| yjiang5 | jroll: Hi, currently with iscsi deploy, if we re-install the ironic-conductor, all the compute-node that was provisioned before can't be reboot anymore because it dpends on the tftp boot to provide the kernel/ramdisk, am I right? Can IPA resolve this issue? | 00:02 |
|---|---|---|
| jroll | yjiang5: IPA only does boot from disk, so that is not a problem | 00:03 |
| yjiang5 | jroll: thanks for your confirmation. I will switch to IPA, now! | 00:03 |
| jroll | lolol | 00:04 |
| jroll | JayF: found new marketing material for IPA ^ | 00:04 |
| yjiang5 | jroll: :) | 00:05 |
| *** Marga_ has joined #openstack-ironic | 00:05 | |
| openstackgerrit | Arun S A G proposed a change to openstack/python-ironicclient: Fix python-ironicclient crash https://review.openstack.org/131307 | 00:06 |
| *** marcoemorais has joined #openstack-ironic | 00:06 | |
| yjiang5 | NobodyCam: I uploaded patch to convert the qcow2 to raw image on the fly, https://review.openstack.org/#/q/status:open+project:openstack/ironic+branch:master+topic:noconvert_iscsi,n,z . Please have a look. I'm not sure the patch to change the force_raw_images config is too invasive, so feedback is welcome. | 00:06 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/python-ironicclient: Revert "Add keystone v3 CLI support" https://review.openstack.org/131304 | 00:06 |
| *** marcoemorais has quit IRC | 00:07 | |
| *** marcoemorais has joined #openstack-ironic | 00:07 | |
| rloo | jroll: so there's another patch that fixes the CLI problem: https://review.openstack.org/131307 | 00:08 |
| rloo | jroll: did we want to revert? | 00:08 |
| NobodyCam | yjiang5: awesome I'll have a look | 00:08 |
| jroll | rloo: I still want to revert | 00:09 |
| yjiang5 | NobodyCam: thanks. | 00:09 |
| jroll | rloo: my main problem at this point is the original patch removed the timeout and insecure options | 00:09 |
| jroll | why? dunno. | 00:09 |
| rloo | jroll: ok, if you/we are sure, we should just let the other person/patch know. | 00:09 |
| jroll | zer0c00l: ^ | 00:10 |
| jroll | :) | 00:10 |
| jroll | rloo: I might be wrong, though, others might not think it's the right thing to do :) | 00:11 |
| zer0c00l | Arrgg | 00:11 |
| zer0c00l | :P | 00:11 |
| jroll | I do appreciate the patch either wauy | 00:11 |
| jroll | way | 00:11 |
| rloo | jroll: which patch. we can't approve both of them. | 00:12 |
| zer0c00l | NP. Thanks | 00:12 |
| jroll | rloo: I'm telling zer0c00l I appreciate 131307. | 00:12 |
| jroll | I know we can't approve both | 00:12 |
| devananda | we haven't had that many changes to the CLI since this, so reverting, fixing, and re-proposing isn't going to be hard | 00:12 |
| jroll | I've honestly been waiting for deva's opinion on this :P | 00:13 |
| rloo | my preference is to revert, just cuz that patch has so much stuff in it. (or maybe to give myself a chance to review but I don't think that's it cuz I don't really want to review it) | 00:13 |
| rloo | the fix in 131307 is easier to review, but if there other issues with the original patch maybe we should revert. | 00:14 |
| jroll | yeah | 00:15 |
| jroll | I want to stab the other patch | 00:15 |
| zer0c00l | so why the insecure and timeout was removed? | 00:15 |
| zer0c00l | Any clues? | 00:15 |
| devananda | i currently see two problems with e3d356476be376ebd3dfb3e84f379e113e5efac7 | 00:15 |
| devananda | zer0c00l: no idea. oversight on my part when approving it | 00:15 |
| jroll | no idea, zer0c00l | 00:15 |
| devananda | 1. removal of those options w/o any indication why | 00:16 |
| devananda | 2. even with all the right OS_* env vars, and not using IRONIC_URL (which actually we should probably stop using, even though it's a handy dev hack), it still fails with incorrect error messages | 00:16 |
| jroll | we need IRONIC_URL support :( | 00:17 |
| jroll | (we being rackspace team) | 00:17 |
| devananda | http://paste.openstack.org/show/125632/ | 00:17 |
| devananda | jroll: ugh | 00:17 |
| jroll | since ironic is an admin-only api, I'd argue it shouldn't be in the service catalog in a multi tenant environment | 00:17 |
| jroll | and I'll always argue that | 00:18 |
| jroll | sorry | 00:18 |
| *** marcoemorais has quit IRC | 00:18 | |
| jroll | wow, that's a great error. | 00:18 |
| rloo | devananda: so you're OK with reverting that patch? e3d356476be376ebd3dfb3e84f379e113e5efac7 | 00:18 |
| devananda | yea. user 'admin' passes auth. user 'demo' does not. but yay for bad errors | 00:18 |
| lifeless | jroll: so, I'll concede that ironic should be usable without being in the service catlog, but I don't agree that it should not be in it | 00:19 |
| *** Marga_ has quit IRC | 00:20 | |
| lifeless | jroll: I don't see the service catalog as being and end user feature. Its plumbing. | 00:20 |
| *** Marga_ has joined #openstack-ironic | 00:20 | |
| jroll | lifeless: so endpoints that are only accessible to certain folks should be there? | 00:20 |
| jroll | seems... weird | 00:21 |
| devananda | service catalog is both user and service facing | 00:21 |
| devananda | it is how users discover what services a cloud offers and how their clients should discover what URLs those services are accessible at | 00:21 |
| devananda | a cloud should only need to publish a single entrypoint, not one per service | 00:22 |
| jroll | right... our cloud doesn't offer ironic. | 00:22 |
| jroll | our internal cloud does | 00:22 |
| devananda | sure | 00:22 |
| devananda | so hide it in the service catalog from tenants taht don't have access to it | 00:22 |
| lifeless | jroll: so it should be in your internal cloud service catalog, and your nova etc should be using that cloud catalog | 00:23 |
| jroll | this is also me selfishly not wanting to infiltrate our identity team and fix our super broken service catalog | 00:23 |
| jroll | :) | 00:23 |
| devananda | and don't make the ironic endpoint accessible from the internet | 00:23 |
| devananda | jroll: exactly. THAT's the problem | 00:23 |
| * jroll hides | 00:23 | |
| lifeless | jroll: so, don't take this badly, but eternalising the cost onto the whole community is not great. | 00:23 |
| lifeless | s/etern/extern/ | 00:23 |
| devananda | jroll: that ^ | 00:23 |
| jroll | yes, I completely agree | 00:23 |
| jroll | however, I still think that's valuable functionality for a client | 00:24 |
| devananda | jroll: I don't entirely disagree | 00:24 |
| devananda | it's certainly very useful for my local debugging | 00:24 |
| jroll | for example, setting up a new cell with ironic in it, testing before linking it in | 00:24 |
| devananda | but we shouldn't be recommending that as the normal way to use our client, IMHO | 00:24 |
| jroll | I agree | 00:24 |
| jroll | and we don't | 00:25 |
| devananda | k k | 00:25 |
| devananda | so, case at hand, this patch | 00:25 |
| devananda | the two options it removed -- timeout and insecure -- I'm not sure how we didn't catch that | 00:26 |
| devananda | L400 is still accessing them | 00:27 |
| devananda | jroll: oh. they are still there | 00:28 |
| jroll | ? | 00:28 |
| devananda | jroll: ironic --help|grep insecure | 00:28 |
| jroll | from kssession.Session.register_cli_options(parser) | 00:29 |
| jroll | ? | 00:29 |
| jroll | yeah | 00:30 |
| jroll | https://github.com/openstack/python-keystoneclient/blob/master/keystoneclient/session.py#L686 | 00:30 |
| jroll | huh | 00:30 |
| devananda | yep | 00:30 |
| jroll | ok | 00:30 |
| jroll | do what you think is right here | 00:30 |
| devananda | jroll: so the error handling is bad in this patch, but it doesn't appear to have removed any options | 00:30 |
| jroll | yeah | 00:31 |
| * jroll feels the rage leave his body | 00:31 | |
| devananda | :) | 00:31 |
| *** marcoemorais has joined #openstack-ironic | 00:32 | |
| *** achanda has quit IRC | 00:32 | |
| *** achanda has joined #openstack-ironic | 00:33 | |
| *** marcoemorais has quit IRC | 00:36 | |
| *** achanda has quit IRC | 00:37 | |
| devananda | jroll: arun has posted a fix for the issue he and you found, right? | 00:48 |
| devananda | the one I found is something else, it seems | 00:49 |
| JayF | zer0c00l: ^ is the guy who originally found the issue iirc | 00:49 |
| JayF | and has a patch up | 00:49 |
| devananda | yes, that's arun | 00:49 |
| * harlowja nobody knows why he picked that name ;) | 00:50 | |
| devananda | heh | 00:50 |
| JayF | it sounds like it'd be from some bad hacker movie | 00:50 |
| harlowja | ;) | 00:50 |
| harlowja | with angelina or something | 00:50 |
| JayF | like with kids skateboarding and hacking mainframes over modems or something | 00:50 |
| devananda | jroll: anyway, is there another reason to revert, or just a remaining bug in the error handling? | 00:51 |
| *** foexle_ has quit IRC | 00:57 | |
| * zer0c00l was addicted to hackers.avi when i was a teenager | 00:58 | |
| zer0c00l | s/i/he | 00:58 |
| harlowja | hackers.avi, questionable | 00:59 |
| harlowja | ha | 00:59 |
| harlowja | i don't think that was the movie name, lol | 01:00 |
| *** yuanying has joined #openstack-ironic | 01:02 | |
| *** rloo has quit IRC | 01:03 | |
| devananda | the error is coming from http://git.openstack.org/cgit/openstack/python-ironicclient/tree/ironicclient/openstack/common/apiclient/exceptions.py#n446 | 01:11 |
| devananda | i dont have time to dig into this any more right now | 01:14 |
| spandhe | jroll: devananda , so when I create a hypervisor using “ironic node-create” it should create a row in nova.compute_nodes and ironic.nodes… is my understanding correct? Or do I need to do something special to create an entry in nova.computes? | 01:16 |
| devananda | spandhe: it creates the row in ironic.nodes immediately. nova will asynchronously create a row in nova.compute_nodes at the next execution of the resource tracker periodic task | 01:17 |
| devananda | spandhe: by default, I believe that runs every 60s | 01:17 |
| spandhe | devananda: oh, ok.. so somethings wrong with that task in my case.. let me check.. thanks! | 01:18 |
| devananda | or with n-cpu's configuration | 01:18 |
| devananda | check the n-cpu log | 01:18 |
| *** Marga_ has quit IRC | 01:20 | |
| *** foexle has joined #openstack-ironic | 01:23 | |
| *** Marga_ has joined #openstack-ironic | 01:24 | |
| *** Marga_ has quit IRC | 01:25 | |
| *** Marga_ has joined #openstack-ironic | 01:25 | |
| *** Marga_ has quit IRC | 01:26 | |
| *** nosnos has joined #openstack-ironic | 01:59 | |
| *** Marga_ has joined #openstack-ironic | 02:14 | |
| Haomeng | morning ironic:) | 02:27 |
| mrda | Hi Haomeng | 02:36 |
| Haomeng | mrda: morning:) | 02:36 |
| *** spandhe has quit IRC | 02:37 | |
| jroll | devananda: I'm not sure what the right thing to do is... I guess if zer0c00l's patch works, let's just roll forward | 02:45 |
| *** killer_prince is now known as lazy_prince | 03:06 | |
| *** ramineni has joined #openstack-ironic | 03:16 | |
| *** ChuckC___ has joined #openstack-ironic | 03:18 | |
| *** ChuckC___ has quit IRC | 03:27 | |
| *** harlowja is now known as harlowja_away | 03:27 | |
| *** harlowja_away is now known as harlowja | 03:30 | |
| *** nosnos has quit IRC | 03:36 | |
| *** nosnos has joined #openstack-ironic | 03:37 | |
| *** Poornima has joined #openstack-ironic | 03:38 | |
| *** marcoemorais has joined #openstack-ironic | 03:40 | |
| *** pcrews has quit IRC | 03:41 | |
| *** nosnos has quit IRC | 03:41 | |
| *** marcoemorais has quit IRC | 03:48 | |
| *** achanda has joined #openstack-ironic | 04:01 | |
| *** Nisha has joined #openstack-ironic | 04:04 | |
| *** nosnos has joined #openstack-ironic | 04:15 | |
| *** foexle_ has joined #openstack-ironic | 04:47 | |
| *** foexle has quit IRC | 04:51 | |
| *** Marga_ has quit IRC | 04:52 | |
| *** achanda has quit IRC | 04:54 | |
| *** achanda has joined #openstack-ironic | 04:56 | |
| *** spandhe has joined #openstack-ironic | 04:56 | |
| *** rameshg87 has joined #openstack-ironic | 05:19 | |
