YorikSar: you're refactoring it? | 18:22 |
| YorikSar | dolphm: Yes. | 18:22 |
| dolphm | sweet | 18:22 |
| YorikSar | dolphm: But don't expect anything too sweet. My goal is to separate options parsing from command execution at first. I want to make sampledata.py run thousand times quicker than sampledata.sh. | 18:23 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 18:27 |
| jk0 | dprince: I'll have a fix for you to test coming in a few minutes | 18:28 |
| dolphm | YorikSar: sounds like a MAJOR improvement :D | 18:28 |
| *** lorin1 has quit IRC | 18:28 | |
| dolphm | YorikSar: ... does that mean we could functional tests using an in-memory sqlite backend? | 18:28 |
| zns | YorikSar: +1 | 18:28 |
| YorikSar | dolphm: If I will break down current command processing logic into something object-oriented, it will definetly be in separate request. | 18:29 |
| dolphm | YorikSar: can't wait! | 18:29 |
| YorikSar | dolphm: I'm not sure about functional tests. I never tried to understand the big difference in details of their implementation. | 18:30 |
| YorikSar | dolphm: btw, can you enlighten me, why do unit tests require running keystone server? | 18:31 |
| dolphm | YorikSar: one of the major issues with testing is that a most of them are dependent on sampledata.sh to populate the db.. which is A) slow, B) obviously not self-contained in the tests, C) sampledata.sh starts and stops the server on every call, which means an in-memory db configuration would be wiped away after every line in sampledata.sh | 18:31 |
| dolphm | YorikSar: most of the "unit" tests are actually functional/system tests... which require a server to run against | 18:32 |
| YorikSar | dolphm: C) caused a huge rewrite of fakeldap from Nova | 18:32 |
| YorikSar | dolphm: But why cannot we just exclude http server and client from testing? | 18:33 |
| YorikSar | dolphm: I mean, make almost direct call to WSGI app and get a response instead of running a stand-alone http server and using httplib to access it. | 18:35 |
| dolphm | YorikSar: we have a few such tests, and more are certainly welcome, but i think we also need higher level tests to test the interaction between middleware & server... so starting keystone in various configs makes sense there | 18:37 |
| *** lorin1 has joined #openstack-dev | 18:37 | |
| dolphm | just as an example | 18:37 |
| jaypipes | YorikSar: keystone already includes entry_points in setup()... | 18:38 |
| jaypipes | dolphm: Glance does functional testing just fine using an in-memory SQLite database. | 18:39 |
| dolphm | jaypipes: yeah, but our tests are currently handicapped by sampledata.sh | 18:39 |
| jaypipes | dolphm: yup. | 18:40 |
| dolphm | jaypipes: e.g. the tests shouldn't depend at all on sampledata.sh | 18:40 |
| jaypipes | dolphm: right. all that stuff should be done in FunctionalTestCase.setUp()... | 18:40 |
| dolphm | jaypipes: ++ | 18:40 |
| YorikSar | jaypipes: But they include only entry points for paste, not console scripts that are installed the way I described. | 18:40 |
| jaypipes | YorikSar: that's what the scripts= param is for... | 18:41 |
| YorikSar | dolphm, jaypipes: I'm almost ready to run back to workplace to finish work on sampledata :) | 18:41 |
| jaypipes | YorikSar: :) | 18:42 |
| jaypipes | YorikSar: where are you based? | 18:42 |
| dolphm | jaypipes: how do other projects bootstrap themselves (in tests or production)? i.e. keystone needs an Admin role, admin user, and a link between the two before the API can be used at all | 18:42 |
| zykes- | has anyone made the thought of making a common way for doing things ? | 18:42 |
