Pradip | ues I think so too. i am trying with that | 00:00 |
---|---|---|
Pradip | NobodyCam: got it. Thanks | 00:01 |
NobodyCam | :) | 00:01 |
*** boris-42 has joined #openstack-ironic | 00:02 | |
*** naohirot has joined #openstack-ironic | 00:02 | |
*** achanda has quit IRC | 00:16 | |
*** davideagnello has quit IRC | 00:16 | |
*** ijw__ has quit IRC | 00:29 | |
openstackgerrit | Sinval Vieira Mendes Neto proposed openstack/ironic: Refactor refresh method in objects for reuse https://review.openstack.org/203914 | 00:34 |
*** ijw has joined #openstack-ironic | 00:35 | |
*** korekhov has joined #openstack-ironic | 00:36 | |
*** Pradip has quit IRC | 00:49 | |
*** ijw has quit IRC | 00:55 | |
*** korekhov has quit IRC | 00:56 | |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Make endpoint type configurable https://review.openstack.org/204326 | 01:04 |
*** puranamr has joined #openstack-ironic | 01:16 | |
*** zhenguo has joined #openstack-ironic | 01:22 | |
zer0c00l | I am running ironic/ipa using devstack. All the ironic nodes created by devstack have is_whole_disk_image=True | 01:23 |
zer0c00l | is there a way to use is_whole_disk_image False | 01:24 |
zer0c00l | is_whole_disk_image is part of driver_internal_info and ironic node-update cannot update it | 01:24 |
zer0c00l | i am trying to test a partition image | 01:25 |
zer0c00l | so this is_whole_disk_image should be set to False | 01:25 |
*** killer_prince has joined #openstack-ironic | 01:25 | |
TheJulia | zer0c00l: trying to write out a partition image how? | 01:31 |
TheJulia | killer_prince: good morning | 01:31 |
killer_prince | TheJulia: Good UGT morning.. | 01:33 |
zer0c00l | TheJulia: So i created a partition image using my custom diskimage-builder | 01:33 |
*** stendulker has joined #openstack-ironic | 01:33 | |
zer0c00l | i mean custom disk-image-create element | 01:33 |
zer0c00l | It doesn't have grub. | 01:34 |
TheJulia | meaning you have to boot via tftp? | 01:34 |
zer0c00l | Yes | 01:34 |
zer0c00l | The coreOS ramdisk kernel and ramdisk are transfered over tftp | 01:34 |
TheJulia | zer0c00l: what driver are you using? | 01:35 |
zer0c00l | agent_ssh | 01:37 |
TheJulia | so i think you need to use pxe_ssh | 01:37 |
zer0c00l | sorry, i might be mistaken | 01:38 |
zer0c00l | ironic node-show says the driver is agent_ssh | 01:38 |
TheJulia | because the agent is really geared for whole disk images. ipa can apparently still be used for the deploy in that case, from what I have been told | 01:38 |
zer0c00l | By whole disk images you mean a disk image with grub and everthing installed? | 01:39 |
TheJulia | whole disk as in a partition table is present | 01:40 |
TheJulia | and generally bootable | 01:40 |
zer0c00l | So partition table and bootloader | 01:40 |
zer0c00l | okay! | 01:40 |
zer0c00l | So there is no way to use partition image (not whole disk) with ironic? | 01:41 |
zer0c00l | or ipa | 01:41 |
zer0c00l | From the looks of it the iscsi_deploy has options to deploy a partition image. | 01:41 |
zer0c00l | it checks this driver_internal_info=is_whole_disk_image | 01:41 |
zer0c00l | if it is false | 01:41 |
zer0c00l | it ipa creates the partition | 01:42 |
TheJulia | zer0c00l: every deployment is naturally different, but we pushing users to utilize the agent driver and whole disk images that are bootable so the machine can boot without the condoctor and at some point live on a different network | 01:42 |
zer0c00l | and dds the parttion image on it | 01:42 |
zer0c00l | i see | 01:43 |
zer0c00l | TheJulia: is there any documentation on how whole disk images are built? | 01:43 |
zer0c00l | TheJulia: or how can i build them? | 01:43 |
TheJulia | zer0c00l: honestly, not sure. I'm fairly sure partition images will work if the right path is taken, but you then have to have the conductor host available to boot the node... forevermore | 01:44 |
TheJulia | zer0c00l: add the vm element to the disk-image-create element list | 01:44 |
zer0c00l | TheJulia: i will add the VM element | 01:44 |
TheJulia | zer0c00l: lower case vm, it is badely named :( | 01:45 |
zer0c00l | TheJulia: I don't think partition images need conductor host all the time. IPA installs the bootloader (grub2) if it is a partition image | 01:45 |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Make endpoint type configurable https://review.openstack.org/204326 | 01:45 |
* TheJulia thinks its time we write a new element | 01:45 | |
zer0c00l | So the machine will be able to boot independent of conductor | 01:45 |
zer0c00l | i will checkout the vm element | 01:46 |
zer0c00l | Thanks for the pointer! | 01:46 |
TheJulia | zer0c00l: so i dont know that code, so i cant speak to that, although that seems highly reliant upon things being grub2 compatible | 01:46 |
zer0c00l | I agree | 01:47 |
zer0c00l | Not sure how it will work on older systems like RHEL6 or RHEL5 | 01:47 |
zer0c00l | where there is no grub 2 | 01:47 |
zer0c00l | i don't see why it shouldn't work | 01:47 |
*** pal has joined #openstack-ironic | 01:47 | |
*** puranamr has quit IRC | 01:47 | |
zer0c00l | but not sure whether there have been tests done with those partition images | 01:47 |
TheJulia | i would doubt it | 01:48 |
zer0c00l | i work at a place where i have to support RHEL6,RHEL7 | 01:48 |
zer0c00l | RHEL6 has grub 0.9 | 01:48 |
zer0c00l | RHEL 7 grub 2 | 01:48 |
zer0c00l | and RHEL 7 needs special parittion (biosboot) to hold that grub | 01:48 |
zer0c00l | not sure ipa handles that | 01:49 |
zer0c00l | so yeah | 01:49 |
zer0c00l | whole disk images solve these kind of problems :) | 01:49 |
TheJulia | yeah, the logic to handle the variations would be fraught with terror and risk :) | 01:49 |
TheJulia | yup | 01:49 |
zer0c00l | or make ironic plugabble to other installers where they can take over and do the installation :) | 01:50 |
openstackgerrit | Tan Lin proposed openstack/ironic: Migrate ObjectListBase to subclass from the Oslo one https://review.openstack.org/201017 | 01:51 |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Load authenticate token from HTTP header https://review.openstack.org/174231 | 01:51 |
TheJulia | i doubt that would happen, there is somethin nice to knowing you've downloaded an image, verified the checksum, and then write thay data to disk without making any additional changes | 01:51 |
TheJulia | s/somethin/something/ | 01:51 |
*** bradjones has quit IRC | 01:54 | |
*** bradjones has joined #openstack-ironic | 01:56 | |
*** bradjones has quit IRC | 01:56 | |
*** bradjones has joined #openstack-ironic | 01:56 | |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Load authenticate token from HTTP header https://review.openstack.org/174231 | 01:59 |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Load authenticate token from HTTP header https://review.openstack.org/174231 | 02:04 |
*** yuanying_ has joined #openstack-ironic | 02:14 | |
*** yuanying has quit IRC | 02:17 | |
*** puranamr has joined #openstack-ironic | 02:18 | |
*** puranamr has quit IRC | 02:18 | |
*** yuanying_ has quit IRC | 02:25 | |
*** yuanying has joined #openstack-ironic | 02:26 | |
*** stendulker has quit IRC | 02:31 | |
*** achanda has joined #openstack-ironic | 02:33 | |
*** thrash is now known as thrash|g0ne | 02:36 | |
jroll | NobodyCam: LOL best typo, I do that all the time | 02:41 |
jroll | s/do/type/ :| | 02:41 |
*** pal has quit IRC | 02:43 | |
openstackgerrit | Cameron.C proposed openstack/ironic: Allow inspector driver to work in standalone mode https://review.openstack.org/202435 | 02:43 |
*** ramineni has joined #openstack-ironic | 02:49 | |
*** hakimo has joined #openstack-ironic | 02:52 | |
*** hakimo_ has quit IRC | 02:55 | |
*** chenglch has joined #openstack-ironic | 02:59 | |
*** Nisha has joined #openstack-ironic | 03:03 | |
*** naohirot has quit IRC | 03:03 | |
*** stendulker has joined #openstack-ironic | 03:17 | |
*** Haomeng has joined #openstack-ironic | 03:17 | |
*** Haomeng|2 has quit IRC | 03:19 | |
*** yuanying has quit IRC | 03:24 | |
*** killer_prince has quit IRC | 03:24 | |
*** yuanying has joined #openstack-ironic | 03:25 | |
*** pal has joined #openstack-ironic | 03:29 | |
*** h00327910_ has quit IRC | 03:35 | |
*** saripurigopi has joined #openstack-ironic | 03:38 | |
*** ubuntus_ has quit IRC | 03:48 | |
*** achanda has quit IRC | 03:50 | |
*** naohirot has joined #openstack-ironic | 03:58 | |
*** coolsvap|away is now known as coolsvap | 04:01 | |
*** kkoski has joined #openstack-ironic | 04:17 | |
*** kkoski has quit IRC | 04:21 | |
*** puranamr has joined #openstack-ironic | 04:24 | |
*** boris-42 has quit IRC | 04:32 | |
*** puranamr has quit IRC | 04:37 | |
*** achanda has joined #openstack-ironic | 04:37 | |
*** puranamr has joined #openstack-ironic | 04:37 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic: Updated from global requirements https://review.openstack.org/204394 | 04:55 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-inspector: Updated from global requirements https://review.openstack.org/204395 | 04:55 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/204396 | 04:55 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/python-ironic-inspector-client: Updated from global requirements https://review.openstack.org/204419 | 04:59 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/204420 | 04:59 |
*** max_lobur has joined #openstack-ironic | 05:00 | |
*** max_lobur has quit IRC | 05:04 | |
*** zz_natorious is now known as natorious | 05:06 | |
*** puranamr has quit IRC | 05:10 | |
*** natorious is now known as zz_natorious | 05:25 | |
*** lazy_prince has joined #openstack-ironic | 05:26 | |
*** ramineni has quit IRC | 05:54 | |
*** ramineni has joined #openstack-ironic | 05:56 | |
SpamapS | NobodyCam: Hm, that looks like a massive refactoring. I will read up on it. Probably fine. | 06:01 |