| *** Marga_ has joined #openstack-ironic | 05:23 | |
| *** harlowja is now known as harlowja_away | 05:37 | |
| *** lazy_prince is now known as killer_prince | 05:38 | |
| *** rakesh_hs has joined #openstack-ironic | 05:45 | |
| *** rameshg87 has quit IRC | 05:45 | |
| *** rameshg87 has joined #openstack-ironic | 05:46 | |
| *** loki184 has joined #openstack-ironic | 05:53 | |
| *** yuanying_ has joined #openstack-ironic | 06:02 | |
| *** yuanying has quit IRC | 06:06 | |
| *** achanda has quit IRC | 06:08 | |
| *** ChuckC___ has joined #openstack-ironic | 06:15 | |
| *** yuanying_ has quit IRC | 06:21 | |
| *** ChuckC___ has quit IRC | 06:21 | |
| *** yuanying has joined #openstack-ironic | 06:21 | |
| *** yuanying has quit IRC | 06:26 | |
| *** spandhe has quit IRC | 06:26 | |
| *** yuanying has joined #openstack-ironic | 06:26 | |
| *** yuanying has quit IRC | 06:42 | |
| *** yuanying has joined #openstack-ironic | 06:42 | |
| *** killer_prince is now known as lazy_prince | 06:46 | |
| *** chenglch has joined #openstack-ironic | 06:55 | |
| *** k4n0 has joined #openstack-ironic | 06:57 | |
| *** achanda has joined #openstack-ironic | 07:09 | |
| *** teju has joined #openstack-ironic | 07:13 | |
| *** achanda has quit IRC | 07:17 | |
| *** rushiagr_away is now known as rushiagr | 07:27 | |
| *** leopoldj has joined #openstack-ironic | 07:36 | |
| *** bmahalakshmi has joined #openstack-ironic | 07:37 | |
| *** bmahalakshmi has left #openstack-ironic | 07:41 | |
| *** andreykurilin_ has joined #openstack-ironic | 07:41 | |
| *** sandhya has joined #openstack-ironic | 07:41 | |
| *** bmahalakshmi has joined #openstack-ironic | 07:42 | |
| *** spandhe has joined #openstack-ironic | 07:47 | |
| *** pradipta_away is now known as pradipta | 07:54 | |
| *** bmahalakshmi has quit IRC | 08:02 | |
| GheRivero | morning @Ironic | 08:15 |
| *** jinxly has quit IRC | 08:26 | |
| *** foexle_ has quit IRC | 08:29 | |
| *** foexle has joined #openstack-ironic | 08:30 | |
| *** bmahalakshmi has joined #openstack-ironic | 08:35 | |
| *** rushiagr is now known as rushiagr_away | 08:40 | |
| *** ndipanov_gone is now known as ndipanov | 08:43 | |
| *** yuanying has quit IRC | 08:53 | |
| *** yuanying has joined #openstack-ironic | 08:53 | |
| *** athomas has quit IRC | 08:55 | |
| *** andreykurilin_ has quit IRC | 08:56 | |
| *** Marga_ has quit IRC | 08:56 | |
| *** Marga__ has joined #openstack-ironic | 08:57 | |
| *** yuanying has quit IRC | 08:57 | |
| *** spandhe has quit IRC | 08:58 | |
| *** vinbs has joined #openstack-ironic | 09:10 | |
| *** achanda has joined #openstack-ironic | 09:15 | |
| *** derekh has joined #openstack-ironic | 09:16 | |
| *** lucasagomes has joined #openstack-ironic | 09:18 | |
| *** vinbs_ has joined #openstack-ironic | 09:19 | |
| *** achanda has quit IRC | 09:19 | |
| *** pradipta is now known as pradipta_away | 09:20 | |
| ramineni | lucasagomes: hi | 09:20 |
| *** vinbs has quit IRC | 09:28 | |
| *** vinbs_ is now known as vinbs | 09:28 | |
| lucasagomes | ramineni, hi there morning | 09:32 |
| lucasagomes | catching up with emails | 09:32 |
| ramineni | lucasagomes : morning :) | 09:35 |
| *** bmahalakshmi has quit IRC | 09:35 | |
| ramineni | lucasagomes: want to check regarding the comment on https://review.openstack.org/#/c/129529/1/specs/kilo/add-boot-mode-mgmt-interface.rst | 09:35 |
| *** bmahalakshmi has joined #openstack-ironic | 09:36 | |
| *** pelix has joined #openstack-ironic | 09:36 | |
| lucasagomes | ramineni, oh will do, I have to catch up with the specs, it was holiday yesterday here so I'm a bit behind | 09:36 |
| lucasagomes | but I will take a look soonish | 09:36 |
| ramineni | lucasagomes: oohok | 09:36 |
| ramineni | lucasagomes : "one problem with @abstractmethod is that it's not backward compat with drivers outside the tree" | 09:37 |
| lucasagomes | ramineni, yeah the decorator, it makes mandatory to all drivers inheriting from that interface to have to implement that method | 09:38 |
| lucasagomes | I think I suggested to leave it without the decorator, and by default raise an NotImplemented exception right? | 09:38 |
| ramineni | lucasagomes: yes , | 09:39 |
| ramineni | lucasagomes: otherwise we need to change all the drivers , to riase the error for the exception right | 09:41 |
| lucasagomes | ramineni, yes | 09:41 |
| ramineni | lucasagomes : got it .. sorry ..was thinking something else .. | 09:41 |
| lucasagomes | ramineni, and we have been thinking about the out-of-tree drivers as well and trying to avoid to break them too | 09:42 |
| ramineni | lucasagomes: what do you mean by out-of-tree drivers? | 09:42 |
| lucasagomes | drivers that are not part of the ironic codebase | 09:42 |
| lucasagomes | some company that wrote a private driver but didn't submit it upstream | 09:42 |
| ramineni | lucasagomes : ooh ok | 09:43 |
| openstackgerrit | Derek Higgins proposed a change to openstack/ironic: Switch to oslo.concurrency https://review.openstack.org/131375 | 09:43 |
| *** Marga__ has quit IRC | 09:44 | |
| *** jinxly has joined #openstack-ironic | 09:44 | |
| *** Marga_ has joined #openstack-ironic | 09:44 | |
| openstackgerrit | Anusha Ramineni proposed a change to openstack/ironic-specs: Add get/set boot mode to Management Interface https://review.openstack.org/129529 | 09:45 |
| *** chenglch has quit IRC | 10:03 | |
| openstackgerrit | Sam Betts proposed a change to openstack/ironic: Add logging to driver vendor_passthru functions https://review.openstack.org/129298 | 10:09 |
| openstackgerrit | Anusha Ramineni proposed a change to openstack/ironic-specs: Management Interface to firmware update https://review.openstack.org/100842 | 10:12 |
| openstackgerrit | Harshada Mangesh Kakad proposed a change to openstack/ironic-specs: Seamicro Serial Console https://review.openstack.org/131382 | 10:16 |
| *** mkerrin has joined #openstack-ironic | 10:16 | |
| *** Nisha has quit IRC | 10:20 | |
| openstackgerrit | Derek Higgins proposed a change to openstack/ironic: Switch to oslo.concurrency https://review.openstack.org/131375 | 10:22 |
| *** jinxly has quit IRC | 10:29 | |
| *** jinxly has joined #openstack-ironic | 10:30 | |
| openstackgerrit | Anusha Ramineni proposed a change to openstack/ironic-specs: Management Interface to firmware update https://review.openstack.org/100842 | 10:33 |
| *** loki184 has quit IRC | 10:41 | |
| *** MattMan has joined #openstack-ironic | 10:42 | |
| *** sambetts has joined #openstack-ironic | 10:42 | |
| *** sandhya has quit IRC | 10:53 | |
| *** ramineni has quit IRC | 10:56 | |
| *** dlaube has quit IRC | 10:59 | |
| *** dlaube has joined #openstack-ironic | 11:00 | |
| *** jinxly has quit IRC | 11:06 | |
| openstackgerrit | A change was merged to openstack/ironic: Put nodes-related API in same section https://review.openstack.org/131202 | 11:11 |
| *** Marga_ has quit IRC | 11:18 | |
| *** Marga_ has joined #openstack-ironic | 11:18 | |
| *** athomas has joined #openstack-ironic | 11:25 | |
| openstackgerrit | John Trowbridge proposed a change to openstack/ironic: Changes default IPMI retry timeout to 15 seconds https://review.openstack.org/131296 | 11:26 |
| *** rameshg87 has quit IRC | 11:31 | |
| *** Poornima has quit IRC | 11:38 | |
| *** Marga_ has quit IRC | 11:40 | |
| *** Haomeng|2 has joined #openstack-ironic | 11:41 | |
| lucasagomes | jroll, https://review.openstack.org/#/c/130513/ when you get some time | 11:41 |
| lucasagomes | devananda, ^ | 11:41 |
| *** Haomeng has quit IRC | 11:41 | |
| openstackgerrit | A change was merged to openstack/ironic: Use docstrings for attributes in api/controllers https://review.openstack.org/131286 | 11:42 |
| *** teju has quit IRC | 11:46 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/131111 | 11:47 |
| *** bmahalakshmi has quit IRC | 11:47 | |
| *** pelix has quit IRC | 11:50 | |
| *** pelix has joined #openstack-ironic | 11:52 | |
| openstackgerrit | Derek Higgins proposed a change to openstack/ironic: Switch to oslo.concurrency https://review.openstack.org/131375 | 11:53 |
| *** pelix has quit IRC | 11:53 | |
| derekh | lucasagomes: ^^ is a more complete version, I had been trying to just do what was needed to get tripleo ci working again by no longer using the lockutils from the novatree | 11:54 |
| derekh | lucasagomes: but these version should make the complete switch | 11:54 |
| lucasagomes | derekh, oh great, I will take a look | 11:55 |
| lucasagomes | thanks for updating it all | 11:55 |
| derekh | lucasagomes: no prob | 11:56 |
| *** igordcard has joined #openstack-ironic | 11:58 | |
| openstackgerrit | Sam Betts proposed a change to openstack/ironic: Remove duplicated _fetch_images function https://review.openstack.org/131405 | 12:11 |
| *** bradjones has quit IRC | 12:12 | |
| *** bradjones has joined #openstack-ironic | 12:13 | |
| *** jinxly has joined #openstack-ironic | 12:24 | |
| *** igordcard has quit IRC | 12:27 | |
| *** ryanpetrello has quit IRC | 12:34 | |
| *** ryanpetrello has joined #openstack-ironic | 12:34 | |
| *** dprince has joined #openstack-ironic | 12:35 | |
| *** linggao has joined #openstack-ironic | 12:46 | |
| *** k4n0 has quit IRC | 12:46 | |
| *** todd_dsm has joined #openstack-ironic | 12:52 | |
| *** lucasagomes is now known as lucas-hungry | 12:52 | |
| jroll | lucas-hungry: +2, leaving it for deva and also left a couple nits :) | 12:57 |
| derekh | Can anybody else review https://review.openstack.org/#/c/131375/3 please, as without it tripleo CI is busted | 12:58 |
| *** rakesh_hs has quit IRC | 12:58 | |
| jroll | derekh: sure | 13:01 |
| jroll | derekh: this is the only part that's breaking tripleo ci, right? | 13:01 |
| jroll | https://review.openstack.org/#/c/131375/3/ironic/nova/compute/manager.py | 13:01 |
| *** ryanpetrello has quit IRC | 13:02 | |
| derekh | jroll: correct, patch set 2 changed only that | 13:02 |
| sambetts | Shrews: Hi! o/ | 13:02 |
| jroll | aha, neat | 13:02 |
| Shrews | sambetts: hi! | 13:03 |
| openstackgerrit | Dmitry Nikishov proposed a change to openstack/ironic: Removed hardcoded IDs from "port" test resources https://review.openstack.org/131087 | 13:03 |
| Shrews | derekh: looks pretty straight-forward to me | 13:04 |
| jroll | derekh: +2+A | 13:04 |
| jroll | cheers | 13:04 |
| * jroll brb | 13:05 | |
| derekh | jroll: thanks, just in time too, I'm off for the rest of the day, tripleo ci jobs should start passing once that is merged | 13:05 |
| jroll | cool | 13:05 |
| *** ryanpetrello has joined #openstack-ironic | 13:05 | |
| *** ryanpetrello has quit IRC | 13:06 | |
| *** derekh has quit IRC | 13:06 | |
| *** igordcard has joined #openstack-ironic | 13:07 | |
| sambetts | Shrews: I totally agree with your comment, which is sort of the reason I designed the function the way it is now, however I couldn't think of a neat place to put it as a decorator because of the way the passthru works, so I created my function as a sort of psudo decorator | 13:07 |
| Shrews | sambetts: couldn't put it in the base class .py file? | 13:08 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Fix python-ironicclient crash https://review.openstack.org/131307 | 13:08 |
| Shrews | sambetts: i haven't thought the use of a decorator 100% thru, so your way may be best. i just want to make sure we've considered the possibility | 13:09 |
| sambetts | Shrews: the blocker I see in my head is where to put the "@somedecoratorname" because of the dynamic nature of vendor_passthru | 13:10 |
| Shrews | sambetts: Could we put it wherever "func" is defined? Maybe not since we also want "method" logged | 13:11 |
| *** ryanpetrello has joined #openstack-ironic | 13:11 | |
| Shrews | maybe that's reason enough if that's a requirement | 13:12 |
| *** rushiagr_away is now known as rushiagr | 13:12 | |
| * Shrews goes to retrieve his coffee to help his brain muscle wake up | 13:12 | |