| *** alekibango has quit IRC | 18:42 | |
| dolphm | zykes-: https://github.com/openstack/openstack-skeleton ? | 18:43 |
| jaypipes | dolphm: the API server should be able to be started without an admin role/user established... is that not the case? | 18:43 |
| zykes- | dolphm: more like common libraries | 18:43 |
| zykes- | etc | 18:43 |
| YorikSar | jaypipes: scripts param defines existing scripts that is already there and should be copied to disribution package and to the target system. | 18:43 |
| jaypipes | zykes-: in skeleton, there is an openstack/ directory which is being used as the place to put common lib stuff.. | 18:43 |
| jaypipes | zykes-: still very new, though :( | 18:44 |
| dolphm | jaypipes: not that i'm aware of | 18:44 |
| *** lorin1 has quit IRC | 18:44 | |
| zykes- | jaypipes: i meant like a openstack-commons | 18:44 |
| dolphm | jaypipes: (you can start it, but it's pointless if you can't login as admin) | 18:44 |
| zykes- | project for doing "common" code | 18:44 |
| YorikSar | jaypipes: console_scripts section in entrypoints makes the system generate a scripts that just run a specific function in specific module. | 18:44 |
| dolphm | from openstack.common import wsgi | 18:44 |
| zykes- | like exceptions, wsgi helper module, utils.py etc | 18:44 |
| jaypipes | zykes-: I know. openstack-common is being "baked" in the skeleton project... | 18:44 |
| YorikSar | jaypipes: I'm in Russia, Saratov. | 18:44 |
| zykes- | ok | 18:45 |
| jaypipes | zykes-: see the skeleton project.. those things are in openstack/common/* | 18:45 |
| jaypipes | YorikSar: cool. :) | 18:45 |
| zykes- | YorikSar: you work for rackspace no ? | 18:45 |
| YorikSar | zykes-: No, Mirantis is my employer. | 18:46 |
| zykes- | ah | 18:46 |
| zykes- | anyone know if reddwarf is started on ? | 18:46 |
| zykes- | or atlas | 18:47 |
| pvo | zykes- reddwarf has been started | 18:47 |
| pvo | though I don't know the status as of today. | 18:47 |
| zykes- | pvo: is it @ github? | 18:47 |
| pvo | zykes- : https://github.com/rackspace/reddwarf | 18:48 |
| jk0 | dprince: https://code.launchpad.net/~jk0/nova/lp827373/+merge/71759 | 18:48 |
| zykes- | oh | 18:48 |
| zykes- | my company is pretty interested in stuff like that | 18:48 |
| zykes- | maybe i can get some folks on it | 18:48 |
| pvo | zykes-: rock on | 18:49 |
| zykes- | we've got more devs then infrastructure folks so :p | 18:49 |
| *** lorin1 has joined #openstack-dev | 18:49 | |
| zykes- | what's monkeyfarm ? | 18:50 |
| dprince | jk0: Thank you sir. I'm cool with what you did. Let me kick the tires on it... | 18:50 |
| jk0 | cool | 18:50 |
| dprince | Question about admin passwords in nova. | 18:51 |
| dprince | <for anyone> | 18:51 |
| dprince | We appear to save admin password in the instances table and I'd like to know why? | 18:51 |
| dprince | That seems like a very bad idea. | 18:51 |
| *** alekibango has joined #openstack-dev | 18:52 | |
| dprince | I'm feel like a ticket storm is coming here.... | 18:53 |
| blamar_ | dprince: yeah....not so much of a good idea | 18:53 |
| blamar_ | dprince: agreed on the bug report :) | 18:53 |
| openstackjenkins | Project nova build #1,257: SUCCESS in 3 min 27 sec: https://jenkins.openstack.org/job/nova/1257/ | 19:00 |
| openstackjenkins | Tarmac: Use netaddr's subnet features to calculate subnets. | 19:00 |
| *** cp16net has quit IRC | 19:01 | |
| *** zns has quit IRC | 19:01 | |
| vishy | heckj: http://wiki.openstack.org/UnderstandingFlatNetworking very nice. I might suggest changing it to say FlatDhcpNetworking | 19:02 |