*** amotoki has joined #openstack-ironic | 06:09 | |
*** max_lobur has joined #openstack-ironic | 06:11 | |
*** puranamr has joined #openstack-ironic | 06:11 | |
*** Marga__ has quit IRC | 06:14 | |
*** Marga_ has joined #openstack-ironic | 06:15 | |
*** puranamr has quit IRC | 06:15 | |
*** vishwanathj has quit IRC | 06:17 | |
*** dtantsur|afk is now known as dtantsur | 06:25 | |
dtantsur | Morning Ironic | 06:25 |
stendulker | Good Morning dtantsur :) | 06:28 |
dtantsur | o/ | 06:28 |
*** lsmola has joined #openstack-ironic | 06:29 | |
*** boris-42 has joined #openstack-ironic | 06:29 | |
*** yuikotakada has joined #openstack-ironic | 06:29 | |
*** ukalifon has joined #openstack-ironic | 06:30 | |
*** Nisha_away has joined #openstack-ironic | 06:36 | |
*** Nisha has quit IRC | 06:36 | |
Nisha_away | dtantsur, GM | 06:42 |
dtantsur | o/ | 06:42 |
Nisha_away | dtantsur, i am getting following error continuously in ironic-inspector service http://paste.openstack.org/show/398792/ . I am not even trying deploy/inspection at this time | 06:44 |
dtantsur | Nisha_away, interesting, is there any logging above/below this message? | 06:46 |
*** yog__ has joined #openstack-ironic | 06:46 | |
*** ramineni has quit IRC | 06:47 | |
*** ramineni has joined #openstack-ironic | 06:48 | |
Nisha_away | dtantsur, i restarted it again to see if there is any failure before the error messages starts but the logs dont show anything like that. The errors starts coming after some time itself. http://paste.openstack.org/show/398835/ | 06:52 |
Nisha_away | this error is looping continuously | 06:52 |
yuikotakada | Hi, Ironic :) | 06:54 |
Nisha_away | yuikotakada, hi | 06:54 |
*** smoriya has joined #openstack-ironic | 06:55 | |
yuikotakada | Nisha_away, hi :) | 06:57 |
Nisha_away | :) | 06:57 |
dtantsur | yuikotakada, o/ | 06:58 |
dtantsur | Nisha_away, please set debug=True in the inspector.conf and restart | 06:59 |
yuikotakada | dtantsur, o/ | 07:00 |
*** coolsvap is now known as coolsvap|away | 07:01 | |
Nisha_away | ok. will do ...had to reboot the system due to some reasons...will see if error still come, will set debug to true | 07:01 |
Nisha_away | i am stacking again | 07:01 |
dtantsur | ack | 07:04 |
dtantsur | probably the problem will go away after reboot... | 07:04 |
* dtantsur brb | 07:06 | |
*** athomas has joined #openstack-ironic | 07:12 | |
*** ifarkas has joined #openstack-ironic | 07:12 | |
openstackgerrit | Anusha Ramineni proposed openstack/ironic: Make driver properties discoverable in REST API https://review.openstack.org/203921 | 07:16 |
*** e0ne has joined #openstack-ironic | 07:22 | |
-openstackstatus- NOTICE: zuul is undergoing an emergency restart. Jobs will be re-queued but some events may be lost. | 07:23 | |
*** achanda has quit IRC | 07:30 | |
*** pal has quit IRC | 07:33 | |
*** achanda has joined #openstack-ironic | 07:39 | |
*** ijw has joined #openstack-ironic | 07:44 | |
*** dguerri` is now known as dguerri | 07:48 | |
*** achanda has quit IRC | 07:49 | |
*** ijw has quit IRC | 07:58 | |
*** jistr has joined #openstack-ironic | 08:03 | |
*** karimb has joined #openstack-ironic | 08:04 | |
*** e0ne has quit IRC | 08:05 | |
*** yuanying has quit IRC | 08:05 | |
*** pal has joined #openstack-ironic | 08:09 | |
*** karimb has quit IRC | 08:13 | |
*** karimb has joined #openstack-ironic | 08:14 | |
-openstackstatus- NOTICE: zuul jobs are currently stuck while problems with gearman are debugged | 08:14 | |
*** ChanServ changes topic to "zuul jobs are currently stuck while problems with gearman are debugged" | 08:14 | |
*** subscope has quit IRC | 08:14 | |
*** derekh has joined #openstack-ironic | 08:17 | |
*** lucasagomes has joined #openstack-ironic | 08:19 | |
*** romainh has joined #openstack-ironic | 08:31 | |
*** boris-42 has quit IRC | 08:32 | |
*** romcheg has joined #openstack-ironic | 08:33 | |
*** dtantsur is now known as dtantsur|brb | 08:36 | |
*** ramineni has quit IRC | 08:42 | |
*** ramineni has joined #openstack-ironic | 08:43 | |
saripurigopi | I'm trying to add driver vendor API, but its not getting listed. failing with "InvalidParameterValue: No handler for method enrol_nodes" | 09:00 |
*** athomas has quit IRC | 09:00 | |
*** subscope has joined #openstack-ironic | 09:01 | |
saripurigopi | Followed the same steps for node vendor apis, node vendor API is working fine. | 09:01 |
saripurigopi | have updated mapping in driver/pxe.py file. anything else I need to modify? | 09:02 |
*** e0ne has joined #openstack-ironic | 09:07 | |
saripurigopi | TheJulia: Hi | 09:09 |
*** karimb has quit IRC | 09:10 | |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic-specs: Lazy delete instances https://review.openstack.org/204162 | 09:10 |
sambetts | saripurigopi: Have you passed you used the driver_passthru decorator on the function | 09:12 |
lucasagomes | saripurigopi, the method is decorated with @driver_passthru? | 09:12 |
sambetts | ? | 09:12 |
sambetts | hahaha hey lucasagomes | 09:13 |
lucasagomes | sambetts, sup hah same time! | 09:13 |
lucasagomes | sambetts, good morning | 09:13 |
sambetts | lucasagomes: Hows it going? | 09:14 |
lucasagomes | sambetts, all is well! How are you? | 09:14 |
saripurigopi | sambetts: o/ | 09:14 |
saripurigopi | lucasagomes, sambetts: yes, http://paste.openstack.org/show/399258/ | 09:14 |
lucasagomes | saripurigopi, and when you make the request are you using /v1/drivers/<driver name>/vendor_passthru/enrol_nodes ? | 09:15 |
lucasagomes | saripurigopi, the method looks correctly decorated | 09:16 |
lucasagomes | saripurigopi, does the uuid method works? | 09:16 |
saripurigopi | lucasagomes: I'm trying from ironic CLI like : ironic driver-vendor-passthru --http-method GET pxe_ucs enrol_nodes address=10.105.214.143 username=admin password=Nbv12345 | 09:16 |
saripurigopi | lucasagomes: yes it works. | 09:16 |
*** karimb has joined #openstack-ironic | 09:17 | |
sambetts | saripurigopi: is there any errors other than the method not found appearing in the ironic logs? | 09:19 |
lucasagomes | saripurigopi, hmm odd, try also to GET v1/drivers/<driver name>/vendor_passthru/methods | 09:19 |
lucasagomes | that will list all the driver vendor passthru methods exposed by that driver | 09:19 |
saripurigopi | sambetts: no, I don't see any other errors. http://paste.openstack.org/show/399259/ | 09:20 |
lucasagomes | or ironic node-driver-vendor-passthru --http-method GET pxe_ucs methods | 09:21 |
saripurigopi | lucasagomes: TheJulia suggested me to try that yesterday, it returns empty for driver vendor apis in this case. | 09:21 |
lucasagomes | somehow it's not getting registered then | 09:21 |
sambetts | saripurigopi: Have you restarted your conductor? | 09:21 |
saripurigopi | yes | 09:21 |
saripurigopi | sambetts: ironic node-vendor-passthru --http-method GET d070d861-cb17-4d94-a0fc-b3bf2b76cc15 uuid {u'uuid': u'd070d861-cb17-4d94-a0fc-b3bf2b76cc15'} | 09:22 |
lucasagomes | it's odd, we have been using the driver vendor passthru for things like lookup in the agent | 09:23 |
lucasagomes | and didn't hit a problem... | 09:23 |
lucasagomes | try to take a look at the code in ironic when it does the mapping for the handlers | 09:23 |
* lucasagomes lemme find the link | 09:23 | |
saripurigopi | lucasagomes: okay. I'm not sure what i'm missing. | 09:24 |
lucasagomes | saripurigopi, seems fine, you just need to decorate then in the vendor interface | 09:24 |
lucasagomes | as you did apparently | 09:24 |
*** karimb has quit IRC | 09:24 | |
saripurigopi | lucasagomes: http://docs.openstack.org/developer/ironic/dev/vendor-passthru.html ? | 09:24 |
sambetts | saripurigopi: Its said on the client that it defaults to a POST request you specified it as a GET request, so I think you might need to add --http-method GET | 09:25 |
sambetts | oh you have... oops XD | 09:25 |