| jroll | sambetts, Shrews, what are y'all looking at? | 13:16 |
| sambetts | Shrews: An issue I see with putting it where func is defined is when func is not specific to just the vendor passthru, leading to incorrect logging in other cases | 13:17 |
| sambetts | jroll: https://review.openstack.org/#/c/129298/7 | 13:17 |
| Shrews | sambetts: got an example of that? | 13:17 |
| *** bmahalakshmi has joined #openstack-ironic | 13:18 | |
| sambetts | Shrews: just looking to see if there are any | 13:18 |
| *** achanda has joined #openstack-ironic | 13:18 | |
| Shrews | jroll: your thoughts on that would be appreciated, too. we're just thinking this through :) | 13:18 |
| jroll | yep, looking :) | 13:18 |
| Shrews | i'm mainly achieving my "Being Difficult Way Too Early in the Morning" badge | 13:19 |
| Shrews | :) | 13:19 |
| sambetts | ironic/drivers/modules/ilo/deploy.py L566 set_boot_device is used in many other places | 13:19 |
| jroll | lol | 13:20 |
| jroll | sambetts: I also like the decorator idea | 13:20 |
| jroll | and for set_boot_device, you could make like a vendor_set_boot_device that wraps set_boot_device | 13:21 |
| jroll | and decorate that and use it when it's coming from the passthru | 13:21 |
| Shrews | jroll: ooh, that's a good idea | 13:21 |
| jroll | also... this is part of lucas-hungry's spec | 13:21 |
| jroll | basically | 13:21 |
| jroll | he calls out this problem explicitly | 13:21 |
| jroll | one thing that could be really neat, that we do similar in IPA | 13:22 |
| jroll | is decorate with @passthru(LOG, 'method') | 13:22 |
| jroll | and have that add to vendor_routes automatically | 13:22 |
| jroll | and/or have vendor_routes calculated from this | 13:22 |
| * jroll tries to find this | 13:23 | |
| *** achanda has quit IRC | 13:23 | |
| jroll | https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/base.py#L120-124 | 13:23 |
| jroll | combined with this decorator (and one other) https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/base.py#L199-200 | 13:24 |
| Shrews | jroll: what's that spec? | 13:25 |
| Shrews | Extended Vendor Passthru? | 13:25 |
| jroll | Shrews: https://review.openstack.org/#/c/130513/ | 13:25 |
| jroll | yeah | 13:25 |
| jroll | check out the paragraph at line 44 | 13:25 |
| Shrews | ah | 13:26 |
| Shrews | sambetts: seems like you might want to work with lucas-hungry on this, then | 13:26 |
| Shrews | i like jroll's ideas | 13:26 |
| *** ryanpetrello_ has joined #openstack-ironic | 13:26 | |
| jroll | I think I like @register_passthru('method') | 13:27 |
| jroll | I think I'm ok with the LOG coming from base.py | 13:27 |
| sambetts | Shall I initially work on changing my patch over into an actual decorator then? which can then be extend in the future to automate things | 13:27 |
| jroll | it's only logging exceptions anyway, so you'll get a full traceback | 13:27 |
| sambetts | ok that makes sense | 13:27 |
| jroll | I mean | 13:27 |
| jroll | I tend to think so | 13:27 |
| jroll | I also haven't had enough coffee :) | 13:28 |
| Shrews | sambetts: however you do it, you should make sure that it meshes with how lucas-hungry sees it working | 13:28 |
| *** rloo has joined #openstack-ironic | 13:29 | |
| Shrews | sambetts: thanks for taking the time to discuss it | 13:29 |
| *** shikui__ has quit IRC | 13:30 | |
| sambetts | I've seen a couple of reviews from lucas-hungry this one I think is most relevant https://review.openstack.org/#/c/129261/6 | 13:30 |
| jroll | oh yeah | 13:31 |
| jroll | I think that's going to end up being similar | 13:31 |
| jroll | or at least having this functionality | 13:32 |
| * jroll looks | 13:32 | |
| openstackgerrit | A change was merged to openstack/ironic: Switch to oslo.concurrency https://review.openstack.org/131375 | 13:32 |
| jroll | sambetts: yeah, lucas covers this here: https://review.openstack.org/#/c/129261/10/ironic/conductor/manager.py | 13:35 |
| Shrews | jroll: crap. i didn't see your client revert before i approved https://review.openstack.org/131307 | 13:36 |
| jroll | lol | 13:36 |
| jroll | Shrews: it's fine, we talked about it last night | 13:36 |
| Shrews | *phew* | 13:36 |
| * jroll abandons | 13:36 | |
| jroll | lucas-hungry: these need to cut over to the BP and not the bug, yeah? https://review.openstack.org/#/q/status:open+project:openstack/ironic+branch:master+topic:bug/1382457-vendor_passthru,n,z | 13:40 |
| jroll | um | 13:41 |
| jroll | https://review.openstack.org/#/c/131416/ | 13:41 |
| jroll | Shrews: chances the same thing just happened to us? | 13:41 |
| *** lucas-hungry is now known as lucasagomes | 13:41 | |
| Shrews | *sigh* | 13:41 |
| jroll | https://review.openstack.org/#/c/131375/ | 13:41 |
| * lucasagomes long long haha | 13:41 | |
| lucasagomes | long log* | 13:42 |
| Shrews | jroll: http://logs.openstack.org/75/131375/3/gate/gate-ironic-python27/e5b14c1/testr_results.html.gz | 13:42 |
| Shrews | so, yep | 13:42 |
| Shrews | i mean | 13:43 |
| Shrews | nope | 13:43 |
| jroll | yay. | 13:43 |
| jroll | although | 13:43 |
| jroll | https://review.openstack.org/#/c/131416/1/requirements.txt | 13:43 |
| lucasagomes | jroll, thanks | 13:43 |
| jroll | tripleo might be broken again now? | 13:44 |
| lucasagomes | jroll, yes I created the bp after the bug, I need to update the patches | 13:44 |
| lucasagomes | but I'm waiting the bp to be approved first | 13:44 |
| jroll | dunno how they install things on their CI systems | 13:44 |
| jroll | lucasagomes: sure | 13:44 |
| lucasagomes | I will tho fix the first patch of the series | 13:44 |
| jroll | lucasagomes: you see that you still need to make the BP? | 13:44 |
| lucasagomes | jroll, what u mean? | 13:44 |
| lucasagomes | in launchpad? I didn't? | 13:45 |
| * lucasagomes checks | 13:45 | |
| jroll | the url in the spec didn't work, anyway | 13:45 |
| lucasagomes | oh, right gotta fix that then | 13:45 |
| lucasagomes | https://blueprints.launchpad.net/ironic/+spec/extended-vendor-passthru | 13:46 |
| lucasagomes | I will update the blueprint as wel | 13:46 |
| lucasagomes | well* | 13:46 |
| Shrews | jroll: just quickly looking at nova's testing, they run their tests differently from the way ironic does it. | 13:47 |
| jroll | yeah | 13:47 |
| Shrews | via run_tests.sh | 13:47 |
| lucasagomes | yeah, I think we had a run_test.sh script before | 13:50 |
| lucasagomes | and then we got rid of it | 13:50 |
| lucasagomes | actually when I run tests in nova I just use tox -e | 13:50 |
| *** pelix has joined #openstack-ironic | 13:51 | |
| *** pradipta_away is now known as pradipta | 13:52 | |
| *** ryanpetrello has quit IRC | 13:59 | |
| *** ryanpetrello_ is now known as ryanpetrello | 13:59 | |
| NobodyCam | good morning Ironic | 14:02 |
| lucasagomes | NobodyCam, morning | 14:03 |
| NobodyCam | morning lucasagomes :) | 14:03 |
| jroll | morning NobodyCam :) | 14:03 |
| Shrews | hey NobodyCam | 14:03 |
| NobodyCam | morning jroll ans Shrews :) | 14:04 |
| *** spandhe has joined #openstack-ironic | 14:04 | |
| *** spandhe has quit IRC | 14:04 | |
| *** nosnos has quit IRC | 14:06 | |
| *** nosnos has joined #openstack-ironic | 14:07 | |
| devananda | g'morning, all | 14:10 |
| jroll | heya devananda | 14:10 |
| lucasagomes | NobodyCam, jroll btw was talking with a friend about it... If you guys get the metro in paris, watch ur stuff, it's full of freaks and pickpocket | 14:11 |
| lucasagomes | devananda, ^ u too | 14:11 |
| lucasagomes | devananda, morning | 14:11 |
| *** nosnos has quit IRC | 14:11 | |
| jroll | fun, thanks lucasagomes | 14:12 |
| *** pradipta is now known as pradipta_away | 14:12 | |
| lucasagomes | yeah ppl think it's fine because its europe and all, but paris is pretty bad about it afaict | 14:13 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Clear maintenance_reason when setting maintenance=False https://review.openstack.org/129694 | 14:13 |
| lucasagomes | devananda, when you get a time please mind looking at https://review.openstack.org/#/c/130513/ ? | 14:13 |
| jroll | lucasagomes: ^ fixed tests on that, you +2'd before | 14:13 |
| NobodyCam | good morning devananda | 14:14 |
| lucasagomes | jroll, will take a look | 14:14 |
| NobodyCam | lucasagomes: good to know!!!! | 14:14 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic-specs: New stateless iPXE driver https://review.openstack.org/130812 | 14:15 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/python-ironicclient: Add node-set-maintenance command https://review.openstack.org/129693 | 14:15 |
| jroll | fixed this guy too ^ | 14:15 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic-specs: New stateless iPXE driver https://review.openstack.org/130812 | 14:16 |
| lucasagomes | jroll, oh haven't seem that _state there | 14:19 |
| lucasagomes | in the client | 14:19 |
| lucasagomes | good catch | 14:19 |
| jroll | heh yeah | 14:19 |
| jroll | tests caught it, not me :P | 14:19 |
| lucasagomes | heh | 14:20 |
| lucasagomes | good | 14:20 |
| lucasagomes | means the change is covered :) | 14:20 |
| NobodyCam | :) w00t for tests | 14:21 |
| *** Marga_ has joined #openstack-ironic | 14:21 | |
| rloo | hi everyone | 14:21 |
| NobodyCam | good morning rloo :) | 14:22 |
| rloo | jroll: do we need a DocImpact for 129694? | 14:22 |
| jroll | maybe? | 14:22 |
| jroll | I don't know | 14:22 |
| jroll | and good morning :) | 14:22 |
| rloo | jroll: cuz we're deprecating? | 14:22 |
| rloo | jroll: ok, in that case, I'd like to have it since it wasn't mentioned in the BP | 14:22 |
| jroll | probably, I guess | 14:22 |
| jroll | just add to commit msg? | 14:22 |
| jroll | "DocImpact" | 14:23 |
| rloo | jroll: yeah. | 14:23 |
| jroll | right? | 14:23 |
| jroll | ok | 14:23 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Clear maintenance_reason when setting maintenance=False https://review.openstack.org/129694 | 14:23 |
| lucasagomes | rloo, morning | 14:23 |
| jroll | done | 14:23 |
| rloo | thx jroll! | 14:23 |
| jroll | bye bye +2s :( | 14:23 |
| rloo | afternoon lucasagomes | 14:23 |
| lucasagomes | jroll, will re-add | 14:23 |
| rloo | no worries jroll. we have lots of +2s to hand out | 14:23 |
| jroll | woohoo | 14:24 |
| Shrews | rloo: yeah, good idea there. /me re-adds +2 | 14:24 |
| rloo | see jroll, you can be happy again :-) | 14:24 |
| devananda | so, hey, trying to find a dinner location for monday and I need a rough head count. | 14:24 |
| jroll | devananda: count me in, I figure JayF and JoshNang would also be in | 14:25 |
| devananda | I'm assuming that we'll have more than just the folks who were in the meeting yesterday | 14:25 |
| lucasagomes | devananda, I'm in too | 14:25 |
| lucasagomes | and I believe dtantsur will go too | 14:25 |
| devananda | non-cores, etc, as well ? | 14:25 |
| lucasagomes | sure | 14:25 |
| rloo | +1 for me. | 14:25 |
| devananda | so 20+ person group? | 14:25 |
| lucasagomes | oh wow all that? | 14:25 |
| rloo | that's a lot. | 14:25 |
| rloo | do we have that many non cores? ;) | 14:26 |
| devananda | i dunno - that's why I'm asking :) | 14:26 |
| rloo | do you need to know today? | 14:26 |
| rloo | maybe (ugh) send email and ask? | 14:26 |
| devananda | I don't want to put it on the ML, because then it will get much much larger .... | 14:26 |
| rloo | yeah, cuz everyone loves ironic and partying | 14:27 |
| NobodyCam | if able to make it i would have a +1 with /me | 14:27 |
| rloo | i'd just hate to exclude anyone. but 20 seems like a lot. how many were we in atlanta at that pizza place. i'm guessing there will be fewer in paris. | 14:27 |
| rloo | NobodyCam: John going to be there? | 14:28 |
| NobodyCam | yep! | 14:28 |
| rloo | NobodyCam: cool! | 14:28 |
| NobodyCam | :) (happydance) | 14:28 |
| devananda | rloo: why fewer? our team is larger now. I think we had 18, and nova had about 20 at the table next to us | 14:28 |