| vishy | since there is a flat mode too | 19:02 |
| zykes- | is there anything going on to use openvswitch =? | 19:04 |
| *** jsavak has joined #openstack-dev | 19:04 | |
| heckj | vishy: Thank you - will do. It needs to be updated with the --multi-host stuff though | 19:04 |
| *** zns has joined #openstack-dev | 19:05 | |
| *** joesavak has quit IRC | 19:05 | |
| heckj | vishy: was trying to describe both Flat and FlatDHCP in the same document (as opposed to VLAN) | 19:06 |
| *** cp16net has joined #openstack-dev | 19:07 | |
| *** cp16net has quit IRC | 19:08 | |
| *** cp16net has joined #openstack-dev | 19:08 | |
| *** lorin1 has quit IRC | 19:12 | |
| *** lorin1 has joined #openstack-dev | 19:13 | |
| *** bengrue has joined #openstack-dev | 19:19 | |
| *** hisaharu has joined #openstack-dev | 19:29 | |
| zykes- | wonder what's gonna happen to the Cortex thingie from citrix | 19:31 |
| zykes- | after they | 19:31 |
| zykes- | they'e got Olympus & cloud.com | 19:31 |
| *** lorin1 has quit IRC | 19:34 | |
| *** hisaharu has quit IRC | 19:34 | |
| *** hisaharu has joined #openstack-dev | 19:37 | |
| *** mattray has quit IRC | 19:39 | |
| *** lorin1 has joined #openstack-dev | 19:42 | |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 19:47 |
| openstackgerrit | Verification of a change to openstack/keystone failed: Additional changes to fix minor service support stuff and increase test coverage. Also making validate token call available using service admin tokens https://review.openstack.org/242 | 19:48 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Added pylint message count as run_tests.sh -l https://review.openstack.org/246 | 19:50 |
| openstackgerrit | A change was merged to openstack/keystone: Added pylint message count as run_tests.sh -l https://review.openstack.org/246 | 19:53 |
| openstackgerrit | Yogeshwar Srikrishnan proposed a change to openstack/keystone: Additional changes to fix minor service support stuff and increase test coverage. Also making validate token call available using service admin tokens https://review.openstack.org/242 | 19:56 |
| ttx | Team meeting in one hour in #openstack-meeting | 19:58 |
| *** bengrue has quit IRC | 19:58 | |
| zykes- | ttx: nova ? | 19:59 |
| *** lorin1 has quit IRC | 19:59 | |
| *** johnpur has joined #openstack-dev | 19:59 | |
| *** ChanServ sets mode: +v johnpur | 19:59 | |
| ttx | zykes-: openstack. | 19:59 |
| *** lorin1 has joined #openstack-dev | 20:01 | |
| *** jakedahn has joined #openstack-dev | 20:03 | |
| ttx | lorin1: hey -- do you plan to submit code for GPU/BareMetal/Tilera ? Or not in time for Diablo ? | 20:03 |
| openstackgerrit | Verification of a change to openstack/keystone failed: Additional changes to fix minor service support stuff and increase test coverage. Also making validate token call available using service admin tokens https://review.openstack.org/242 | 20:05 |
| openstackgerrit | Joe Savak proposed a change to openstack/keystone: Updates to XML and JSON changes for validateToken https://review.openstack.org/250 | 20:07 |
| openstackgerrit | A change was merged to openstack/keystone: Updates to XML and JSON changes for validateToken https://review.openstack.org/250 | 20:11 |
| *** dprince has quit IRC | 20:11 | |
| *** joesavak has joined #openstack-dev | 20:12 | |
| *** zul has quit IRC | 20:14 | |
| *** johnpur has quit IRC | 20:14 | |
| *** jsavak has quit IRC | 20:15 | |
| *** joesavak has quit IRC | 20:16 | |
| *** cp16net has quit IRC | 20:21 | |
| zykes- | ttx: tilera ? | 20:24 |