saripurigopi | sambetts: :-) | 09:25 |
lucasagomes | saripurigopi, https://github.com/openstack/ironic/blob/3a46559c8e03997237e51454cda0a9e63f3c42b2/ironic/drivers/base.py#L650-L674 | 09:25 |
*** puranamr has joined #openstack-ironic | 09:25 | |
lucasagomes | that's how the mapping get's created | 09:25 |
saripurigopi | lucasagomes: okay | 09:26 |
lucasagomes | saripurigopi, are you using MixinVendorInterface ? | 09:26 |
lucasagomes | that could be a problem there too, https://github.com/openstack/ironic/blob/a0760043e5b10b8e7fad2db53e23786c481c2b84/ironic/drivers/utils.py#L46-L70 | 09:27 |
lucasagomes | I would investigate those bits for looking for a problem there | 09:27 |
lucasagomes | saripurigopi, note that for driver level mapping, the MixinVendorInterface uses "driver_passthru_mapping" not "mapping" | 09:28 |
saripurigopi | sambas, lucasagomes: sorry, It was my mistake. I had the mappings defined in ironic/driver/pxe.py as well, like we specified in Juno. I've removed that, seems it listening the method now. | 09:28 |
*** coolsvap|away is now known as coolsvap | 09:28 | |
saripurigopi | *sambetts: | 09:29 |
lucasagomes | saripurigopi, yeah it's not "mappings" for driver stuff | 09:29 |
*** karimb has joined #openstack-ironic | 09:29 | |
*** e0ne has quit IRC | 09:29 | |
saripurigopi | lucasagomes: okay. I just removed that, seems working fine now. | 09:30 |
lucasagomes | self.vendor = utils.MixinVendorInterface(self.mapping, driver_passhru_mapping=self.driver_mapping) | 09:30 |
lucasagomes | saripurigopi, cool! | 09:30 |
*** puranamr has quit IRC | 09:30 | |
*** romcheg has quit IRC | 09:31 | |
saripurigopi | lucasagomes: do I still need to specify "self.vendor = utils.MixinVendorInterface(self.mapping, driver_passhru_mapping=self.driver_mapping) " in pxe.py? | 09:34 |
lucasagomes | saripurigopi, probably yes because pxe drivers depends on some passthru methods | 09:34 |
lucasagomes | pass_deploy_info | 09:34 |
lucasagomes | pass_bootloader_info | 09:34 |
saripurigopi | lucasagomes: okay, | 09:35 |
saripurigopi | sambetts, lucasagomes: thank you. | 09:35 |
lucasagomes | no problem | 09:35 |
sambetts | :) | 09:36 |
*** yury_yekovenko has joined #openstack-ironic | 09:39 | |
*** romcheg has joined #openstack-ironic | 09:39 | |
*** yury_yekovenko has quit IRC | 09:40 | |
*** naohirot has quit IRC | 09:47 | |
*** karimb has quit IRC | 09:50 | |
*** MattMan has quit IRC | 10:05 | |
*** dtantsur|brb is now known as dtantsur | 10:07 | |
dtantsur | gate is down completely | 10:09 |
dtantsur | good time to get some downstream work done :D | 10:09 |
lucasagomes | :-( | 10:14 |
lucasagomes | dtantsur, the jobs were called? Or is it broken? | 10:14 |
lucasagomes | oh I see ZUUL now... yeha | 10:14 |
lucasagomes | damn | 10:14 |
dtantsur | yep, nothing passes gate now | 10:15 |
*** coolsvap is now known as coolsvap|away | 10:16 | |
*** Nisha_away has quit IRC | 10:16 | |
openstackgerrit | Debayan Ray proposed stackforge/proliantutils: common changes for firmware update https://review.openstack.org/203543 | 10:16 |
*** e0ne has joined #openstack-ironic | 10:17 | |
lucasagomes | if people wants to review something: https://review.openstack.org/#/c/204162/ :-) | 10:18 |
*** yuikotakada has quit IRC | 10:19 | |
*** pelix has joined #openstack-ironic | 10:26 | |
*** Haomeng|2 has joined #openstack-ironic | 10:38 | |
*** chenglch has quit IRC | 10:40 | |
*** cing has joined #openstack-ironic | 10:41 | |
*** Haomeng has quit IRC | 10:41 | |
*** davideagnello has joined #openstack-ironic | 10:43 | |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic-specs: Lazy delete instances https://review.openstack.org/204162 | 10:49 |
*** athomas has joined #openstack-ironic | 10:56 | |
*** romcheg has quit IRC | 11:01 | |
*** stendulker has quit IRC | 11:02 | |
*** davideagnello has quit IRC | 11:04 | |
sambetts | haha lucasagomes using the word blueprint in your commit message has confused the hell out of the gerrit commit message linking thing :-P it now thinks your blueprint is called "adds" | 11:07 |
lucasagomes | sambetts, LOL damn... yeah I should say spec | 11:07 |
*** ramineni has quit IRC | 11:09 | |
*** romcheg has joined #openstack-ironic | 11:16 | |
*** bradjones has quit IRC | 11:18 | |
*** bradjones has joined #openstack-ironic | 11:20 | |
*** bradjones has quit IRC | 11:20 | |
*** bradjones has joined #openstack-ironic | 11:20 | |
*** pal has quit IRC | 11:24 | |
*** ramineni has joined #openstack-ironic | 11:25 | |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Only take exclusive lock in sync_power_state if node is updated https://review.openstack.org/202562 | 11:26 |
dtantsur | this was a complex beast ^^ | 11:26 |
* lucasagomes adds to his todo | 11:28 | |
lucasagomes | oh cool, that's the one that was WIP when I reviewed | 11:29 |
openstackgerrit | Zhenguo Niu proposed openstack/ironic: Use utils.mkfs directly in deploy_utils https://review.openstack.org/204557 | 11:30 |
* sambetts goes to make coffee before a meeting | 11:36 | |
*** ramineni has quit IRC | 11:38 | |
dtantsur | lucasagomes, yep, I believe it's no longer WIP now :) | 11:43 |
TheJulia | Good morning everyone | 11:46 |
sambetts | Morning TheJulia o/ | 11:48 |
dtantsur | TheJulia, morning! I hope you don't need some gate, because we don't have any for you | 11:49 |
dtantsur | :D | 11:49 |
lazy_prince | Good Morning TheJulia | 11:49 |
lucasagomes | dtantsur, :-) w00t! Yeah that's a good patch, specially when get_power_state() can take a lot of time! Not holding a lock there is great | 11:49 |
lucasagomes | TheJulia, good morning | 11:49 |
* lucasagomes goes grab some food | 11:49 | |
*** lucasagomes is now known as lucas-hungry | 11:49 | |
dtantsur | lucasagomes :) | 11:49 |
TheJulia | dtantsur: gate broken? | 11:49 |
lucas-hungry | TheJulia, it's down | 11:50 |
TheJulia | :( | 11:50 |
dtantsur | see the topic | 11:50 |
TheJulia | ahh | 11:50 |
TheJulia | Good to know | 11:50 |
* TheJulia is still drinking her first cup of coffee | 11:51 | |
*** thrash|g0ne is now known as thrash|brb | 11:52 | |
*** david-lyle has joined #openstack-ironic | 11:52 | |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Only prevent updates to driver_info if power transition is in progres https://review.openstack.org/204565 | 11:58 |
dtantsur | discussable ^^ | 11:58 |
*** dprince has joined #openstack-ironic | 11:59 | |
TheJulia | dtantsur: I think expanding on the problem case may help a reviewer's understanding | 12:05 |
dtantsur | TheJulia, you mean what specifically? why is it a problem? | 12:06 |
TheJulia | I kind of understand why, but the picture of why is 100% clear in my brain | 12:06 |
TheJulia | yes | 12:06 |
dtantsur | oh hmm, I'll try to | 12:06 |
TheJulia | err, is not 100% clear | 12:06 |
* TheJulia clearly needs more coffee | 12:06 | |
ifarkas | hi all | 12:09 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Only prevent updates to driver_info if power transition is in progres https://review.openstack.org/204565 | 12:09 |
ifarkas | dtantsur, a question regarding your comment: https://review.openstack.org/#/c/203991/2/zuul/layout.yaml | 12:09 |
dtantsur | ifarkas, yep | 12:10 |
ifarkas | dtantsur, so python-jobs isn't there because there's no test in the lib yet, they will be added later. or did I misunderstand something? | 12:10 |
dtantsur | ifarkas, ah, I didn't know it, sorry. I was assuming you just forgot about the 2nd place to define jobs (which I personally often do) | 12:11 |
ifarkas | dtantsur, ahh, okay. thanks. I will leave a comment | 12:11 |
dtantsur | ifarkas, then it's better not do add python-jobs to jenkins/jobs.yaml to avoid confusion | 12:11 |
ifarkas | dtantsur, ok, will remove that | 12:11 |
dtantsur | * jenkins/jobs/projects.yaml | 12:11 |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Strip ports list from inventory data https://review.openstack.org/204571 | 12:21 |
*** dguerri is now known as dguerri` | 12:25 | |
*** cing has quit IRC | 12:26 | |
*** karimb has joined #openstack-ironic | 12:39 | |
*** jjohnson2 has joined #openstack-ironic | 12:41 | |
* dtantsur has a week of emails to the ML.. | 12:41 | |
*** pal has joined #openstack-ironic | 12:48 | |
*** dguerri` is now known as dguerri | 12:51 | |
*** trown|outttypeww is now known as trown | 12:51 | |
*** lazy_prince has quit IRC | 12:55 | |
*** lucas-hungry is now known as lucasagomes | 12:56 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Add nics information to ironic inventory https://review.openstack.org/204595 | 13:04 |
*** rvasilets has joined #openstack-ironic | 13:06 | |
rvasilets | lucasagomes, Hi. Did you member this patch https://review.openstack.org/#/c/186064/21 ?) One of our cores require you +1 again to put his +2. Could you vote once more? | 13:07 |
lucasagomes | rvasilets, hi there, I will re-review it soonish | 13:08 |