| rloo | devananda: guessing cuz it is in Paris, fewer people will be there. Like no Shrews | 14:29 |
| rloo | devananda: but I could be wrong. | 14:29 |
| devananda | rloo: we were missing several EU folks in atlanta, too | 14:29 |
| NobodyCam | and I roman will be there too | 14:29 |
| Shrews | well, Paris is _certainly_ no Atlanta :) | 14:29 |
| NobodyCam | gah | 14:30 |
| rloo | true. could always book for 20. will it matter if there are fewer? | 14:30 |
| lucasagomes | well it's better to have more places just in case | 14:30 |
| devananda | yes | 14:30 |
| lucasagomes | so 20 is it | 14:30 |
| NobodyCam | romcheg will be there too i should have said | 14:30 |
| Shrews | devananda: you should do Eventbrite tickets | 14:31 |
| devananda | capping around 10 makes the choices of restaurant much broader. having a loose number of about 20 makes the choices smaller, though I think I found one, but have not heard back from them | 14:31 |
| NobodyCam | :) | 14:31 |
| rloo | ohh in that case, cap at 10. ha ha. | 14:31 |
| devananda | heh | 14:31 |
| lucasagomes | lol | 14:31 |
| devananda | it also means better food | 14:31 |
| rloo | oh, now I think we should really cap it at 10. | 14:31 |
| devananda | :) | 14:32 |
| lucasagomes | btw there's a very cool restaurant in paris, but def doesn't fit a group of 20 | 14:32 |
| devananda | lucasagomes: only one? ;) | 14:32 |
| lucasagomes | it's very old school, the way the cook/prepare the food | 14:32 |
| lucasagomes | devananda, http://www.polidor.com/ | 14:32 |
| rloo | i'm even fine not going if there are more than 10. I think better food should win | 14:32 |
| lucasagomes | the site is kinda crap, but I've been there and the food is pretty nice | 14:32 |
| lucasagomes | many locals go there, so | 14:33 |
| rloo | lucasagomes: even reasonable prices! but is that walking distance? (they open at 7pm) | 14:33 |
| lucasagomes | rloo, well, the thing about paris is that the metro system is very good | 14:34 |
| lucasagomes | I haven't check to see if it's near the venue (idk where the venue is yet :P) | 14:34 |
| lucasagomes | lemme check | 14:34 |
| lucasagomes | very good = goes everywhere | 14:34 |
| NobodyCam | brb | 14:34 |
| *** lazy_prince is now known as killer_prince | 14:36 | |
| lucasagomes | rloo, so not a walkable distance :/ it's 6km | 14:37 |
| rloo | lucasagomes: I'm fine with it but people might want something near the Palais du Congres. There's a bunch of stuff happening there or from there or whatever. | 14:38 |
| lucasagomes | yeah, sure | 14:38 |
| rloo | lucasagomes: and then there's the 10 vs 20 bodies issue ;) | 14:39 |
| lucasagomes | and I don't even know how to book there, probably u have to call and hope that someone in the restaurant speak english | 14:39 |
| lucasagomes | yeah, maybe another time with a smaller group | 14:39 |
| *** PaulCzar has joined #openstack-ironic | 14:40 | |
| *** achanda has joined #openstack-ironic | 14:48 | |
| PaulCzar | hey folks, I'm trying to run haproxy in front of ironic and having issues getting an appropriate httpchk to work ... everything I've tried to far results in the following in the ironic-api logs and the haproxy refusing to balance it - https://gist.github.com/paulczar/810881bd204cbf273168 | 14:49 |
| lucasagomes | PaulCzar, hmm, what curl command r u using to test it? | 14:54 |
| *** zz_jgrimm is now known as jgrimm | 14:58 | |
| PaulCzar | lucasagomes: I can run 'curl -X GET http://openstack.example.com:6385/v1' and it responds appropriately | 15:02 |
| PaulCzar | but using `option httpchk GET /v1` in haproxy.cfg i get those errors | 15:03 |
| jroll | it probably disconnects before the body is sent :/ | 15:03 |
| devananda | lucasagomes: a few nits on the passthru changes. are you spinning another rev of it? | 15:03 |
| *** foexle_ has joined #openstack-ironic | 15:03 | |
| lucasagomes | devananda, yup fixing the link and stuff, wanted to get ur opnion before spinning it | 15:04 |
| PaulCzar | jroll: that's what I'm suspecting ... yet to find a option to tell haproxy to wait a while longer | 15:04 |
| jroll | PaulCzar: try HEAD maybe? | 15:04 |
| *** foexle has quit IRC | 15:07 | |
| PaulCzar | jroll: yeah, tried HEAD and OPTIONS | 15:08 |
| PaulCzar | same thing | 15:08 |
| jroll | huh | 15:08 |
| devananda | lucasagomes: comments posted | 15:08 |
| PaulCzar | well actually OPTIONS gives a 404 | 15:08 |
| *** achanda has quit IRC | 15:09 | |
| *** vinbs has quit IRC | 15:13 | |
| JayF | hi PaulCzar :) | 15:17 |
| *** pcrews has joined #openstack-ironic | 15:18 | |
| devananda | I'm going to be only slightly around for the next few hours, then offline the rest of the day - flying to paris! (yes, early) | 15:19 |
| JayF | PaulCzar: I presume haproxy has a TCP check? That should work to give you basic functionality | 15:19 |
| devananda | also, I'll be in EU time for the rest of the week | 15:19 |
| jroll | devananda: woot, enjoy :) | 15:19 |
| devananda | if anyone's looking for client bugs to fix -- https://bugs.launchpad.net/python-ironicclient/+bug/1386470 | 15:20 |
| rloo | devananda: you can scout out the restaurants! | 15:24 |
| NobodyCam | have a great flight devananda :) | 15:24 |
| jroll | devananda: seems like a candidate bug for an in-flight hackathon | 15:25 |
| *** Marga_ has quit IRC | 15:30 | |
| *** jlk has joined #openstack-ironic | 15:35 | |
| PaulCzar | hey JayF ... thanks that's probably the next thing I try | 15:36 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add node-set-maintenance command https://review.openstack.org/129693 | 15:41 |
| *** sambetts_ has joined #openstack-ironic | 15:42 | |
| lucasagomes | devananda, awesome thanks will take a look at the comments | 15:43 |
| *** sambetts has quit IRC | 15:43 | |
| openstackgerrit | Sam Betts proposed a change to openstack/ironic: Add logging to driver vendor_passthru functions https://review.openstack.org/129298 | 15:44 |
| sambetts_ | Shrews, jroll: ^ is that more like what you guys had in mind | 15:45 |
| sambetts_ | lucasagomes: ^ I've done quite a few changes to my logging patch what do you think? | 15:45 |
| lucasagomes | sambetts_, hi there, will take a look | 15:46 |
| jroll | sambetts_: yeah, something like that, nie :) | 15:48 |
| jroll | nice | 15:48 |
| sambetts_ | then in the future I guess it could be extended to include some of the routing stuff etc. | 15:48 |
| jroll | indeed | 15:49 |
| NobodyCam | _LE is for error not exception right? http://docs.openstack.org/developer/oslo.i18n/guidelines.html | 15:51 |
| jroll | both | 15:51 |
| jroll | afaik | 15:51 |
| NobodyCam | :( crys | 15:51 |
| jroll | # RIGHT | 15:51 |
| *** bmahalakshmi has quit IRC | 15:51 | |
| jroll | LOG.exception(_LE('There was an error.')) | 15:51 |
| jroll | I mean | 15:52 |
| jroll | they're both error | 15:52 |
| jroll | one has a traceback attached | 15:52 |
| *** hemna has joined #openstack-ironic | 15:54 | |
| NobodyCam | root@ubuntu:~# swift list | 16:00 |
| NobodyCam | ('Connection aborted.', error(111, 'Connection refused')) | 16:00 |
| NobodyCam | humm something still not right | 16:01 |
| *** athomas has quit IRC | 16:03 | |
| rloo | NobodyCam, jroll: yes, LE is for both error and exception. It is mentioned in that Note box. | 16:07 |
| rloo | NobodyCam: I had them update that wiki wrt LE and they put it in the Note instead of under the 'Level'. I was wondering if it was clear enough. I guess not. | 16:07 |
| *** marcoemorais has joined #openstack-ironic | 16:08 | |
| *** Marga_ has joined #openstack-ironic | 16:08 | |
| NobodyCam | rloo: no I think that is clear | 16:09 |
| NobodyCam | I stopped reading at the table | 16:09 |
| NobodyCam | :( | 16:09 |
| NobodyCam | my bad | 16:09 |
| rloo | NobodyCam: No, not your bad. That was my concern, that people would stop reading at the table. I guess I should have pointed it out to them but I didn't. I was just happy they were willing to update the wiki. | 16:10 |
| jroll | you can update the wiki | 16:11 |
| NobodyCam | :-p | 16:12 |
| *** Marga_ has quit IRC | 16:14 | |
| rloo | jroll: I don't think so. I think that was generated. (I reviewed their changes, so there was a patch.) | 16:14 |
| lucasagomes | sambetts_, jroll Shrews I like the decorator... but one thing, if we are going that path to make the vendor_routes() driver_routes() to be dynamic generated | 16:14 |
| rloo | jroll: I could be wrong though; it was awhile ago. But I recall reviewing something for them ;) | 16:14 |
| lucasagomes | we also would need a metaclass for that | 16:14 |
| jroll | oh, right, that's a docs page | 16:14 |
| *** Marga_ has joined #openstack-ironic | 16:15 | |
| lucasagomes | to actually "register" the methods | 16:15 |
| lucasagomes | idk how black magic it will look like | 16:15 |
| jroll | it doesn't need to be a metaclass :P | 16:15 |
| jroll | but that might be the better way | 16:15 |
| lucasagomes | jroll, no? I can't think about any other way | 16:15 |
| jroll | attach an attribute that you look for | 16:16 |
| lucasagomes | jroll, to the method right? | 16:16 |
| jroll | https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/base.py#L120-124 | 16:16 |
| jroll | yeah | 16:16 |
| jroll | it's dirty :) | 16:16 |
| jroll | the decorators that do that are at the bottom of that file | 16:17 |
| jroll | we need to attach the command name anyway so it isn't as bad here | 16:17 |
| lucasagomes | right, yeah you can inspect the attributes of the class and find a marker | 16:17 |
| *** Marga_ has quit IRC | 16:19 | |
| *** killer_prince has quit IRC | 16:21 | |
| sambetts_ | I think thats quite a neat way of doing it, we'd have to either add a second decorator or some kinda of flag to destinguish between the vendor passthru and driver passthru | 16:22 |
| *** killer_prince has joined #openstack-ironic | 16:22 | |
| sambetts_ | but that's an easy extension | 16:22 |
| *** killer_prince is now known as lazy_prince | 16:22 | |
| *** sambetts_ is now known as sambetts | 16:24 | |
| lucasagomes | sambetts, right yeah I can modify the spec to talk about it | 16:30 |
| lucasagomes | jroll, rloo devananda ^ r u guys ok with it? | 16:31 |
| *** Marga_ has joined #openstack-ironic | 16:31 | |
| lucasagomes | so instead of having the vendor_routes() driver_routes() we have a decorator on the methods which then will create those routes dynamically? | 16:31 |
| NobodyCam | lucasagomes: I like that idea | 16:32 |
| *** marcoemorais has quit IRC | 16:32 | |
| NobodyCam | dynamically is good | 16:32 |
| *** marcoemorais has joined #openstack-ironic | 16:32 | |
| *** rbowen has joined #openstack-ironic | 16:33 | |
| rloo | lucasagomes: I wasn't paying attention. In theory, it sounds good. Let me think. Does sambetts' change reflect that? | 16:33 |
| rbowen | NobodyCam: lucasagomes reported the problem you had on the RDO website. It appears to have gone away ... or are you still seeing the problem? | 16:33 |
| rbowen | NobodyCam: I presume the database server must have gone away briefly, but I don't see anything promising in the logs to suggest why. | 16:33 |
| lucasagomes | rloo, yeah, the new changes does create a decorator to runs the vendor methods | 16:34 |
| openstackgerrit | Sam Betts proposed a change to openstack/ironic: Remove duplicated _fetch_images function https://review.openstack.org/131405 | 16:34 |
| NobodyCam | rbowen: nope its all working now... | 16:36 |
| lucasagomes | rloo, lemme know what u think | 16:36 |
| rbowen | ok, cool. | 16:36 |
| *** Marga_ has quit IRC | 16:36 | |
| *** pensu has joined #openstack-ironic | 16:37 | |
| *** Nisha has joined #openstack-ironic | 16:38 | |
| *** marcoemorais has quit IRC | 16:41 | |
| *** harlowja_away is now known as harlowja | 16:46 | |
| rloo | lucasagomes: yeah, that seems fine to me. You'd modify the decorator fn to add that info to a dict, right? | 16:47 |