| ttx | zykes-: https://blueprints.launchpad.net/nova/+spec/heterogeneous-tilera-architecture-support | 20:24 |
| *** lorin1 has quit IRC | 20:30 | |
| openstackjenkins | Project nova build #1,258: SUCCESS in 4 min 20 sec: https://jenkins.openstack.org/job/nova/1258/ | 20:33 |
| openstackjenkins | * Tarmac: Fixes issue with exceptions getting eaten in image/s3.py if there is a failure during register. The variables referenced with locals() were actually out of scope. | 20:33 |
| openstackjenkins | * Tarmac: Allow local_gb size to be 0. libvirt uses local_gb as a secondary drive, but XenServer uses it as the root partition's size. Now we support both. | 20:33 |
| openstackjenkins | Include some random PEP8 fixes. | 20:33 |
| *** Gordonz has joined #openstack-dev | 20:36 | |
| openstackgerrit | Kevin L. Mitchell proposed a change to openstack/keystone: Port of glance-control to keystone. This will make writing certain keystone integration functional tests a little easier to do. https://review.openstack.org/251 | 20:37 |
| jk0 | vishy: https://bugs.launchpad.net/nova/+bug/827634 | 20:42 |
| uvirtbot | Launchpad bug 827634 in nova "local_gb is not used consistently between hypervisors" [Undecided,New] | 20:42 |
| vishy | jk0: thx | 20:43 |
| jk0 | np | 20:44 |
| vishy | jk0 if you are booting a whole disk image, how does xs handle it? | 20:45 |
| jk0 | it just takes the local_gb value and converts it to gb | 20:46 |
| jk0 | something like local_gb * 1024 * 1024 * 1024 | 20:47 |
| vishy | right but does it try to resize whole disk images? | 20:47 |
| vishy | what if the filesystem is something funky like ntfs? | 20:47 |
| jk0 | oh, I'm not sure of those details offhand | 20:48 |
| vishy | jk0: this seems like a big gap to me. Ok guess I will dig into the code :) | 20:48 |
| *** jroault has joined #openstack-dev | 20:48 | |
| *** jroault has quit IRC | 20:49 | |
| jk0 | I think it basically just creates a vdi and links it to the instance, and then the image picks it up from there | 20:49 |
| jk0 | but yeah I'd have to look too | 20:49 |
| jk0 | I'm curious how the other hypervisors use it | 20:49 |
| *** jrouault has joined #openstack-dev | 20:49 | |
| *** jrouault has left #openstack-dev | 20:51 | |
| *** statik has quit IRC | 20:53 | |
| *** statik has joined #openstack-dev | 20:53 | |
| *** statik has joined #openstack-dev | 20:53 | |
| *** joesavak has joined #openstack-dev | 20:54 | |
| *** rnirmal has quit IRC | 20:54 | |
| vishy | jk0: we use local_gb as a second drive so it works with any image. | 20:54 |
| vishy | we used to repartition the file system, but that is always dangerous | 20:55 |
| ttx | Meeting starts in 2 minutes in #openstack-meeting, please join | 20:58 |
| Vek | really? I thought the leafs would be more flammable </tounge-in-cheek> | 20:59 |
| *** mfer has quit IRC | 20:59 | |
| *** Tushar85 has joined #openstack-dev | 21:00 | |
| *** Tushar has quit IRC | 21:00 | |
| *** Tushar85 has quit IRC | 21:00 | |
| openstackgerrit | Verification of a change to openstack/glance failed: Adds SSL configuration params to the client https://review.openstack.org/190 | 21:02 |
| *** mattray has joined #openstack-dev | 21:02 | |
| *** mattray has quit IRC | 21:07 | |
| *** dolphm_ has joined #openstack-dev | 21:08 | |
| openstackjenkins | Project burrow build #31: SUCCESS in 11 sec: https://jenkins.openstack.org/job/burrow/31/ | 21:16 |
| openstackjenkins | Tarmac: Test case cleanup, moved backend base test cases to test.backend (out of memory test). | 21:16 |
| *** zul has joined #openstack-dev | 21:18 | |