* lucasagomes adds to his todo | 13:08 | |
*** [1]cdearborn has joined #openstack-ironic | 13:09 | |
*** linggao has joined #openstack-ironic | 13:10 | |
*** linggao has quit IRC | 13:10 | |
*** linggao has joined #openstack-ironic | 13:11 | |
linggao | Hi folk, long time no see. :-) | 13:12 |
dtantsur | o/ | 13:13 |
linggao | Someone asked me a queston about Ironic and I think this is a good place to ask. :-) | 13:13 |
linggao | on http://docs.openstack.org/ page why isn't ironic api mentioned? | 13:14 |
linggao | And why isn't Ironic mentioned in the installation guide? | 13:15 |
*** rloo has joined #openstack-ironic | 13:16 | |
*** Pradeep has joined #openstack-ironic | 13:17 | |
Pradeep | Ironic power state is not in sync with nova power state, can some one help me on this | 13:20 |
Pradeep | When i power off the node through ironic driver,the power of the baremetal goes down immediately but it takes sometime for the nova (nova list) to change the power state to shutdown | 13:21 |
Pradeep | When i power on the node again through ironic (ironic node-set-power-state <> on ), Power in ironic node-list and baremetal is up | 13:22 |
Pradeep | but the power of instance in nova (nova list) is still down and the instance is not coming up | 13:22 |
sambetts | Pradeep: you shouldn't be shuting down the machine via ironic like that if you've got an instance on there | 13:23 |
jroll | right, nova is seeing that it shouldn't be powered on | 13:24 |
jroll | Pradeep: 'nova reboot --hard' should kick it back on fine | 13:24 |
NobodyCam | good morning Ironicers | 13:24 |
jroll | in the future if there's an instance, use nova to start/stop | 13:24 |
lucasagomes | linggao, yeah it's a pitty... we are the only integrate project to have no official docs | 13:24 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic-inspector: Convert functional test to a framework and decouple from client https://review.openstack.org/204607 | 13:25 |
lucasagomes | linggao, people are working on that AFAIK! | 13:25 |
*** amotoki has quit IRC | 13:25 | |
Pradeep | ok sam, you mean we should not use "ironic node-set-power-state " | 13:25 |
jroll | lucasagomes: docs are there, just not on the home page | 13:26 |
jroll | right? | 13:26 |
jroll | http://docs.openstack.org/developer/ironic/ | 13:26 |
*** david-lyle has quit IRC | 13:27 | |
sambetts | Pradeep: No if you've got an instance on a baremetal compute host, and you want to reboot it you should do it via nova | 13:27 |
* jroll bbiab | 13:27 | |
*** puranamr has joined #openstack-ironic | 13:27 | |
*** cdearborn has joined #openstack-ironic | 13:28 | |
linggao | Hi lucasagomes. thanks for the prompt response.:-) | 13:28 |
lucasagomes | jroll, right but that's not the official docs right? I mean that's our docs but it's not someone from the docs team sanitizing and making release of those docs for us | 13:28 |
sambetts | Pradeep: as jroll says you can bring a node back up if you've gone out of sync like that by doing nova reboot --hard | 13:28 |
lucasagomes | as you would expect for the other integrated projects | 13:28 |
*** [1]cdearborn has quit IRC | 13:28 | |
Pradeep | ok Sam, then when shall we use "ironic node-set-power-state" | 13:29 |
sambetts | Pradeep: for maintenance, or non-nova related tasks. Once you've got your node into ironic and they are bootable etc then you shouldn't have to touch the ironic api at all except for maintenence on the nodes, all instance management should be done via nova | 13:31 |
*** mjturek1 has joined #openstack-ironic | 13:31 | |
*** puranamr has quit IRC | 13:32 | |
rloo | hello ironickers | 13:32 |
*** cdearborn has quit IRC | 13:33 | |
sambetts | Hey rloo o/ how are you? | 13:33 |
lucasagomes | Pradeep, it's expected that the nova instance delays a little to update the power state | 13:33 |
rloo | hi sambetts, good, although not ready to work yet. going through my emails... :) | 13:33 |
lucasagomes | Pradeep, nova runs a periodic tasks which calls get_info(<instance>) on the driver to actually get the information about that instance | 13:33 |
lucasagomes | rloo, good morning | 13:33 |
lucasagomes | Pradeep, https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L5671-L5673 | 13:34 |
lucasagomes | and seems that by default that periodic task runs every 10 minutes, which is a pretty long interval | 13:35 |
lucasagomes | rloo, welcome back! | 13:36 |
*** ndipanov has joined #openstack-ironic | 13:37 | |
rloo | lucasagomes: thanks. (Back for about a week then gone again for another weekish) | 13:37 |
dtantsur | rloo, morning! | 13:38 |
Pradeep | thank you Sam, lucas | 13:38 |
*** thiagop has joined #openstack-ironic | 13:39 | |
rloo | hi dtantsur! | 13:39 |
thiagop | morning Ironic | 13:40 |
NobodyCam | morning sambetts rloo lucasagomes Pradeep and dtantsur :) | 13:40 |
dtantsur | o/ | 13:40 |
sambetts | o/ NobodyCam | 13:40 |
NobodyCam | and thiagop :) | 13:40 |
*** subscope has quit IRC | 13:41 | |
rloo | morning NobodyCam | 13:41 |
NobodyCam | :) | 13:41 |
NobodyCam | how are you doing rloo ? | 13:42 |
*** cdearborn has joined #openstack-ironic | 13:42 | |
rloo | NobodyCam: good. Catchin' up still :) | 13:42 |
NobodyCam | TheJulia: can we unblock this now that glean pushed a new package to pypi? | 13:42 |
NobodyCam | https://review.openstack.org/#/c/203000 | 13:42 |
NobodyCam | :) | 13:43 |
*** subscope has joined #openstack-ironic | 13:43 | |
*** zhenguo has quit IRC | 13:43 | |
TheJulia | Hmm... Yes :) | 13:44 |
thiagop | Guys, can I get some eyes on this Ramesh's patch? | 13:44 |
thiagop | https://review.openstack.org/#/c/166513/ | 13:44 |
TheJulia | NobodyCam: firing it up locally to test | 13:44 |
NobodyCam | woot :) | 13:44 |
lucasagomes | rloo, ah nice one! | 13:45 |
*** jcoufal has quit IRC | 13:45 | |
thiagop | I think it's pretty mature after rloo reviews and it'll bring a lot of flexibility to new drivers | 13:45 |
* lucasagomes will review it soonish | 13:46 | |
TheJulia | NobodyCam: that whole using cirros for gate testing doesn't really help validate that dib changes work :( | 13:46 |
*** Pradeep has quit IRC | 13:49 | |
openstackgerrit | Chris Krelle proposed openstack/bifrost: Vagrantfile and vagrant.yml for testing https://review.openstack.org/191110 | 13:50 |
*** kkoski has joined #openstack-ironic | 13:55 | |
*** bnemec has joined #openstack-ironic | 13:58 | |
TheJulia | NobodyCam: so... looks like it might not be working on debian still, going to try minimal, then going to try to dig into whats going on | 14:01 |
*** kkoski has quit IRC | 14:02 | |
*** kkoski has joined #openstack-ironic | 14:02 | |
NobodyCam | TheJulia: ack :) | 14:03 |
jroll | rloo: welcome back \o/ | 14:06 |
rloo | jroll: :) | 14:07 |
BadCub | Mornin' y'all | 14:09 |
dtantsur | BadCub, o/ | 14:09 |
thiagop | morning BadCub | 14:09 |
*** jcoufal has joined #openstack-ironic | 14:12 | |
*** ndipanov has quit IRC | 14:13 | |
-openstackstatus- NOTICE: CI is currently recovering from an outage overnight. It is safe to recheck results with NOT_REGISTERED errors. It may take some time for zuul to work through the backlog. | 14:15 | |
*** ChanServ changes topic to "CI is currently recovering from an outage overnight. It is safe to recheck results with NOT_REGISTERED errors. It may take some time for zuul to work through the backlog." | 14:15 | |
*** boris-42 has joined #openstack-ironic | 14:16 | |
lucasagomes | jroll, if you have time see sambetts comment here https://review.openstack.org/#/c/204162/ | 14:18 |
lucasagomes | jroll, I just want to confirm with you if the network bits are fine | 14:18 |
jroll | lucasagomes: yeah, was looking at that this morning | 14:20 |
jroll | need to think about it more | 14:20 |
lucasagomes | ack | 14:20 |
jroll | I think in the flat network case (today) we'll have problems | 14:20 |
lucasagomes | jroll, why is that? | 14:20 |
jroll | because nova tears down the neutron ports | 14:21 |
jroll | and you no longer have netwroking | 14:21 |
jroll | oh wait no | 14:21 |
sambetts | jroll: thats what I was thinking, but I'm not sure if it would | 14:21 |
lucasagomes | it's ok, I mean we don't need to boot anything for deleting | 14:21 |
jroll | you're right, we don't do anything that needs networking until cleaning | 14:21 |
lucasagomes | and for cleaning it will recconfigure the network | 14:21 |
*** mtanino has joined #openstack-ironic | 14:21 | |
jroll | at which point... yeah | 14:21 |
lucasagomes | and neutron ports etc | 14:21 |
jroll | mhm | 14:21 |
jroll | so yeah, the future is the same idea, only tenant ports are torn down by nova, I think it's fine | 14:22 |
jroll | unconfigure_tenant_networks() might need to be called immediately rather than with the delete, but I think that's ok | 14:22 |
jroll | lucasagomes: so is the lock the only reason we can't delete while in DEPLOYING? | 14:23 |