| lucasagomes | rloo, hmm no, as the dict should be per vendor class but the decorator has no idea about the class that function lives | 16:48 |
| lucasagomes | rloo, we should on the init method of the class inspect it's attributes and find the methods decorated by that decorator | 16:48 |
| lucasagomes | and build the dict from that | 16:48 |
| lucasagomes | something like jroll posted here https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/base.py#L120-124 | 16:49 |
| rloo | lucasagomes: oh, was wondering how you'd get the info. | 16:49 |
| lucasagomes | another way to do that would be to use a metaclass, but seems overkill | 16:49 |
| lucasagomes | rloo, right yeah | 16:50 |
| rloo | lucasagomes: where's the description of the method. I guess that can be added to the decorator. | 16:50 |
| rloo | lucasagomes: (just thinking of the API part to return the available methods, http method...) | 16:51 |
| lucasagomes | rloo, yeah, if I'm going to extend it, the description of the method should go in the decorator | 16:51 |
| lucasagomes | as well as things like what http methods it supports, sync/async etc | 16:51 |
| lucasagomes | rloo, if u want to make it more dynamic we can even use the docstring as the description | 16:52 |
| lucasagomes | but idk if that's too much blackmagic | 16:52 |
| lucasagomes | because u can get the docstring via __doc__ | 16:52 |
| openstackgerrit | A change was merged to openstack/ironic: Updated from global requirements https://review.openstack.org/131111 | 16:53 |
| sambetts | mmmm that python blackmagic haha | 16:53 |
| lucasagomes | :D | 16:53 |
| rloo | lucasagomes: so the conductor.manager doesn't care; it'll still access a dictionary. it is the base driver. A change to that API. I don't see pros/cons to either way, that would sway me either way. So I think I'm fine either way. | 16:53 |
| lucasagomes | rloo, alright, I will base my changes on sambetts changes then | 16:54 |
| rloo | I guess my question would be which is easier to maintain going forward, which would be easier to upgrade from, if we end up changing (again). | 16:54 |
| rloo | from a driver point of view, either should be easy to understand/code against. | 16:54 |
| lucasagomes | yeah, I think it's not super complicated if the devs understand how python decorators works | 16:55 |
| lucasagomes | and they should anyway | 16:55 |
| rloo | lucasagomes: in the spec, you could mention the alternative to this. | 16:56 |
| lucasagomes | rloo, ohhhhh that's a good idea | 16:56 |
| lucasagomes | so I don't have to rewrite it all? | 16:56 |
| lucasagomes | tho I think I should, cause it changes quite a bit of things | 16:57 |
| lucasagomes | like no vendor_routes or driver_routes anymore | 16:57 |
| rloo | lucasagomes: Oh, I meant the other way around. write it as using the decorator, and mention the alternative is to use the vendor_routes etc. | 16:58 |
| lucasagomes | gotcha | 16:58 |
| lucasagomes | right ;( | 16:58 |
| sambetts | sorry lucasagomes :/ | 17:01 |
| lucasagomes | sambetts, it's all good, I'm happy we are getting in a consensus | 17:01 |
| *** marcoemorais has joined #openstack-ironic | 17:01 | |
| *** marcoemorais has quit IRC | 17:01 | |
| lucasagomes | sambetts, it's because all I want was to have GET in the vendor passthru, but ended up having to refactor it all | 17:01 |
| lucasagomes | but it's good :) | 17:01 |
| *** marcoemorais has joined #openstack-ironic | 17:01 | |
| sambetts | Yeah, getting GET support is something that'll be v.useful | 17:02 |
| lucasagomes | yeah, other http methods will be supported as well, in case ppl need | 17:04 |
| lucasagomes | and a big ass doc talking about not using sync methods that talks to the BMC | 17:04 |
| *** marcoemorais has quit IRC | 17:05 | |
| *** marcoemorais has joined #openstack-ironic | 17:05 | |
| *** marcoemorais has quit IRC | 17:07 | |
| *** marcoemorais has joined #openstack-ironic | 17:07 | |
| *** marcoemorais has quit IRC | 17:07 | |
| *** marcoemorais has joined #openstack-ironic | 17:08 | |
| *** marcoemorais has quit IRC | 17:08 | |
| *** marcoemorais has joined #openstack-ironic | 17:09 | |
| *** jmank has joined #openstack-ironic | 17:10 | |
| NobodyCam | walkies time ... brb | 17:11 |
| *** Marga_ has joined #openstack-ironic | 17:11 | |
| *** marcoemorais has quit IRC | 17:15 | |
| *** Marga_ has quit IRC | 17:19 | |
| *** marcoemorais has joined #openstack-ironic | 17:20 | |
| * NobodyCam is back | 17:26 | |
| *** dprince has quit IRC | 17:27 | |
| *** Marga_ has joined #openstack-ironic | 17:28 | |
| *** achanda has joined #openstack-ironic | 17:32 | |
| *** spandhe has joined #openstack-ironic | 17:33 | |
| *** achanda has quit IRC | 17:42 | |
| *** ndipanov is now known as ndipanov_gone | 17:45 | |
| *** Marga_ has quit IRC | 17:46 | |
| *** Marga_ has joined #openstack-ironic | 17:46 | |
| *** achanda has joined #openstack-ironic | 17:47 | |
| *** foexle_ has quit IRC | 17:47 | |
| *** pensu has quit IRC | 17:48 | |
| *** achanda has quit IRC | 17:49 | |
| *** achanda has joined #openstack-ironic | 17:49 | |
| lucasagomes | sambetts, you think it makes sense to add ur patch as part of the blueprint I'm working on? So I can add ur name as a contributor of that blueprint | 17:56 |
| *** marcoemorais has quit IRC | 17:59 | |
| *** marcoemorais has joined #openstack-ironic | 17:59 | |
| sambetts | lucasagomes: yeah, I'm happy for it to become part of the blueprint :-) | 18:00 |
| lucasagomes | sambetts, aight :) | 18:00 |
| *** marcoemorais has quit IRC | 18:01 | |
| *** marcoemorais has joined #openstack-ironic | 18:01 | |
| *** marcoemorais has quit IRC | 18:01 | |
| *** marcoemorais has joined #openstack-ironic | 18:02 | |
| *** MattMan has left #openstack-ironic | 18:02 | |
| rloo | sambetts: in that case, I'll put a -2 on it because it can't be approved before the BP is approved. | 18:03 |
| jroll | ha, you've been tricked | 18:03 |
| sambetts | I have been! :-P | 18:03 |
| rloo | I was wondering if I should have mentioned that before or not. It isn't too late? | 18:03 |
| sambetts | what are other peoples opinions one if it should be part of the BP or not? | 18:03 |
| sambetts | s/one/on | 18:04 |
| jroll | I mean, I think it should | 18:04 |
| jroll | that BP has 2x+2 anyway | 18:04 |
| jroll | so it shouldn't take much to approve | 18:04 |
| rloo | the question is whether your patch can stand alone on it's own, to fix that bug. Or whether those guys tricked you into doing it so that it does more and fits with the BP :-) | 18:04 |
| rloo | jroll: as long as people agree with the new driver API change. | 18:05 |
| jroll | I mean | 18:05 |
| jroll | does this change the driver API any more than it was? | 18:05 |
| jroll | the only difference is now driver_routes etc are built automatically | 18:06 |
| *** Haomeng has joined #openstack-ironic | 18:06 | |
| jroll | instead of using a new method | 18:06 |
| rloo | jroll: it changes how people write drivers, before they added to that *_routes table. Now they use the decorator. | 18:06 |
| lucasagomes | rloo, jroll oh, right yeah... hmmm | 18:07 |
| jroll | right, but it was already changing that | 18:07 |
| jroll | just in a different way | 18:07 |
| jroll | who is going to be mad that instead of defining a dict, they use a decorator? :/ | 18:07 |
| rloo | exactly. So as long as people are fine with the decorator method, etc. | 18:07 |
| *** Haomeng|2 has quit IRC | 18:07 | |
| jroll | s/mad/something else, idk | 18:07 |
| jroll | ok | 18:07 |
| rloo | ha ha. Most likely it is fine, it is fine with me. But I can't speak for others. | 18:07 |
| jroll | regardless, it won't take long to approve, I think | 18:07 |
| jroll | ok | 18:07 |
| jroll | let's both +2 it now | 18:07 |
| jroll | :) | 18:08 |
| rloo | i can't +2. I only have +1 powers. | 18:08 |
| lucasagomes | I'm rewritting the spec :( I will put a new version up tomorrow it's late here | 18:08 |
| jroll | oh :( | 18:08 |
| *** Marga_ has quit IRC | 18:08 | |
| rloo | i did that on purpose, so people couldn't bribe me :-) | 18:08 |
| jroll | cool | 18:08 |
| jroll | thanks lucas | 18:08 |
| lucasagomes | but seems people have agreeded with the direction of it | 18:08 |
| lucasagomes | including devananda which is good | 18:08 |
| *** ociuhandu has joined #openstack-ironic | 18:09 | |
| Shrews | wow. you all moved on that decorator fast while i was out | 18:09 |
| lucasagomes | :) | 18:10 |
| *** dprince has joined #openstack-ironic | 18:10 | |
| Shrews | making it part of the BP makes sense to me | 18:11 |
| rloo | sambetts: you aren't in a hurry, are you? | 18:12 |
| sambetts | Not really, I was just trying to fix a bug and get more involved so this is all good by me :-) | 18:13 |
| lucasagomes | I'm asking to know how I should write the spec, where I say I'm going to extend the register_passthru decorator | 18:13 |
| lucasagomes | or add the register_passthru decorator as part of the spec | 18:13 |
| lucasagomes | cause I can't write a spec that depends on a change waiting for review yet | 18:14 |
| ociuhandu | hi all, I’m trying to setup a devstack + ironic all-in-one and when I try to boot any node I just get an error. Can anyone plz give me a hint on what to look for? Here’s the screen-ir-cond.log output: http://paste.openstack.org/show/126072/ | 18:15 |
| lucasagomes | right I will call it a day and go eat something, night everyone :) | 18:16 |
| *** lucasagomes is now known as lucas-dinner | 18:16 | |
| openstackgerrit | A change was merged to openstack/ironic: Remove duplicated _fetch_images function https://review.openstack.org/131405 | 18:17 |
| sambetts | night lucas-dinner | 18:18 |
| sambetts | rloo, Shrews, jroll: Are you all going to be in Paris?? | 18:18 |
| jroll | sambetts: I am :) | 18:18 |
| rloo | sambetts: I'll be in Paris, but not Shrews :-( | 18:19 |
| NobodyCam | night lucas-dinner | 18:19 |
| sambetts | I'm looking forward to putting faces to names :-) | 18:20 |
| rloo | night lucas-dinner. wrt to the spec, don't say you're going to extend the register_passthru decorator. | 18:20 |
| rloo | lucas-dinner: say you're adding the decorator (I guess there will be two, one for vendor, one for driver?) as part of the spec | 18:21 |
| rloo | great sambetts, will be good to meet you! | 18:21 |
| NobodyCam | will be good to see all the old faces and meet all the new ones | 18:22 |
| *** foexle has joined #openstack-ironic | 18:28 | |
| *** romcheg has joined #openstack-ironic | 18:38 | |
| *** marcoemorais has quit IRC | 18:40 | |
| *** foexle has quit IRC | 18:42 | |
| *** Haomeng has quit IRC | 18:43 | |
| *** Haomeng has joined #openstack-ironic | 18:46 | |
| *** romcheg has quit IRC | 18:47 | |
| *** romcheg1 has joined #openstack-ironic | 18:47 | |
| *** romcheg has joined #openstack-ironic | 18:49 | |
| *** Nisha has quit IRC | 18:50 | |
| *** romcheg1 has quit IRC | 18:51 | |
| ociuhandu | can anyone give me a clue on ironic + devstack setup? It’s failing on booting any node and I can not find any error in the logs | 18:54 |
| ociuhandu | I just get a no valid host was found message | 18:56 |
| *** andreykurilin_ has joined #openstack-ironic | 18:56 | |
| *** foexle has joined #openstack-ironic | 18:57 | |
| NobodyCam | ociuhandu: have you seen http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html#deploying-ironic-with-devstack | 18:57 |
| ociuhandu | NobodyCam: yes, followed that | 18:57 |
| sambetts | Night ironic o/ | 18:58 |
| NobodyCam | night sambetts | 18:58 |
| ociuhandu | NobodyCam: here is the ironic api log message I see: http://paste.openstack.org/show/126072/ | 18:58 |
| *** sambetts has quit IRC | 18:58 | |
| ociuhandu | NobodyCam: it just loops through the already created VMs | 18:59 |
| ociuhandu | and can’t seem to find any error | 18:59 |
| NobodyCam | ociuhandu: how about ironic-conductor and nova-compute | 18:59 |
| jroll | "no valid host" comes from nova-scheudler | 18:59 |