| *** asomya has joined #openstack-dev | 21:18 | |
| *** lts has quit IRC | 21:20 | |
| *** markvoelker has quit IRC | 21:22 | |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 21:24 |
| *** jsavak has joined #openstack-dev | 21:32 | |
| *** joesavak has quit IRC | 21:32 | |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: More contract changes https://review.openstack.org/252 | 21:33 |
| vishy | dprince, tr3buchet: can you recheck https://code.launchpad.net/~danwent/nova/linuxnet_vifplug/+merge/70254 to make sure concrns have been addressed? | 21:36 |
| openstackgerrit | A change was merged to openstack/keystone: Port of glance-control to keystone. This will make writing certain keystone integration functional tests a little easier to do. https://review.openstack.org/251 | 21:39 |
| openstackgerrit | A change was merged to openstack/keystone: Admin WADL updates https://review.openstack.org/249 | 21:42 |
| tr3buchet | vishy: sure | 21:43 |
| *** YorikSar has quit IRC | 21:46 | |
| *** zul has quit IRC | 21:52 | |
| *** chipc has quit IRC | 21:55 | |
| mwhooker | ugh. anyone else have problems with circular dependencies will all this come in __init__.py? | 21:57 |
| mwhooker | s/come/code/ | 21:57 |
| *** vladimir3p has joined #openstack-dev | 22:00 | |
| openstackgerrit | James E. Blair proposed a change to openstack/openstack-ci: Add blueprint update script. https://review.openstack.org/253 | 22:04 |
| openstackgerrit | James E. Blair proposed a change to openstack/openstack-ci-puppet: Put blueprint updates into production. https://review.openstack.org/254 | 22:07 |
| vishy | mwhooker: yes, that is why i try to never put code in __init__ | 22:08 |
| vishy | mwhooker: where are you hitting it? | 22:08 |
| mwhooker | nova/api/ec2/__init__.py | 22:09 |
| mwhooker | thinking of just breaking though out into another file | 22:09 |
| mwhooker | http://pastebin.com/WrMxTvXA | 22:10 |
| vishy | mwhooker: yes, seems fine. I had to put ec2utils into a separate file because of circular deps | 22:11 |
| *** bsza has quit IRC | 22:12 | |
| openstackgerrit | A change was merged to openstack/openstack-ci: Add blueprint update script. https://review.openstack.org/253 | 22:12 |
| mwhooker | alright | 22:12 |
| vishy | mwhooker: a lot of those are tesing fixtures and/or docstrings and licenses as opposed to actual code | 22:12 |
| mwhooker | while I've got you on the horn, can you help me figure out how "image_location", as is used in the ec2 api, should be validated? | 22:13 |
| vishy | hmm | 22:13 |
| mwhooker | sure, but the couple near the bottom | 22:13 |
| mwhooker | not sure if it's a url path, a unix path, or any os path | 22:13 |
| vishy | it is supposed to be a bucket/manifest.xml | 22:14 |
| vishy | it is the s3 path to an xml file | 22:14 |
| mwhooker | ok, so url | 22:14 |
| mwhooker | though I notice os.path.split is used | 22:14 |
| mwhooker | so who knows how that will work where the os path separator isn't '/' | 22:14 |
| mwhooker | join, rather. not split | 22:15 |
| openstackgerrit | A change was merged to openstack/openstack-ci-puppet: Put blueprint updates into production. https://review.openstack.org/254 | 22:15 |
| vishy | jaypipes, notmyname: any ideas on jason's question in #openstack ? | 22:18 |
| vishy | bug 827660 | 22:21 |
| uvirtbot | Launchpad bug 827660 in glance "Cannot store files greater than 5GB in Swift" [Critical,Confirmed] https://launchpad.net/bugs/827660 | 22:21 |
| *** amccabe has quit IRC | 22:22 | |
| *** zul has joined #openstack-dev | 22:25 | |
| *** jsavak has quit IRC | 22:28 | |