lucasagomes | jroll, not really, we would need to stop the worker thread doing the job as well | 14:23 |
lucasagomes | which is the hard part | 14:23 |
jroll | mmm, right | 14:24 |
jroll | agent doesn't really have this problem :P | 14:24 |
lucasagomes | if we break the lock and put it in "deleting" the worker thread may overwirte it | 14:24 |
lucasagomes | so I prefer to be safe and wait until we are able to acquire the lock | 14:24 |
lucasagomes | then then delete the instance | 14:24 |
lucasagomes | to avoid races | 14:24 |
jroll | yeah, but | 14:24 |
jroll | if we stop supporting the bash ramdisk | 14:24 |
lucasagomes | even tho it may mean that the instance goes to active for a short time before it gets deleted | 14:24 |
jroll | we don't really need this, right? | 14:24 |
*** vishwanathj has joined #openstack-ironic | 14:24 | |
lucasagomes | jroll, the problem is the iscsi methodology | 14:25 |
lucasagomes | where the conductor does the buck of work | 14:25 |
jroll | oh right | 14:25 |
jroll | dang | 14:25 |
jroll | down with iscsi! | 14:25 |
jroll | :P | 14:25 |
lucasagomes | sure for the agent_ methodology this will be pretty good | 14:25 |
lucasagomes | cause it mostly wait on DEPLOYWAIT state | 14:25 |
lucasagomes | jroll, hehe yeah | 14:25 |
jroll | yeah, I forgot IPA supports iscsi drivers | 14:25 |
lucasagomes | so it does affect IPA with iscsi still | 14:25 |
jroll | ok fiiiine | 14:25 |
lucasagomes | but it's not a big problem | 14:25 |
jroll | :P | 14:25 |
lucasagomes | hah | 14:25 |
*** kkoski has quit IRC | 14:26 | |
lucasagomes | jroll, but anyway re bash ramdisk! Damn we should def deprecate it | 14:26 |
lucasagomes | start putting some warnings on the pass_deploy_info methods and stuff | 14:26 |
jroll | yeah, indeed | 14:28 |
jroll | maybe we should put up a small spec on it | 14:28 |
jroll | as much as I hate specs | 14:29 |
jroll | just to get on the same page for the plan | 14:29 |
lucasagomes | jroll, right... yeah lemme put one up | 14:29 |
jroll | heh, I wasn't going to make you do it, but thanks in advance! | 14:29 |
jroll | :P | 14:29 |
lucasagomes | jroll, btw, that patch reorganizing the specs needs little work on | 14:29 |
lucasagomes | we should approved that asap | 14:29 |
lucasagomes | jroll, it's ok, the bash ramdisk annoys me big time | 14:30 |
sambetts | so would that mean deprecating the pxe_ipmi driver over agent_ipmi? | 14:30 |
lucasagomes | cause I still have to deal with it | 14:30 |
lucasagomes | so... | 14:30 |
dtantsur | sambetts, not necessary | 14:30 |
lucasagomes | sambetts, nop, IPA works with all pxe_* drivers | 14:30 |
dtantsur | pxe_ipmi can work with IPA, they differ in image delivery methods | 14:30 |
lucasagomes | it just means deprecating the bash ramdisk | 14:30 |
thiagop | lucasagomes: paste the link to the spec of reorganizing, please | 14:30 |
thiagop | I'd like to take a look on that | 14:30 |
lucasagomes | thiagop, 1 sec | 14:30 |
jroll | lucasagomes: what work does the reorg need, other than probably a rebase and also fixing tests | 14:30 |
lucasagomes | thiagop, https://review.openstack.org/#/c/203628/ | 14:31 |
jroll | thiagop: https://review.openstack.org/#/c/203628/ | 14:31 |
jroll | heh | 14:31 |
thiagop | lucasagomes: thanks | 14:31 |
lucasagomes | jroll, none... fixing tests | 14:31 |
jroll | oh I see your comment now lucas | 14:31 |
jroll | yeah so the tests... require everything to be in a toctree | 14:31 |
jroll | but idk if the placeholders should | 14:31 |
jroll | maybe I can make a hidden page | 14:31 |
jroll | :P | 14:31 |
jroll | I'm writing another spec now, I'll put that up and then fix the reorg | 14:32 |
lucasagomes | ++ | 14:33 |
-openstackstatus- NOTICE: CI is slowly recovering, please be patient while the backlog is worked through. | 14:33 | |
*** ChanServ changes topic to "Bare Metal Provisioning | Status: http://bit.ly/ironic-whiteboard | Docs: http://docs.openstack.org/developer/ironic/ | Bugs: https://bugs.launchpad.net/ironic" | 14:33 | |
sambetts | lucasagomes, dtantsur: now I'm more confused about the differences between agent_* and pxe_*, there is a greater difference than just the RAMDISK they are booted with? I feel like a noob again asking all these questions :-P | 14:35 |
*** linggao has left #openstack-ironic | 14:36 | |
*** vishwana_ has joined #openstack-ironic | 14:36 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic-specs: Add node reservations spec https://review.openstack.org/204641 | 14:37 |
lucasagomes | sambetts, agent_* doesn't use iscsi to deploy the image. They fetch the image from a URL, convert and write it in-band onto the local disk | 14:37 |
dtantsur | sambetts, it's our fault that the names are so confusing :) | 14:37 |
lucasagomes | and it currently doesn't support partition images as well | 14:37 |
lucasagomes | so it doesn't do partitions | 14:37 |
dtantsur | sambetts, I've proposed https://review.openstack.org/#/c/188370/ to change how we compose and name drivers | 14:38 |
lucasagomes | the pxe_* uses iscsi to copy the image over the network onto the local disk | 14:38 |
*** pal has quit IRC | 14:38 | |
jroll | dtantsur: yes please | 14:38 |
lucasagomes | sambetts, dtantsur ++ we named things really badly! | 14:38 |
dtantsur | I don't have much time to update that spec though :( | 14:39 |
sambetts | so with IPA supports both the "agent" way of doing things, and the iscsi way | 14:39 |
dtantsur | it requires some good thinking | 14:39 |
dtantsur | yep | 14:39 |
jroll | dtantsur: I put it on the list for the midcycle | 14:39 |
dtantsur | jroll, great! I won't be there, but if you come up with some ideas feel free to communicate back or just simply update the spec | 14:39 |
*** vishwanathj has quit IRC | 14:39 | |
*** linggao has joined #openstack-ironic | 14:40 | |
*** penick has joined #openstack-ironic | 14:40 | |
jroll | dtantsur: yep, that's my goal | 14:42 |
sambetts | I'm am much less confused now thanks, we need that spec merged though because I would guess others are probably in my situation ;) | 14:42 |
thiagop | jroll: The spec seems fine, but I would suggest a new section on README to indicate how a newcomer would add a spec to the repository | 14:43 |
thiagop | jroll: (the spec to change the spec tree) | 14:43 |
* sambetts now wonders which ramdisk he's running | 14:43 | |
jroll | thiagop: ok, can you leave that feedback on the review please? | 14:44 |
thiagop | jroll: Yeah, sure | 14:44 |
jroll | thanks! | 14:44 |
sambetts | if I used "ramdisk-image-builder ubuntu deploy-ironic -o ..." does that make an IPA ramdisk or a bash one? | 14:47 |
dtantsur | bash one | 14:48 |
sambetts | how do I go about trying the IPA one, there isn't anything in the install-guide about it | 14:51 |
sambetts | ? | 14:51 |
jroll | cd ironic-python-agent/imagebuild/coreos && make | 14:51 |
jroll | or | 14:51 |
jroll | there's one you can download | 14:51 |
jroll | http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 14:52 |
jroll | those are built on every commit | 14:52 |
lucasagomes | sambetts, ^ there's also an element called "ironic-agent" that you can replace the "deploy-ironic" with to build an IPA image using DIB | 14:53 |
lucasagomes | tho I haven't really tested that element | 14:53 |
openstackgerrit | Merged openstack/ironic-specs: Fix typo in release model spec https://review.openstack.org/203631 | 14:53 |
sambetts | ah thanks :D | 14:58 |
*** absubram has joined #openstack-ironic | 14:58 | |
*** marzif has quit IRC | 15:00 | |
*** pelix has quit IRC | 15:01 | |
*** Nisha has joined #openstack-ironic | 15:02 | |
*** absubram has quit IRC | 15:03 | |
*** absubram_ has joined #openstack-ironic | 15:03 | |
jroll | dtantsur: do you still think we shouldn't move specs to $cycle-implemented? | 15:04 |
dtantsur | sorry, I'm on 2 meetings right now :) | 15:05 |
*** pal has joined #openstack-ironic | 15:07 | |
dtantsur | ok, only one meeting now | 15:09 |
openstackgerrit | Thiago Paiva Brito proposed openstack/ironic: OneView Driver for Ironic https://review.openstack.org/191822 | 15:09 |
*** david-lyle has joined #openstack-ironic | 15:09 | |
jroll | heh no worries | 15:09 |
dtantsur | jroll, I'd prefer we don't move them at all, rather change some magic header, and move them in *rendered documentation* | 15:10 |
dtantsur | that's not at any rate a well thought idea though | 15:10 |
openstackgerrit | Merged openstack/ironic-inspector: Fix and enhance contributing documentation on writing plugins https://review.openstack.org/202963 | 15:10 |
* lucasagomes likes the header idea, we probably could get it from the blueprints bits of launchpad | 15:11 | |
lucasagomes | since it says whether the blueprint is completed or not | 15:11 |
*** dprince has quit IRC | 15:12 | |
*** athomas has quit IRC | 15:19 | |
dtantsur | folks, could you please have a look at my mail to the ml? I'm trying to understand whether we can kill NoFreeConductorWorker with move to futurist | 15:22 |