| Shrews | ociuhandu: latest devstack? | 18:59 |
| *** marcoemorais has joined #openstack-ironic | 18:59 | |
| NobodyCam | can you paste the nova-scheudler log | 19:00 |
| ociuhandu | Shrews: git clone from few hours ago on master | 19:00 |
| * Shrews suggests setting RECLONE=yes in local.conf and trying again | 19:00 | |
| ociuhandu | NobodyCam: the paste is from ir-cond.log, my mistack above | 19:00 |
| ociuhandu | NobodyCam: nova scheduler log: http://paste.openstack.org/show/126085/ | 19:02 |
| *** pelix has quit IRC | 19:02 | |
| ociuhandu | NobodyCam: the arch extra_spec is not working | 19:02 |
| jroll | extra_spec requirement '{i686|x86_64}' does not match 'x86_64' | 19:02 |
| ociuhandu | saaw that only now | 19:02 |
| jroll | where are people getting {i686|x86_64} from | 19:02 |
| jroll | ociuhandu: set that extra_spec on your flavor to just x86)64 | 19:03 |
| jroll | err, x86_64 | 19:03 |
| ociuhandu | jroll: yep, got that now, did a blind copy-paste without thinking from http://docs.openstack.org/developer/ironic/deploy/install-guide.html | 19:04 |
| jroll | aha | 19:04 |
| Shrews | those two guides are not compatible | 19:06 |
| NobodyCam | yep jroll http://docs.openstack.org/developer/ironic/deploy/install-guide.html#flavor-creation | 19:07 |
| *** dprince has quit IRC | 19:07 | |
| Shrews | running devstack is different from setting up an installation of your own | 19:07 |
| NobodyCam | Shrews: we shuold *NOTE that on one or both of them | 19:07 |
| jroll | I mean | 19:08 |
| jroll | they install guide has relevant parts | 19:08 |
| ociuhandu | Shrews: on the devstack guide, can I also use it for real hw node pxe? My goal is to get this devstack to support hw nodes boot with pxe | 19:08 |
| ociuhandu | Shrews: looked also on that one as the devstack guide is so short I could not really figure out if all bits are there and if real bare-metal nodes are supported | 19:09 |
| *** marcoemorais has quit IRC | 19:10 | |
| NobodyCam | does swift require a proxy even for single node setup... anyone happen to know off the tops of their heads? | 19:10 |
| Shrews | ociuhandu: you likely can do that, though we haven't documented the steps to do that. i've never tried that. | 19:11 |
| jroll | I don't see why not | 19:11 |
| Shrews | yeah. just need to change configs, register nodes, restart some services | 19:12 |
| Shrews | (in theory) | 19:13 |
| ociuhandu | Shrews: any more detailed guide I can look into for hints on what needs to be changed? | 19:13 |
| Shrews | ociuhandu: not that i'm aware of | 19:13 |
| ociuhandu | Shrews: or should I just start with the regular deployment from the start? :) | 19:13 |
| ociuhandu | and just forget about the devstack one? | 19:14 |
| rloo | ociuhandu: my guess is that you can run through the instructions for devstack + ironic, and then look at the install guide to see what else needs to be done. (I haven't actually looked at the devstack instructions recently) | 19:15 |
| ociuhandu | rloo: thanks | 19:16 |
| *** mjturek has quit IRC | 19:16 | |
| rloo | ociuhandu: and please let us know if the docs can be improved (or submit a patch!) | 19:16 |
| NobodyCam | yay for /shamelessplug | 19:17 |
| ociuhandu | rloo: will document my steps and do that | 19:17 |
| *** andreykurilin_ has quit IRC | 19:17 | |
| NobodyCam | ociuhandu: awesome :) | 19:17 |
| rloo | NobodyCam: what are you talking about? Just trying to make everyone feel 'at home' here ;) | 19:17 |
| *** andreykurilin_ has joined #openstack-ironic | 19:18 | |
| NobodyCam | hehehe the (or patch) | 19:18 |
| ociuhandu | I do have the feeling I’m going to have to bug you for a little while until I manage to have the whole setup running :) | 19:18 |
| ociuhandu | my target is to have everything working in a couple of days :) | 19:18 |
| NobodyCam | :) | 19:19 |
| ociuhandu | one last question for now: if my hardware node does not have ipmi/ilo/idrac can I use the fake driver and just power it on manually for now? We’re also working on a special driver for our setup and will make that available once it’s running | 19:20 |
| ociuhandu | but in order to test the driver I need to have the other bits running :) | 19:20 |
| jroll | yep, that should work fine :) | 19:21 |
| jroll | I think HP has a "log" driver somewhere that does basicaly that | 19:21 |
| ociuhandu | whew, I was hoping that but was not sure :) | 19:21 |
| * NobodyCam thinks log ..... power ..... driver... lol | 19:21 | |
| * NobodyCam also ducks | 19:21 | |
| ociuhandu | looking forward to meeting you guys in Paris at the summit :) | 19:22 |
| NobodyCam | :) | 19:22 |
| rloo | didn't someone submit a patch for some sort of log driver? | 19:22 |
| ociuhandu | and show you also the new driver, in the hope that all will be working by then :P | 19:22 |
| NobodyCam | ociuhandu: cannt wait to see | 19:23 |
| NobodyCam | rloo: yea | 19:23 |
| Shrews | this reminds me, our installation guide doesn't mention needing tftp installed at all. it just assumes it is. oversight? | 19:23 |
| * NobodyCam looks | 19:24 | |
| NobodyCam | Shrews: http://docs.openstack.org/developer/ironic/deploy/install-guide.html#pxe-setup | 19:25 |
| Shrews | NobodyCam: right. that's the section that assumes it is already installed | 19:25 |
| Shrews | unless it comes as part of the syslinux pkg | 19:26 |
| Shrews | maybe that's it | 19:26 |
| ociuhandu | NobodyCam: it never says it should be installed or reference to it’s config file | 19:26 |
| ociuhandu | Shrews: on ubuntu it does not from all I have seen | 19:26 |
| ociuhandu | I had to manually install it | 19:27 |
| Shrews | ociuhandu: great feedback. thank you | 19:27 |
| *** rbowen has left #openstack-ironic | 19:27 | |
| ociuhandu | I hope to be able to provide a doc with all the steps I had to follow to get to the end result and we could review them once I get closer to the working setup :) | 19:28 |
| Shrews | ociuhandu: that would be AWESOME | 19:29 |
| *** mjturek has joined #openstack-ironic | 19:29 | |
| NobodyCam | Shrews: so we need to add tftpd-hpa to step #2 of the pxe-setup section | 19:29 |
| *** mjturek has left #openstack-ironic | 19:30 | |
| ociuhandu | also on the devstack one, as devstack is not installing it also | 19:32 |
| ociuhandu | though it’s creating the config file | 19:32 |
| Shrews | ociuhandu: devstack should install tftp for you | 19:32 |
| ociuhandu | Shrews: I have checked and I had no tftpd or tfptd-hpa on ubuntu 14.04 | 19:33 |
| ociuhandu | let me check the logs | 19:33 |
| Shrews | ociuhandu: https://github.com/openstack-dev/devstack/blob/master/files/apts/ironic | 19:34 |
| Shrews | if it didn't install everything it needed, our gate would fail | 19:36 |
| ociuhandu | Shrews: digging into that now | 19:36 |
| BLZbubba | looks like the Juno cloud-archive has a version of python-sqlalchemy that is too new for python-ironic | 19:37 |
| ociuhandu | Shrews: looks like my setup has some issue: Server /usr/sbin/in.tftpd is not executable [file=/etc/xinetd.d/tftp] [line=8] | 19:39 |
| *** marcoemorais has joined #openstack-ironic | 19:43 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update install guide to install tftp https://review.openstack.org/131547 | 19:49 |
| Shrews | NobodyCam: ^^^ | 19:49 |
| *** Marga_ has joined #openstack-ironic | 19:50 | |
| Shrews | I'm not 100% about the RH package name, so if someone with one of those systems could verify that for me, much thanks | 19:50 |
| rloo | Shrews: I just looked online, and that seems to be the right name for fedora | 19:52 |
| rloo | Shrews: just wondering if anything needs to be set up after installing. | 19:52 |
| rloo | Shrews: http://docs.fedoraproject.org/en-US/Fedora/17/html/Installation_Guide/s1-netboot-pxe-config.html | 19:53 |
| Shrews | rloo: great, thx. not sure about post install stuff... config maybe? | 19:53 |
| Shrews | hoping someone with more operation experience will chime in about that | 19:54 |
| * Shrews notes it is very challenging to develop on ironic without actual h/w to run/test against/blow up | 19:55 | |
| rloo | Shrews: I know what you mean! | 19:56 |
| jroll | it's really hard :( | 19:56 |
| Shrews | heh. and here i thought i was the only one | 19:56 |
| rloo | Shrews: can we just add some handwaving in the guide, they need to find out how to install & configure? | 19:59 |
| rloo | Shrews: are we allowed to put #TODO in the docs? :-) | 20:00 |
| Shrews | uh... | 20:00 |
| Shrews | *shrug* x 2 | 20:00 |
| Shrews | if there is tftpd magic they need to perform, i think it would be nice to show that at some point | 20:01 |
| * Shrews check devstack magic | 20:01 | |
| rloo | I agree. Thought I'd help you out with this patch though. Pass the buck, you know ;) | 20:02 |
| Shrews | rloo: https://github.com/openstack-dev/devstack/blob/master/lib/ironic#L529 | 20:03 |
| rloo | Shrews: good stuff. How do you describe that? :-) | 20:03 |
| Shrews | gobbledy-gook | 20:03 |
| * Shrews studies | 20:03 | |
| rloo | and that's for ubuntu only. | 20:04 |
| Shrews | yeah :( | 20:04 |
| Shrews | ugh. gotta clone that repo to grep it | 20:05 |
| Shrews | https://github.com/openstack-dev/devstack/blob/master/tools/ironic/templates/tftpd-xinetd.template | 20:06 |
| Shrews | so, the way devstack does it might not be the way we want to document it. seems overly complicated | 20:07 |
| Shrews | *sigh* | 20:08 |
| Shrews | rloo: i'm beginning to like your #TODO idea ;) | 20:08 |
| *** todd_dsm has quit IRC | 20:08 | |
| rloo | Shrews: it seems like if it is going to take days to get the rest of the info in, then we should get this in now, because something is better than nothing. | 20:09 |
| jroll | just... "set up tftpd to serve $tftproot" | 20:09 |
| *** Marga_ has quit IRC | 20:09 | |
| rloo | yeah, something like what jroll says. | 20:09 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update install guide to install tftp https://review.openstack.org/131547 | 20:12 |
| Shrews | like so? ^^^ | 20:12 |
| Shrews | maybe adding the quotes... | 20:12 |
| mrda | Morning Ironic | 20:13 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update install guide to install tftp https://review.openstack.org/131547 | 20:13 |
| Shrews | mrda: morning | 20:13 |
| mrda | hey Shrews | 20:14 |
| yjiang5 | ociuhandu: Shrews: I think at least currently devstack does not support ironic+hw, I have a patch https://review.openstack.org/#/c/127658/ on devstack. It at least works on my mini-cloud. | 20:20 |
| NobodyCam | I've used https://github.com/openstack/tripleo-image-elements/blob/master/elements/ironic-conductor/install.d/69-ironic-tftp-support as a base | 20:21 |
| jroll | Shrews: can I get a +A? :) https://review.openstack.org/#/c/129694/ | 20:22 |
| Shrews | yjiang5: argh... i hadn't considered the networking part. the vm's currently bridge in to the private network to make ironic work. | 20:22 |
| Shrews | jroll: NO! | 20:22 |
| jroll | :( | 20:22 |
| * jroll hangs his head and walks away | 20:22 | |
| *** romcheg has quit IRC | 20:22 | |
| *** romcheg1 has joined #openstack-ironic | 20:22 | |
| rloo | Shrews: added a comment | 20:23 |
| rloo | morning mrda | 20:23 |
| yjiang5 | Shrews: yes, we have to change the network side to use provider network, and also to avoid the VM creation. | 20:23 |
| Shrews | NobodyCam: that looks like exactly what devstack does | 20:24 |
| rloo | oh mrda, was going to mention. I think we're good to delete most of the ironic/nova/* files. Did you want to do that? | 20:24 |
| jroll | rloo: I thought we were waiting for end of cycle for that? | 20:25 |
| Shrews | ociuhandu: you may want to try out yjiang5's change | 20:25 |
| jroll | or were we waiting for end of juno? | 20:25 |
| yjiang5 | ociuhandu: and review it :) | 20:25 |
| rloo | jroll: start of kilo. ie, it will NOT be available when kilo is released. | 20:26 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update install guide to install tftp https://review.openstack.org/131547 | 20:26 |