| *** bengrue has joined #openstack-dev | 22:33 | |
| *** anotherjesse has quit IRC | 22:42 | |
| *** anotherjesse has joined #openstack-dev | 22:42 | |
| *** asomya has quit IRC | 22:51 | |
| *** yogirackspace has left #openstack-dev | 22:51 | |
| jaypipes | vishy: yeah, working on it... | 22:52 |
| vishy | jaypipes: coolness. As i mentioned on the bug, we should have a content-length header, no? | 22:52 |
| *** martine has quit IRC | 22:55 | |
| jaypipes | vishy: not necessarily. we do chunked-transfer encoding to avoid having to read the entire request into memory... | 22:55 |
| jaypipes | vishy: but that said, if the client does know the size of the image, we *can* set the content-length header and read off of that... | 22:56 |
| vishy | jaypipes: hmm. Yeah, seems like it is reasoable to require the client to send in the right content-length | 22:56 |
| jaypipes | vishy: sure, as long as it's cheap and doesn't interfere with the chunked-transfer... | 22:57 |
| *** zns has quit IRC | 22:57 | |
| *** zns has joined #openstack-dev | 22:57 | |
| openstackgerrit | Monty Taylor proposed a change to openstack/openstack-ci: Added a version of the rfc script from gluster. https://review.openstack.org/235 | 23:00 |
| *** zns has quit IRC | 23:01 | |
| *** yogirackspace has joined #openstack-dev | 23:05 | |
| openstackjenkins | Project nova build #1,259: SUCCESS in 3 min 28 sec: https://jenkins.openstack.org/job/nova/1259/ | 23:06 |
| openstackjenkins | Tarmac: I'm taking Thierry at his word that I should merge early and merge often :) | 23:06 |
| openstackjenkins | In Diablo-3 we introduced "vif-plugging" to the hypervisor "virt" layer, allowing flexibility in how vNICs are attached to the network switch. This allowed non-linux bridge switch technologies (e.g., Open vSwitch, 802.1qbh) to be used with nova. | 23:06 |
| openstackjenkins | This blueprint adds a similar capability to linux_net.py, allowing the L3/DHCP capabilities to be "plugged" into Quantum networks. Like in the virt layer, we created a vif driver that represents the behavior of Nova prior to the change (LinuxBridgeInterfaceDriver) and make it the default. We also add a new driver for Open Vswitch that can be enabled using a flag (LinuxOVSInterfaceDriver). The code is designed to su | 23:06 |
| openstackjenkins | Most of the interesting code is at the bottom of linux_net.py, where the drivers are defined. I had to pull some common code related to setting IPs on devices out of ensure_bridge() so it could be used by either approach. The driver's plug() method is invoked by the VlanManager's _setup_network() method. Currently unplug() is unused, which seems to be inline with how the existing nova code works. | 23:06 |
| openstackjenkins | In many places in linux_net.py, I had to tweak functions to accept the name of the linux device to configure, rather than just assuming it was the 'bridge' field in the network object, since with OVS it could be any linux device. The code I am least sure about are the changes to bin/nova-dhcpbridge. I changed to this key off of the network ID, rather than the bridge name. | 23:06 |
| openstackjenkins | I've tested this with the linux bridge and with the OVS vif-plugging driver. I was able to confirm that L3 forwarding and DHCP were operating correctly. | 23:06 |
| *** anotherjesse_ has joined #openstack-dev | 23:08 | |
| s1rp | jaypipes: just added a note to the large-file glance tix | 23:08 |
| s1rp | jaypipes: i think we can do the same 5GB segmentation we do w/ the ibacks now | 23:09 |