rloo | dtantsur: i might be missing something but I don't even know why you're asking. | 15:25 |
rloo | dtantsur: I mean, why are you asking whether to drop that exception? I don't see any need to keep it | 15:25 |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic-specs: Reorganize specs tree https://review.openstack.org/203628 | 15:26 |
*** pelix has joined #openstack-ironic | 15:26 | |
dtantsur | rloo, I'm afraid that I'm missing some case that e.g. devananda remembers (I wasn't here when this code appeared) | 15:26 |
dtantsur | jroll, https://review.openstack.org/204641 what are you planning to do here? I think we in inspector are doing *something* that may be of use for standalone scheduling | 15:27 |
rloo | dtantsur: should be simple enough to take a look at the code and see where/when that exception is raised? | 15:27 |
*** athomas has joined #openstack-ironic | 15:27 | |
dtantsur | rloo, dropping it is kind of a behavior change. previously we didn't accept tasks if we don't have free workers *now*. after this change we'll enqueue them instead | 15:28 |
jroll | dtantsur: so I want to make: POST /v1/reservations {'root_gb': 2, 'memory_mb': 512, 'instance_uuid': 'blablabla', ...} and have that reserve a node (with instance id) and return that node object | 15:29 |
*** athomas has quit IRC | 15:29 | |
dtantsur | jroll, and by *something* I mean e.g. this: https://blueprints.launchpad.net/ironic-inspector/+spec/advanced-matching and https://blueprints.launchpad.net/ironic-inspector/+spec/assign-profiles | 15:29 |
jroll | dtantsur: basically we want ironic to do the scheduling instead of nova | 15:29 |
dtantsur | jroll, ah, I see it's not entirely related | 15:29 |
rloo | dtantsur: are you asking whether we want to raise that exception if the task is enqueue'd? I don't think we do. | 15:30 |
dtantsur | jroll, but we would appreciate if this API would support capabilities too | 15:30 |
*** athomas has joined #openstack-ironic | 15:30 | |
jroll | dtantsur: I think that's a clear next step, yes :) | 15:30 |
rloo | dtantsur: seems like that question should be in the futurist spec then? | 15:30 |
dtantsur | rloo, I'm rather asking whether we're ok with enqueue instead of raising. Yeah, it should, but I've missed it :( | 15:30 |
rloo | dtantsur: oh, I'm ok with enqueu'ing. Seems like what you're asking is whether that spec should be approved or not... | 15:31 |
dtantsur | jroll, good to hear. What we're trying to do is to have our own matching instead of pretty simple Nova's disk/memory/cpu | 15:31 |
dtantsur | rloo, no, just a follow-up | 15:32 |
dtantsur | jroll, which might or might not overlap with what you're planning. I'd really like to see this spec continued :) | 15:32 |
rloo | dtantsur: ok, just re-read the futurist spec. it doesn't say anything about enqueuing; just executing parallel tasks | 15:34 |
jroll | dtantsur: indeed, I'll be fleshing it out today | 15:34 |
rloo | dtantsur: so i think when i read the spec, my impression was that we could still run out of workers. | 15:34 |
dtantsur | rloo, yep. it's a (potential) side effect of switching to futurist, that's why I raise it | 15:34 |
jroll | dtantsur: my main motivation for this is running multiple nova-computes, however I think it will bring a ton of benefit for many other cases | 15:34 |
dtantsur | and I see it as a positive side effect for reasons stated in the mail | 15:35 |
rloo | dtantsur: so yes, enqueuing is a change in behaviour. which i think should mean an update to the spec. | 15:35 |
rloo | dtantsur: is there a way with futurist to NOT enqueue? | 15:35 |
dtantsur | rloo, no, and I would not want to propose it. that's arguably a wrong thing to do | 15:35 |
rloo | dtantsur: then I think the spec needs to be updated. maybe updated based on the feedback in the email then, dunno. | 15:36 |
rloo | dtantsur: will reply to your email | 15:37 |
dtantsur | rloo, yeah, maybe... lets wait for more people to weigh in | 15:37 |
dtantsur | thanks! | 15:37 |
dtantsur | jroll, yeah, sounds cool | 15:37 |
dtantsur | now calling it a day, see you | 15:38 |
*** dtantsur is now known as dtantsur|afk | 15:38 | |
sambetts | Cya dtantsur|afk | 15:38 |
*** saripurigopi has quit IRC | 15:38 | |
TheJulia | NobodyCam: debian minimal exploded in the dib build :( going to centos | 15:40 |
*** mgoddard has quit IRC | 15:40 | |
*** ukalifon has quit IRC | 15:41 | |
TheJulia | NobodyCam: centos7 currently doesn't build :( | 15:42 |
*** cascone has quit IRC | 15:43 | |
*** cascone has joined #openstack-ironic | 15:44 | |
NobodyCam | uggh | 15:44 |
jroll | lucasagomes: let's land https://review.openstack.org/#/c/203628/ | 15:44 |
jroll | assuming tests pass | 15:45 |
*** aignatov has quit IRC | 15:45 | |
TheJulia | NobodyCam: firing up centos6, its thinking at least | 15:45 |
TheJulia | NobodyCam: (grumpycat) | 15:46 |
* lucasagomes looks | 15:47 | |
lucasagomes | jroll, btw cool thing re /v1/reservations | 15:47 |
jroll | :D | 15:47 |
lucasagomes | I have wondered if we need to have a "instance" resource in our api | 15:47 |
jroll | I need to actually write it | 15:47 |
jroll | mmmmm | 15:47 |
lucasagomes | or something likely what you suggested | 15:47 |
jroll | yeah I guess this is similar | 15:48 |
*** achanda has joined #openstack-ironic | 15:48 | |
lucasagomes | currently the node carry all instance information, that probably could be passed to an instance resource | 15:48 |
lucasagomes | that would pick a node for it | 15:48 |
lucasagomes | anyway yeah... | 15:48 |
* lucasagomes reviews the reorganize the spec | 15:48 | |
jroll | it's a matter of "reserve me a node that matches this" vs "deploy me a node that matches..." | 15:48 |
jroll | imo | 15:48 |
lucasagomes | yeah | 15:48 |
* jroll will consider it | 15:49 | |
lucasagomes | jroll, which could be possible also if we have advaced filters in our API | 15:49 |
lucasagomes | say the json stuff were indexable | 15:49 |
lucasagomes | we could create a query | 15:49 |
jroll | yeah | 15:49 |
lucasagomes | v1/nodes?cps=2&memory_mb>=X etc... | 15:49 |
lucasagomes | that would be cool too | 15:50 |
*** aignatov has joined #openstack-ironic | 15:50 | |
jroll | mhm | 15:50 |
jroll | I think we'll need to split those to a table one day | 15:50 |
*** Sukhdev has joined #openstack-ironic | 15:50 | |
lucasagomes | yeah the json fields would be a table | 15:51 |
lucasagomes | like extra_specs for flavor | 15:51 |
lucasagomes | or metadata for other resources etc | 15:51 |
*** achanda has quit IRC | 15:52 | |
*** mgoddard has joined #openstack-ironic | 15:53 | |
*** achanda has joined #openstack-ironic | 15:55 | |
*** romcheg has quit IRC | 15:55 | |
*** karimb has quit IRC | 15:56 | |
*** coolsvap|away is now known as coolsvap | 15:56 | |
*** achanda has quit IRC | 15:58 | |
*** ijw has joined #openstack-ironic | 16:03 | |
*** uggla_ has joined #openstack-ironic | 16:07 | |
*** absubram_ has quit IRC | 16:08 | |
*** uggla_ has quit IRC | 16:09 | |
devananda | split the json fields into indexable tables -- yes | 16:12 |
sambetts | Night all o/ | 16:13 |
thiagop | night sambetts | 16:14 |
*** mjturek1 has quit IRC | 16:14 | |
*** jistr has quit IRC | 16:19 | |
*** ijw__ has joined #openstack-ironic | 16:20 | |
*** jcoufal has quit IRC | 16:21 | |
*** trown is now known as trown|lunch | 16:22 | |
*** achanda has joined #openstack-ironic | 16:24 | |
*** ijw has quit IRC | 16:24 | |
*** dprince has joined #openstack-ironic | 16:24 | |
NobodyCam | night sambetts | 16:25 |
lucasagomes | devananda, morning | 16:27 |
*** pal has quit IRC | 16:28 | |
lucasagomes | jroll, we have to keep those placeholders under /liberty ? | 16:32 |
jroll | lucasagomes: I mean, the links exist today | 16:33 |
jroll | the entire goal of the placeholder is not to break links | 16:34 |
lucasagomes | oh I see | 16:34 |
*** achanda has quit IRC | 16:34 | |
*** max_lobur has quit IRC | 16:35 | |
*** coolsvap is now known as coolsvap|away | 16:38 | |
*** yog__ has quit IRC | 16:41 | |
*** bizarrochristy has joined #openstack-ironic | 16:48 | |
*** bizarroc_ has joined #openstack-ironic | 16:49 | |
*** ijw has joined #openstack-ironic | 16:51 | |
*** bizarrochristy has quit IRC | 16:53 | |
*** ijw__ has quit IRC | 16:55 | |
*** pal has joined #openstack-ironic | 16:56 | |
*** athomas has quit IRC | 16:56 | |
lucasagomes | folks I will call it a day, if you guys have some time to take a look at https://review.openstack.org/#/c/204162/ today that would be great! | 16:57 |
lucasagomes | have a good night everyone! | 16:57 |
jroll | night lucas! | 17:00 |
NobodyCam | night lucasagomes | 17:01 |
*** romainh has quit IRC | 17:02 | |
lucasagomes | see y'all | 17:03 |