| Shrews | rloo: thx | 20:26 |
| jroll | hmm | 20:26 |
| rloo | jroll: or that's my interpretation anyway. nova already deleted the bm code. | 20:26 |
| Shrews | rloo: wha??? | 20:27 |
| jroll | rloo: right, so, I think tripleo is still using the host manager from our tree :/ | 20:27 |
| Shrews | i thought the bm code wasn't being deleted until next cycle | 20:27 |
| *** Marga_ has joined #openstack-ironic | 20:27 | |
| * rloo checks | 20:27 | |
| NobodyCam | oh its gone | 20:28 |
| Shrews | heh. ok then | 20:28 |
| jroll | Shrews: supposedly people don't pull code between releases | 20:28 |
| mrda | rloo: oh that's cool | 20:28 |
| rloo | Shrews, you weren't going to use that code, were you? ;) | 20:28 |
| jroll | I'd triple check with SpamapS before removing it from our tree | 20:29 |
| rloo | hmm, do we have release notes for icehouse, mentioning what will be deprecated in K? | 20:29 |
| mrda | yeah, nova-bm going is a surprise | 20:29 |
| mrda | that's not consistent with my conversations | 20:29 |
| JayF | nova-bm last supported release was juno | 20:30 |
| Shrews | rloo: there go my plans for a competing product based off of nova-bm | 20:30 |
| Shrews | :( | 20:30 |
| JayF | so it got removed from K | 20:30 |
| JayF | that's how I understood it and how I think it happened? | 20:30 |
| rloo | Shrews: you can still get the code! just need to know where to look... | 20:30 |
| rloo | well, mrda, thought I'd leave you with the pleasure of deleting those files, whenever we think they can be deleted :-) | 20:31 |
| jroll | right, it was removed from kilo | 20:31 |
| * Shrews was going to call his competitor ColdSteel | 20:31 | |
| mrda | lol, ok, let's discuss this f2f in Paris | 20:31 |
| mrda | (that sounds crazy) | 20:31 |
| Shrews | no i haz cool name, no codez | 20:31 |
| *** Marga_ has quit IRC | 20:31 | |
| Shrews | now, even | 20:32 |
| mrda | oh, (because i missed yesterday's meeting), there's an Ironic dinner on Monday night? No meetup on Sunday? | 20:32 |
| *** todd_dsm has joined #openstack-ironic | 20:32 | |
| jroll | mrda: officially, yeah, dinner monday night | 20:32 |
| jroll | some of us still might meet up sunday night | 20:32 |
| mrda | ok, cool. thanks jroll | 20:32 |
| NobodyCam | un-officially there may also be a breakfast that jroll is having on tues | 20:33 |
| NobodyCam | lol | 20:33 |
| jroll | lol | 20:33 |
| rloo | jroll is cooking for everyone! | 20:33 |
| jroll | I'm having breakfast | 20:33 |
| jroll | it will be cooked for me | 20:33 |
| mrda | jroll: Feel free to grab me on Sunday. Happy to spend time with y'all | 20:33 |
| jroll | there may be extra chairs | 20:33 |
| jroll | y'all can grab them | 20:33 |
| NobodyCam | :) lol | 20:34 |
| NobodyCam | jroll: now do we also get to eat ... or just watch you eat? | 20:34 |
| * NobodyCam stops | 20:35 | |
| mrda | By Tuesday you won't be sick of baguettes yet | 20:35 |
| jroll | lol NobodyCam, that's up to you | 20:35 |
| NobodyCam | :-p | 20:36 |
| openstackgerrit | A change was merged to openstack/ironic: Clear maintenance_reason when setting maintenance=False https://review.openstack.org/129694 | 20:53 |
| *** marcoemorais has quit IRC | 21:01 | |
| *** marcoemorais1 has joined #openstack-ironic | 21:01 | |
| NobodyCam | brb | 21:08 |
| *** yjiang5 has quit IRC | 21:11 | |
| *** linggao has quit IRC | 21:16 | |
| *** yjiang5 has joined #openstack-ironic | 21:19 | |
| *** harlowja is now known as harlowja_away | 21:21 | |
| *** rushiagr is now known as rushiagr_away | 21:27 | |
| *** Marga_ has joined #openstack-ironic | 21:35 | |
| *** spandhe has quit IRC | 21:39 | |
| *** andreykurilin_ has quit IRC | 21:41 | |
| *** romcheg1 has quit IRC | 21:41 | |
| ociuhandu | yjiang5: how do I configure hardware_info file? | 21:41 |
| ociuhandu | yjiang5: IRONIC_HWINFO_FILE=${IRONIC_HWINFO_FILE:-$IRONIC_DATA_DIR/hardware_info | 21:42 |
| yjiang5 | The example in my side is : | 21:42 |
| yjiang5 | 192.168.110.107 00:1e:67:57:50:4c root otc123 | 21:42 |
| *** marcoemorais1 has quit IRC | 21:42 | |
| yjiang5 | ipmi addr, ipmi mac, ipmi user, ipmi password. | 21:42 |
| ociuhandu | yjiang5: can this work with the fake driver? from the code I saw it seems like not | 21:42 |
| SpamapS | jroll, rloo: you don't have to triple check with me btw. Just run check experimental. | 21:43 |
| *** Marga__ has joined #openstack-ironic | 21:43 | |
| jroll | SpamapS: true, but I feel that we'll forget | 21:43 |
| SpamapS | jroll, rloo: but I think jroll is right, we are probably still using ironic namespaced things. | 21:43 |
| SpamapS | if you have a bug for that, please add a task for tripleo. | 21:43 |
| jroll | idk that we do | 21:44 |
| yjiang5 | ociuhandu: I just use the pxe_ipmitool. What's the fake driver? Sorry I use pxe_ipmitool on my environment. | 21:44 |
| *** marcoemorais has joined #openstack-ironic | 21:45 | |
| *** Marga_ has quit IRC | 21:45 | |
| *** romcheg has joined #openstack-ironic | 21:45 | |
| ociuhandu | yjiang5: in this case it means that if the machine does not have ipmi enabled you can’t use it | 21:45 |
| rloo | thx SpamapS! | 21:46 |
| *** harlowja_away is now known as harlowja | 21:46 | |
| rloo | mrda, ^^ wrt the ironic/nova code. | 21:46 |
| *** spandhe has joined #openstack-ironic | 21:46 | |
| ociuhandu | yjiang5: afaik the fake driver is just a “placeholder” so I can manually power on the machine if it does not have ipmi support and it would still boot ok (this is my understanding) | 21:46 |
| yjiang5 | ociuhandu: will that work? I think the ironic has more interaction with the compute node more than power-on, right? | 21:47 |
| *** marcoemorais has quit IRC | 21:48 | |
| *** marcoemorais has joined #openstack-ironic | 21:48 | |
| *** Marga__ has quit IRC | 21:48 | |
| ociuhandu | yjiang5: my understanding is that once it is powered on it will use pxe to install so it should work | 21:49 |
| jroll | yjiang5: it should work, ironic waits for callbacks from ramdisks or whatever after powering on a node | 21:49 |
| ociuhandu | yjiang5: have to say I’m new with ironic so I hope I’m not wrong | 21:49 |
| ociuhandu | yjiang5: the rest of the nteraction should not be part of ipmi (except getting metrics maybe for ceilometer and system status) | 21:50 |
| mrda | rloo: ack | 21:50 |
| yjiang5 | ociuhandu: I have 1 month experience with Ironic now, or maybe 3 weeks, not sure if I can say I'm new with Ironic :) jroll, but that means the ironic-set-power-state will not work, and console either, right? | 21:50 |
| jroll | yjiang5: set power state will work, it just won't actually do anything :) | 21:50 |
| yjiang5 | jroll: :) | 21:50 |
| jroll | ociuhandu will have to watch for logs and push the button | 21:51 |
| yjiang5 | ociuhandu: so how will you create the node when using the fake driver? Any special information should be passed? | 21:53 |
| *** andreykurilin_ has joined #openstack-ironic | 21:53 | |
| openstackgerrit | yunhong-jiang proposed a change to openstack/ironic: Change the force_raw_image config usage https://review.openstack.org/130880 | 21:55 |
| openstackgerrit | yunhong-jiang proposed a change to openstack/ironic: Convert qcow2 image to raw format when deploy https://review.openstack.org/130881 | 21:55 |
| ociuhandu | yjiang5: i should be able to get the IP and start the PXE based on the mac address of the node | 21:55 |
| ociuhandu | yjiang5: I’m actually working on setting a test environment for this now | 21:56 |
| yjiang5 | ociuhandu: So I think possibly we can extend the IRONIC_HARDWARE to be more specific talking about the real connection to the hardware. Currently, as in my code's statement , "he 'pxe_ipmitool' is only valid when IRONIC_HARDWARE set". So basically IRONIC_HARDWARE can be ipmitool, or fake driver, or even ssh, am I right? | 21:57 |
| ociuhandu | yes | 21:59 |
| NobodyCam | humm WARNING keystone.common.wsgi [-] Authorization failed. The request you have made requires authentication. from 127.0.0.1 thats a new one for me :-p | 22:00 |
| ociuhandu | though that would mean handling in a separate way the node registration you do in that while loop, based on the type of driver selected | 22:00 |
| ociuhandu | yjiang5: as currently you only handle the ipmi options list | 22:01 |
| *** romcheg has quit IRC | 22:02 | |
| ociuhandu | yjiang5: it would be nice to also have the cpu_arch as parameter in that file, I think, instead of hardcoding it to x86_64 | 22:02 |
| jroll | there's all sorts of power drivers | 22:03 |
| yjiang5 | ociuhandu: or, basically there are two things, one is the type of the computpe node , is it real hardware, or a VM. Another one is the connection to the compute node, ipmi, ssh, or fake driver etc. My current patch combine them, thus need change to use two config. These two configs are related in one point: If a VM, then the connection information comes from the VM creation process. | 22:03 |
| yjiang5 | jroll: is my understanding correct? | 22:03 |
| jroll | I think so | 22:03 |
| jroll | link to your change? | 22:03 |
| yjiang5 | jroll: https://review.openstack.org/#/c/127658/3 | 22:04 |
| ociuhandu | yjiang5: it’s related to the bare-metal node, not the compute node, no? | 22:04 |
| yjiang5 | ociuhandu: what do you mean of the baremetal node? Isn't it same as compute node? | 22:04 |
| ociuhandu | yjiang5: in my general openstack terminology not really, as the compute in this case is using ironic as a driver | 22:05 |
| ociuhandu | so it’s one compute node that handles all the VM/hardware nodes that will be deployed | 22:06 |
| lucas-dinner | rloo, yup 2 decorators sounds good | 22:06 |
| lucas-dinner | (sorry for the delay was having dinner :) ) | 22:06 |
| jroll | compute node handles everything | 22:07 |
| ociuhandu | those can be whatever you want them to be based on the image to be deployed. They can have no openstack compenonents running | 22:07 |
| yjiang5 | jroll: and then for the neutron changes, when should we change the ip/route? I'm not quite sure, should it happen to the compute node type, or the compute ndoe connection? | 22:07 |
| jroll | and compute node has nothing to do with the baremetal node's drivers | 22:07 |
| yjiang5 | ociuhandu: Not sure of the difference. I worked on nova RT(Resource Tracker) before. At least from RT point of view , compute node is more like the baremetal node? The separation of compute node and the compute service has been discussed a lot on nova side, I think. | 22:08 |
| jroll | yjiang5: I think this is a good start. I think maybe you should be setting IRONIC_POWER_DRIVER=(ssh|ipmitool|ilo|etc) instead of just IRONIC_HARDWARE=True... and combine that with like a IRONIC_DEPLOY_DRIVER to make e.g. pxe_ipmitool | 22:09 |
| yjiang5 | jroll: a silly question, since I think I don't have the whole picture of ironic yet. When 'ironic create-node', we should input driver_info, is it for power driver, or deploy driver, or for both? | 22:11 |
| jroll | both | 22:11 |
| jroll | the "driver" is a combination of deploy and power drivers | 22:11 |
| ociuhandu | yjiang5: from my understanding, any bare-metal node being deployed is similar to regular VM instance in openstack. The main difference is that not being a virtual instance will not “live” on top of a compute node, but rather you will have one compute node managing all bare-metal instances | 22:13 |
| jroll | right | 22:13 |
| jroll | let me see if this helps | 22:13 |
| jroll | think of ironic as the hypervisor | 22:13 |
| yjiang5 | jroll: got it. So back to the devstack support, I think the first step is to decide if we use VM or real hardware as the node (compute node, or baremetal node). I supect if IRONIC_POWER_DRIVER/IRONIC_DEPLOY_DRIVER can make the distinguish. | 22:13 |
| jroll | there are one or more nova-compute "services" | 22:13 |