| s1rp | jaypipes: the difference, is in the Ibacks we maintain the manifest ourselves; with swift's large-file support, we can use their manifest for keeping track of each chunk | 23:09 |
| *** zns has joined #openstack-dev | 23:10 | |
| jaypipes | s1rp: sure. we still need to have the client calculate the filesize before transferring though. Not a huge deal, but that's what needs to happen... | 23:11 |
| s1rp | jaypipes: we do? even if we're using chunked-transfer-encoding? | 23:11 |
| *** anotherjesse has quit IRC | 23:12 | |
| *** anotherjesse_ is now known as anotherjesse | 23:12 | |
| jaypipes | s1rp: if we're going to signal to Swift to do large-object transfer, we need the size up front... | 23:13 |
| *** troytoman is now known as troytoman-away | 23:13 | |
| jaypipes | s1rp: at some point, you need to know the size of the image file... and there's no "add this chunk" method in Swift, so we can't calculate the number of bytes sent to Swift via some iterator. | 23:15 |
| s1rp | jaypipes: the add this chunk is the PUT request to the container, if i'm reading this correctly... then once you've PUT'd all the chunks, then you register the chunks by uploading the manifest | 23:16 |
| jaypipes | s1rp: furthermore, calling body_file.seek() on the webob.Request leads webob to copy the entire request body into a temporary file on disk... | 23:16 |
| s1rp | jaypipes: as long as CTE is upported by the PUT, i think you can do this w/o knowing size upfront | 23:16 |
| s1rp | jaypipes: we shouldn't be seeking | 23:17 |
| s1rp | jaypipes: we have no reason to go backwards in the stream | 23:17 |
| s1rp | jaypipes: we 1) create container 2) cte upload each chunk 3) write manifest | 23:17 |
| jaypipes | s1rp: so you would loop on: while chunk = body_file.read(chunksize), swift_client.put_object(chunk)? | 23:19 |
| s1rp | jaypipes: yeah, basically, though you have to add extra code to know when to terminate the request to start the next chunk | 23:20 |
| s1rp | jaypipes: did this on the iback, don't see a reason why this wouldn't work in glance as well | 23:20 |
| jaypipes | s1rp: ok, lemme stew on it... | 23:20 |
| s1rp | jaypipes: cools | 23:21 |
| *** rods has quit IRC | 23:22 | |
| jaypipes | s1rp: or we could just set up a GlusterFS cluster and use the filesystem backend in Glance ;) | 23:23 |
| *** yogirackspace has left #openstack-dev | 23:23 | |
| s1rp | jaypipes: hehe :) | 23:24 |
| *** zns has quit IRC | 23:27 | |
| *** ohnoimdead has quit IRC | 23:31 | |
| *** anotherjesse has quit IRC | 23:41 | |
| *** anotherjesse_ has joined #openstack-dev | 23:41 | |
| *** anotherjesse_ is now known as anotherjesse | 23:41 | |
| vishy | mtaylor, Daviey: ping | 23:47 |
| mtaylor | vishy: whazzyp? | 23:53 |
| *** dragondm has quit IRC | 23:53 | |
| vishy | mtaylor: so lets suppose there is a really cool piece of software that has a couple of extra things in contrib that compile to useful binaries that i want to use | 23:53 |
| mtaylor | vishy: ok. good supposition | 23:53 |
| vishy | mtaylor: and this piece of software doesn't include those binaries in the ubuntu packaging | 23:54 |
| mtaylor | vishy: fail | 23:54 |
| vishy | what is my best strategy? | 23:54 |
| vishy | specifically dnsmasq has a cool binary called dhcp_release | 23:54 |
| mtaylor | vishy: well - we could make packages of that for our repos that have those binaries included - and then send a patch upstream | 23:54 |
| mtaylor | vishy: lemme look real quick | 23:54 |
| vishy | which would be quite useful for us | 23:54 |
| vishy | it is in contrib/wrt/ | 23:54 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!