*** lucasagomes is now known as lucas-dinner | 17:03 | |
*** derekh has quit IRC | 17:04 | |
*** athomas has joined #openstack-ironic | 17:05 | |
*** zz_natorious is now known as natorious | 17:06 | |
jroll | last call for eyes on https://review.openstack.org/#/c/188528/ | 17:08 |
jroll | I'm going to land it later today if nobody objects | 17:08 |
jroll | Sukhdev: ^^ fyi, that will likely land today | 17:08 |
thiagop | night lucas-dinner | 17:08 |
NobodyCam | jroll: added my +2 | 17:09 |
NobodyCam | :) | 17:09 |
*** stendulker has joined #openstack-ironic | 17:09 | |
Sukhdev | jroll: I think you should go for it….. | 17:09 |
Sukhdev | jroll: the approval is written all over it :-):-) | 17:10 |
Sukhdev | NobodyCam: thanks for taking time to review it - | 17:10 |
NobodyCam | jroll: Sukhdev: and everyone else... awesome work on that spec | 17:10 |
NobodyCam | I have been following its progression | 17:11 |
*** puranamr has joined #openstack-ironic | 17:12 | |
*** puranamr has quit IRC | 17:13 | |
jroll | if NobodyCam says it's ready than it must be ready :D | 17:13 |
jroll | and +a | 17:14 |
*** marzif has joined #openstack-ironic | 17:15 | |
*** puranamr has joined #openstack-ironic | 17:15 | |
*** absubram has joined #openstack-ironic | 17:18 | |
*** e0ne has quit IRC | 17:24 | |
*** stendulker_ has joined #openstack-ironic | 17:25 | |
*** marzif has quit IRC | 17:25 | |
*** marzif has joined #openstack-ironic | 17:26 | |
*** stendulker has quit IRC | 17:28 | |
NobodyCam | w00t | 17:30 |
openstackgerrit | Merged openstack/ironic-specs: Update of the Ironic Neutron Integration spec https://review.openstack.org/188528 | 17:30 |
NobodyCam | :) | 17:30 |
*** marzif has quit IRC | 17:31 | |
*** stendulker_ has quit IRC | 17:31 | |
*** marzif has joined #openstack-ironic | 17:32 | |
TheJulia | NobodyCam: re-blocking changing of the default, issues with simple-init. Going to rebuild my image with devuser so I can get into it and debug | 17:33 |
*** Sukhdev has quit IRC | 17:33 | |
NobodyCam | TheJulia: ack :) | 17:35 |
*** trown|lunch is now known as trown | 17:37 | |
TheJulia | NobodyCam: the totally not fun part of this, building images of various flavors and then poking around trying to figure out what went sideways :( | 17:37 |
*** linggao has quit IRC | 17:38 | |
openstackgerrit | Merged openstack/bifrost: Removal of legacy roles and related updates https://review.openstack.org/203768 | 17:42 |
*** davideagnello has joined #openstack-ironic | 17:43 | |
NobodyCam | SpamapS: please check that ^^^ will not effect your efforts | 17:44 |
*** davideagnello has quit IRC | 17:45 | |
*** pelix has quit IRC | 17:45 | |
*** tab____ has joined #openstack-ironic | 17:48 | |
*** stendulker has joined #openstack-ironic | 17:52 | |
SpamapS | NobodyCam: I perused it yesterday, and really, I'll deal. | 17:53 |
SpamapS | NobodyCam: I suspect you're making it better faster harder stronger .. so carry on. :) | 17:53 |
*** pal has quit IRC | 17:55 | |
*** kkoski has joined #openstack-ironic | 17:57 | |
NobodyCam | SpamapS: woot ;) that removing some old things so wanted to give ya heads up just in case | 17:57 |
NobodyCam | :) | 17:57 |
*** kkoski has quit IRC | 17:58 | |
SpamapS | Yeah I mean, I have a baremetal csv file, but I made it from an extensive set of yaml files that greghaynes made | 17:59 |
SpamapS | so I suspect I can make the new enrolling input thing work with those | 17:59 |
*** stendulker has quit IRC | 18:01 | |
*** marzif has quit IRC | 18:02 | |
NobodyCam | SpamapS: see the "Conversion from CSV to JSON formats" section of https://github.com/openstack/bifrost/blob/master/README.rst#L186 | 18:02 |
SpamapS | NobodyCam: I'll just go right from my yaml to json thankyouverymuch | 18:03 |
*** ijw has quit IRC | 18:03 | |
NobodyCam | hehehe | 18:03 |
NobodyCam | :) | 18:03 |
*** marzif has joined #openstack-ironic | 18:04 | |
rloo | hi devananda, if you have a few minutes, this spec has two +2, but dmitry wants to make sure your concerns were addressed. | 18:05 |
rloo | devananda: and the spec is ... https://review.openstack.org/#/c/189341/ | 18:06 |
*** Sukhdev has joined #openstack-ironic | 18:06 | |
*** krtaylor has quit IRC | 18:06 | |
jroll | rloo: "But v2 seems to only allow UUID" | 18:07 |
jroll | seems problematic to me? | 18:07 |
*** marzif_ has joined #openstack-ironic | 18:07 | |
jroll | which seems like one of deva's concerns | 18:08 |
rloo | jroll: oh, you're making me look at it? ... | 18:08 |
jroll | rloo: "glance v2 api doesn't allow querying images by id" | 18:08 |
jroll | rloo: also concerned about glance image names with spaces/unicode/etc | 18:09 |
*** mgoddard has quit IRC | 18:09 | |
rloo | jroll: i don't think i had looked at deva's comments before. sec. | 18:09 |
*** marzif has quit IRC | 18:11 | |
*** Sukhdev has quit IRC | 18:11 | |
*** alex_xu has quit IRC | 18:11 | |
jroll | rloo: lucas made the same comment (but +2'd it?) | 18:12 |
rloo | jroll: yeah, just noticed that. Sigh. need to -1 I think. | 18:12 |
NobodyCam | rloo: deva is at oscon atm so he may be a bit slow to reply | 18:12 |
jroll | rloo: I also -1'd | 18:13 |
rloo | jroll: are we moving/using v2 in glance? i can't keep track. | 18:13 |
jroll | rloo: we have code for it, at least some calls to glance use it | 18:13 |
rloo | NobodyCam: thx for letting us know. | 18:13 |
jroll | I believe agent driver depends on v2 | 18:13 |
*** puranamr has quit IRC | 18:13 | |
jroll | but am not 100% | 18:13 |
*** alex_xu has joined #openstack-ironic | 18:13 | |
rloo | jroll: wondering if it makes sense to allow it for v1, but i don't like that. | 18:13 |
jroll | rloo: agree, I really don't like allowing features depending on glance version | 18:14 |
jroll | I also don't see why name is easier than UUID, either way I think you'll end up doing a copy/paste operation | 18:15 |
jroll | but that isn't for me to decide I guess | 18:15 |
rloo | jroll: well, names are always easier than uuids! | 18:15 |
jroll | are they though? | 18:16 |
jroll | I can't remember if my Centos image is Centos or CENTOS or CentOS | 18:16 |
jroll | rloo: for example, the image used in devstack is named cirros-0.3.4-x86_64-disk | 18:18 |
jroll | good luck remembering that :) | 18:18 |
rloo | jroll: ok, you're unusual. or maybe you're normal. I dunno. Guess it depends on the users/operators. | 18:18 |
jroll | rloo: I may be weird :) | 18:18 |
rloo | jroll: why would glance v2 allow you to specify a name for an image, but not filter on it. Well, I don't want to dwell on this spec. moving along... | 18:19 |
jroll | in general, I can remember enough to grep, but not to write it verbatim | 18:19 |
*** marzif_ has quit IRC | 18:19 | |
*** puranamr has joined #openstack-ironic | 18:19 | |
jroll | also this is a one time thing, right? it's not like day-to-day use involves putting glance IDs in the database | 18:19 |
* jroll questioning usefulness but will let it go for now | 18:19 | |
rloo | jroll: even if you can't remember the name, if you saw a list of names vs a list of UUIDs, it'll be easier to know which one you wanted. | 18:19 |
rloo | jroll: will see if vladyslav can elaborate on use case or whatever | 18:20 |
jroll | rloo: but "image list" gives a list of names AND uuids, and at that point I can copy paste | 18:20 |
jroll | but yeah it's fine | 18:20 |
jroll | in terms of use cases | 18:20 |
rloo | jroll: so we could do a glance 'image list' and find the uuid corresponding to the name, which will take a few seconds or whatever longer, but might make it easier for the user... | 18:21 |
jroll | rloo: sure, but I want that called out if that's what we want to do :) | 18:21 |
thiagop | it's easier even to 'sed', I think... | 18:22 |
* devananda pops in for a bit | 18:27 | |
*** davideagnello has joined #openstack-ironic | 18:28 | |
trown | Is there any documentation on using/writing IPA extensions? | 18:30 |
*** Sukhdev has joined #openstack-ironic | 18:31 | |
jroll | trown: apparently not, for extensions :( | 18:32 |
jroll | there are hardware manager docs though | 18:32 |
JoshNang | trown: what are you trying to do? | 18:32 |
jroll | so just to be clear, you're looking for docs on a command extension? | 18:32 |
trown | we want to use IPA in place of the inspector ramdisk | 18:33 |
trown | I think dtantsur|afk has a better idea of how to do that, but I finally got a working CentOS based IPA ramdisk to boot and start heartbeating, so a bit excited to start hacking on it | 18:35 |
JoshNang | trown: nice! | 18:35 |
JoshNang | extensions are pretty straight forward..a class that extends base.BaseAgentExtension, some functions decorated with @base.async_command or @base.sync_command to expose them to the API | 18:35 |
trown | JoshNang: how are those then exposed? directly as vendor-passthru methods? | 18:38 |
*** Nisha has quit IRC | 18:38 | |
NobodyCam | brb | 18:39 |