| jroll | there are one or more ironic nodes akak "compute_nodes" managed by the nova-compute services | 22:14 |
| *** todd_dsm has quit IRC | 22:14 | |
| jroll | yjiang5: actually....... | 22:14 |
| jroll | yjiang5: we already have an IRONIC_DEPLOY_DRIVER | 22:15 |
| jroll | maybe if that's set to e.g. pxe_ipmitool, you can do the hardware stuff | 22:15 |
| yjiang5 | jroll: hmm, a quick check of "ironic/drivers/base.py", seems in ironic, it's ONE driver, with TWO interface. So IRONIC_DEPLOY_DRIVER is already the ironic DRIVER, and it does include both power and deploy, like pxe_ssh etc. | 22:18 |
| jroll | yjiang5: right, that's why I just changed my mind :) | 22:19 |
| yjiang5 | jroll: :) current the issue is, ociuhandu want to use hardware stuff with the fake driver. | 22:19 |
| *** ryanpetrello has quit IRC | 22:19 | |
| jroll | right, so I think pxe_fake is a "real hardware" driver | 22:19 |
| jroll | if you're using the virtual bare metal nodes, there's no reason to use the fake driver, we have the ssh driver for that | 22:20 |
| jroll | but I see your point | 22:20 |
| yjiang5 | jroll: so how do you think we distinguish VM-based and hardware-based? I think possibly we can't distinguish from IRONIC_DEPLOY_DRIVER? | 22:23 |
| jroll | I think we can | 22:24 |
| NobodyCam | omg seed deployed with swift :-p I should be a ble to test something, humm what was I going to test? | 22:26 |
| jroll | woohoo! | 22:26 |
| yjiang5 | jroll: waiting for your suggestion :) | 22:26 |
| jroll | couldn't be the agent | 22:26 |
| jroll | yjiang5: *_ssh is virtual, anything else is not | 22:26 |
| jroll | IMO | 22:26 |
| yjiang5 | jroll: can the fake works with virtual also, manually shutdown the VM, just like push power button in the real hardware. Hmm possibly no one want to do that :) So I will take that way. | 22:30 |
| jroll | it could, but we have the ssh driver that does that | 22:30 |
| jroll | for kvm, virtualbox, parallels, vmware | 22:30 |
| jroll | no reason to use fake there | 22:31 |
| jroll | if you do, fix devstack yourself :P | 22:31 |
| yjiang5 | jroll: I will re-work the patch. Hopefully it will get more review after this discussion. | 22:31 |
| ociuhandu | jroll: yes but that complicates too much I think yjiang5’s life and patch as his patch is also handling automatic node registration from a file and based on the driver selected, different parameters have to be set, no? | 22:31 |
| jroll | just different driver(_info) params, yes | 22:32 |
| jroll | I mean | 22:32 |
| jroll | yjiang5: set things up so that if IRONIC_HARDWARE is set, it does interesting things to set up for hw | 22:32 |
| jroll | and then a second patch can automatically set that if "pxe_ipmitool" is the deploy driver | 22:32 |
| jroll | is how I would do it | 22:32 |
| yjiang5 | ociuhandu: engineer life is always complicate :) | 22:32 |
| ociuhandu | :) | 22:33 |
| yjiang5 | jroll: why do we still need IRONIC_HARDWARE? We can simply decided based on ironic_deploy_driver as you just said, right? | 22:34 |
| jroll | just to make it simpler down below | 22:34 |
| ociuhandu | yjiang5: let’s say I started setting up an ironic env today and by the end of the week i have to have a lot of things working on it :) | 22:34 |
| jroll | instead of `IRONIC_DEPLOY_DRIVER == pxe_ipmitool or pxe_ipminative or agent_ipmitool.........` | 22:34 |
| jroll | but I guess you could use a function call | 22:35 |
| yjiang5 | jroll: yes, exactly. I will work on it and hopefully push for review tomorrow. | 22:35 |
| NobodyCam | ipa folk will I be able to see anything on the console when agent is loaded? | 22:36 |
| yjiang5 | ociuhandu: can you please share me the config needed for your FAKE driver? I didn't try fake driver before. Or I can checking the code myself. | 22:37 |
| spandhe | jroll: devananda… hello! I have a qn related to compute_nodes, ironic nodes discussion that we had yesterday.. | 22:40 |
| ociuhandu | yjiang5: I am working on getting to that point, once I have one that I am sure works fine I’ll get back to you for sure | 22:40 |
| spandhe | what should be the service_id for an ironic node (hypervisor) in nova.compute_nodes ? | 22:40 |
| ociuhandu | yjiang5: from all I’ve seen all I need to set is the mac address for the network interface but I have to check that | 22:41 |
| jroll | spandhe: it doesn't need to be anything specific | 22:41 |
| spandhe | jroll: its set to 4 (nova-compute) in my case.. | 22:43 |
| spandhe | jroll: so https://github.com/openstack/nova/blob/2014.1.1/nova/scheduler/host_manager.py#L450 call returns http://paste.openstack.org/show/126180/ | 22:44 |
| *** jgrimm is now known as zz_jgrimm | 22:44 | |
| yjiang5 | ociuhandu: so, can I ignore your pxe_fake driver now, and add it later when you figure out, as a separate patch? | 22:44 |
| *** ryanpetrello has joined #openstack-ironic | 22:44 | |
| spandhe | the response has “service” component in it.. is it supposed to be there ? | 22:44 |
| jroll | spandhe: yeah, seems fine | 22:45 |
| ociuhandu | yjiang5: yes | 22:45 |
| spandhe | jroll: ok.. what happens after that is.. when applying the filters, initial host list https://github.com/openstack/nova/blob/2014.1.1/nova/filters.py#L70 is something like http://paste.openstack.org/show/126182/ | 22:48 |
| spandhe | jroll: and in the end, the scheduler ends up picking nova-compute hostname for scheduling | 22:48 |
| jroll | spandhe: seems fine? | 22:50 |
| jroll | other than the negative disk | 22:50 |
| jroll | and 0 ram | 22:51 |
| jroll | are any instances assigned to that nova-compute service? | 22:51 |
| jroll | are you doing something weird with host aggregates or something? | 22:51 |
| jroll | are there VMs happening there? | 22:51 |
| spandhe | jroll: nope.. no VMs or host aggregates.. | 22:51 |
| jroll | huh | 22:52 |
| spandhe | jroll: does the initial host list look fine? currently, based on that initial list, the scheduler is picking nova-compute.. | 22:52 |
| *** ryanpetrello has quit IRC | 22:53 | |
| spandhe | jroll: the database shows correct values for free ram, etc.. | 22:53 |
| jroll | spandhe: it looks like there's an instance on that ironic node | 22:53 |
| jroll | or on some ironic node on that nova-compute service | 22:53 |
| jroll | e.g. u'memory_mb_used': 24576L, | 22:55 |
| * jroll brb | 22:55 | |
| spandhe | jroll: I havent added the nova-compute node to ironic.. | 22:55 |
| jroll | um | 22:56 |
| spandhe | yeah I just spotted that.. let me check how that happened.. | 22:56 |
| jroll | you don't add a nova-compute node to ironic | 22:56 |
| jroll | you add a server to ironic | 22:56 |
| jroll | nova-compute polls ironic for information about your baremetal servers | 22:56 |
| spandhe | jroll: right.. I added a server to Ironic.. | 22:56 |
| jroll | uh huh | 22:56 |
| jroll | and then nova-compute hit the ironic api and saw that server | 22:56 |
| jroll | think of ironic as the hypervisor | 22:56 |
| jroll | your hypervisor now has room | 22:57 |
| jroll | for an instance | 22:57 |
| jroll | (or it should, not sure why it shows those resources as used | 22:57 |
| jroll | ) | 22:57 |
| jroll | I have to walk my dog, I'll be back shortly | 22:57 |
| spandhe | jroll: right… so it should pick the hypervisor.. | 22:57 |
| spandhe | jroll: sure.. will talk later.. thanks! | 22:57 |
| *** yuanying has joined #openstack-ironic | 23:06 | |
| jroll | spandhe: so... yeah. nova-scheduler gets all of the nova-compute services | 23:06 |
| jroll | and looks for one with enough space for your instance | 23:07 |
| jroll | which in this case is none of them | 23:07 |
| spandhe | jroll: yeah.. looking at nova.db.sqlalechemy.api.compute_node_get_all // | 23:07 |
| jroll | nova internals are not the problem here :) | 23:07 |
| spandhe | jroll: when I fire the query manually on the database, I get the correct values.. but the api returns weird values for “used” | 23:07 |
| jroll | huh, what do you get in the db? | 23:08 |
| spandhe | jroll: in db, all “used” columns are 0.. including running_vms | 23:09 |
| jroll | I may be misunderstanding something here | 23:10 |
| jroll | can you run "nova hypervisor-stats" | 23:10 |
| spandhe | jroll: http://paste.openstack.org/show/126190/ | 23:11 |
| jroll | ........ weird. | 23:11 |
| jroll | something seems out of sync here | 23:13 |
| jroll | or I'm misunderstanding things | 23:13 |
| jroll | on the side since I was thinking about it... JayF, NobodyCam, do either of you see a reason why we shouldn't be able to call delete on an instance that is currently deploying? | 23:14 |
| jroll | e.g. why can't this be false for states.DEPLOYING https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L321-323 | 23:15 |
| jroll | spandhe: maybe restart your nova-compute service? | 23:15 |
| spandhe | jroll: let me try that.. | 23:15 |
| jroll | spandhe: or maybe not, maybe it's fine... what's the end output of the scheduler filtering? nothing? | 23:15 |
| spandhe | jroll: I tried nova boot one more time.. this time I got all the correct values for the query response.. but it picked the nova-compute host | 23:17 |
| jroll | sounds like a good thing | 23:17 |
| jroll | is there something wrong with that? | 23:18 |
| spandhe | jroll: it picked nova-compute instead of hypervisor.. | 23:18 |
| spandhe | jroll: digging deeper.. something went wrong in self.host_manager.get_weighed_hosts call | 23:19 |
| NobodyCam | jroll: so close: Error: Swift temporary URLs require a shared secret to be created. You must provide "swift_temp_url_key" as a config option. | 23:19 |
| jroll | it's unclear to me what nova-compute and hypervisor mean in this context... | 23:19 |
| jroll | NobodyCam: provide the config option! | 23:19 |
| NobodyCam | lol | 23:19 |
| spandhe | jroll: by nova-compute, I mean the machine where service nova-compute runs | 23:20 |
| jroll | NobodyCam: you have to configure swift as well... swift post -m something /me looks | 23:20 |
| spandhe | sorry for the confusion.. should have been more clear.. | 23:20 |
| jroll | spandhe: yes, the scheduler picks a nova-compute service, which picks a "hypervisor" | 23:20 |
| spandhe | jroll: oh, | 23:20 |
| jroll | NobodyCam: swift post -m "Temp-URL-Key: $SWIFT_TEMPURL_KEY | 23:21 |
| jroll | " | 23:21 |
| jroll | and set the same key in ironic conf | 23:21 |
| jroll | spandhe: so what do you see now? good things or bad things? | 23:23 |
| spandhe | jroll: that clarified things a little bit.. but I dont see the hypervisor being picked.. The boot errors out in the end.. and the error msg mentions nova-compute node as the hypervisor name.. let me see if I can find anything in nova-compute logs | 23:25 |
| jroll | yeah, check those logs, and an exact error would be cool | 23:25 |
| jroll | if you get the request id (req-$some_uuid) from any of the services, you can use that to trace it through the system | 23:25 |
| jroll | though ironic doesn't log it, need to fix that :( | 23:25 |
| spandhe | jroll: I dont see the request id in nova-compute logs.. | 23:31 |
| jroll | huh. | 23:32 |
| spandhe | jroll: the scheduler should eventually return the hypervisor, right? | 23:32 |
| jroll | maybe that's a rackspace thing, sorry (I really don't think it is but who knows) | 23:32 |
| jroll | spandhe: I... think so? I'm fuzzy on nova scheduler things | 23:32 |
| jroll | I'm doing my best not to read the entire scheduler code right now :( | 23:33 |
| spandhe | jroll: let me do that myself :) thanks! | 23:34 |
| jroll | lol | 23:34 |
| jroll | I'm happy to help if you find specific things in there :P | 23:34 |
| *** rushiagr_away is now known as rushiagr | 23:36 | |
| *** rloo has quit IRC | 23:38 | |
| spandhe | jroll: sure :) thanks! digging.. will let you know if I find anything interesting.. | 23:40 |
| *** andreykurilin_ has quit IRC | 23:40 | |
| *** igordcard has quit IRC | 23:47 | |
| *** yuanying has quit IRC | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!