*** achanda has joined #openstack-ironic | 18:40 | |
*** Nisha has joined #openstack-ironic | 18:41 | |
JoshNang | trown: they'll be available at /v1/commands in the agent API, not the ironic API. to kick them off you POST to that url with a json body with two keys: {"name": "$extension.$method", "params": {$some_params}} | 18:41 |
JoshNang | https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent_client.py for reference how we interact with the agent API | 18:41 |
*** krtaylor has joined #openstack-ironic | 18:42 | |
trown | JoshNang: ah perfect, that gives me something to play around with. Thanks! | 18:42 |
JoshNang | trown: np! feel free to ping if you get stuck at all | 18:42 |
*** e0ne has joined #openstack-ironic | 18:45 | |
*** Nisha has quit IRC | 18:46 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic-specs: Add node reservations spec https://review.openstack.org/204641 | 18:46 |
jroll | devananda: ^^ | 18:46 |
jroll | trown: there's also a --standalone option if you want to be able to spin up an agent without having an ironic environment around | 18:47 |
trown | jroll: then it does not heartbeat right? | 18:49 |
JoshNang | trown: correct | 18:50 |
*** igordcard has joined #openstack-ironic | 18:54 | |
*** krtaylor has quit IRC | 18:55 | |
*** davideagnello has quit IRC | 19:01 | |
*** Marga_ has quit IRC | 19:04 | |
*** Marga_ has joined #openstack-ironic | 19:05 | |
devananda | jroll: erm, that is very much not what jaypipes and i discussed in YVR | 19:06 |
devananda | jroll: what I recall -- we should expose an API to search for nodes having some properties, and an atomic way to reserve one or more such nodes | 19:07 |
devananda | we already present the second part of that through updating the instance_uuid (there's some db foo that makes it atomic) | 19:07 |
*** Marga_ has quit IRC | 19:07 | |
*** Marga_ has joined #openstack-ironic | 19:08 | |
devananda | jroll: this proposal doesn't say anything about integrating or using the logic in the nova scheduler, or what scheduling mechanisms it would support | 19:09 |
openstackgerrit | Merged openstack/ironic: Clean up tftp files if agent deployed disk image https://review.openstack.org/204265 | 19:13 |
jroll | devananda: this does both in one endpoint and is roughly what we discussed here | 19:16 |
jroll | devananda: I'm writing a nova spec for the nova bits (which will be in the compute and not in the scheduler) | 19:16 |
devananda | jroll: does it leverage the nova scheduler's functionality for, well, scheduling? | 19:16 |
jroll | devananda: and I'm not sure what you mean by "scheduling mechanisms it would support" | 19:16 |
devananda | jroll: this does not implement search | 19:17 |
jroll | devananda: no, the nova scheduler at this point would be reduced to "pick a compute host at random because it doesn't matter" | 19:17 |
devananda | it implements scheduling/reservation | 19:17 |
jroll | uh | 19:17 |
jroll | right. | 19:17 |
jroll | that is my goal. a simple scheduler in ironic | 19:17 |
devananda | which is not "return a list of nodes that have X, Y, !Z" | 19:17 |
jroll | to be clear, this is a first draft | 19:17 |
devananda | sure | 19:18 |
jroll | but my general plan is "ironic, reserve me a node that has X, Y, !Z" | 19:18 |
jroll | nova virt driver does that, and then deploys to it | 19:18 |
devananda | uh | 19:18 |
devananda | virt driver? | 19:18 |
jroll | yes. | 19:18 |
devananda | that seems like a violation of nova's architecture | 19:18 |
jroll | the goal is to reduce the scheduler to choosing a host, not a (host, node) | 19:18 |
openstackgerrit | Merged openstack/ironic: Fix the DEPLOYWAIT check for agent_* drivers https://review.openstack.org/203157 | 19:18 |
devananda | shouldn't the nova scheduler be doing htat/ | 19:18 |
jroll | the nova team generally agreed this was a good path | 19:19 |
devananda | scheduer should be choosing a (node) | 19:19 |
devananda | omg :( | 19:19 |
jroll | no | 19:19 |
jroll | because that violates nova's architecture even more | 19:19 |
openstackgerrit | Merged openstack/ironic: Don't do a premature reservation check in the provision API https://review.openstack.org/204081 | 19:19 |
devananda | for capabilities like affinity / anti affinity, we'll need to essentialy re-implement ALL of the logic thta's in the nova scheduler already | 19:20 |
devananda | and push down all the knowledge nova has about user and instance | 19:20 |
devananda | into Ironic | 19:20 |
devananda | we've ALWAYS discussed pushing Ironic's (potential) rack / location / hardware awareness UP to Nova | 19:20 |
devananda | not pushing Nova's user / instance information DOWN | 19:21 |
*** saripurigopi has joined #openstack-ironic | 19:22 | |
devananda | I gotta run - but I'm clearly not in favor of re-implementing scheduling inside Ironic | 19:22 |
devananda | and would much rather make our API provide the information needed for external scheduling services | 19:23 |
lucas-dinner | jroll, rloo re glance image by name | 19:23 |
rloo | hi lucas-dinner | 19:23 |
lucas-dinner | I will just double check, but v2 still supports name | 19:23 |
openstackgerrit | Merged openstack/python-ironicclient: Add provision_updated_at field to nodes https://review.openstack.org/203585 | 19:23 |
openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Out-of-band hardware introspection support for Cisco B/C/M-series servers https://review.openstack.org/204733 | 19:23 |
lucas-dinner | but would require 2 API calls I think to get it... 1 sec lemme recheck | 19:24 |
lucas-dinner | long time I made that comment so I forgot | 19:24 |
*** puranamr has quit IRC | 19:27 | |
*** romainh has joined #openstack-ironic | 19:28 | |
*** puranamr has joined #openstack-ironic | 19:28 | |
jroll | devananda: what about the standalone use case | 19:31 |
lucas-dinner | hmmmm I may be wrong damn I thought there was a way to get by name on v2 | 19:34 |
*** saripurigopi has quit IRC | 19:39 | |
lucas-dinner | JoshNang, re instance_uuid... My initial idea was to allow it to be removed on DEPLOY* like it works for CLEAN* | 19:43 |
lucas-dinner | JoshNang, ideally we should clean that up at the end of tear_down as we do with instance_info() | 19:43 |
NobodyCam | anyone happen to know if there is a spec for this: https://blueprints.launchpad.net/ironic/+spec/remove-swift-dependency-for-ilo-drivers | 19:44 |
lucas-dinner | but that would be a non-backward compat change since Nova will pool for the node using the instance UUID | 19:44 |
*** [1]cdearborn has joined #openstack-ironic | 19:46 | |
JoshNang | lucas-dinner: hrm :/ | 19:50 |
lucas-dinner | JoshNang, I can investigate see if there's a better way to do it | 19:51 |
lucas-dinner | otherwise we make updates really hard | 19:51 |
JoshNang | lucas-dinner: ++ i'm looking at the nova driver code right now | 19:51 |
*** romainh has left #openstack-ironic | 19:52 | |
lucas-dinner | JoshNang, I'm not sure what would happen if we remove it as part of the tear_down() I don't think it may break nova but it will raise some ugly exceptions when it's pooling for the state | 19:52 |
lucas-dinner | the looping call will basically exit due the exception NodeNotFound | 19:52 |
lucas-dinner | which is not handled there | 19:53 |
JoshNang | i'm assuming the instance would go to error then | 19:54 |
*** karimb has joined #openstack-ironic | 19:54 | |
lucas-dinner | yeah I would have to test it | 19:56 |
lucas-dinner | that's why allowing it to be changed in DEPLOY* may be safier | 19:56 |
lucas-dinner | safer* | 19:56 |
lucas-dinner | tho it's ugly | 19:56 |
lucas-dinner | but can be fixed all at once CLEAN* and DEPLOY* | 19:57 |
JoshNang | eh allowing during DEPLOY* doesn't seem that bad. you could check that the delete_instance flag is set | 19:59 |
lucas-dinner | yeah | 19:59 |
lucas-dinner | actually I'm looking at the nova code and thinking... because if it's an old nova code it won't even issue the "deleted" command for nodes in DEPLOYING | 20:01 |
*** puranamr has quit IRC | 20:01 | |
lucas-dinner | https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L865-L869 | 20:01 |
lucas-dinner | JoshNang, so delete_instance will never be set and that should fail to remove the instance_uuid | 20:01 |
*** cdearborn has quit IRC | 20:02 | |
JoshNang | lucas-dinner: ah! | 20:02 |
lucas-dinner | so if we don't allow it to be removed and just remove it as part of the tear_down it may work | 20:03 |
lucas-dinner | (it's late here my brain is a bit slow) | 20:03 |
lucas-dinner | JoshNang, I will investigate more tomorrow, but yeah, in any case I will keep it backward compat, even if that means allowing changes on DEPLOY* | 20:04 |
JoshNang | lucas-dinner: alright | 20:04 |
*** krtaylor has joined #openstack-ironic | 20:04 | |
*** marzif_ has joined #openstack-ironic | 20:05 | |
JoshNang | i must be missing something, because this doesn't look like it should work well today..if the node is DEPLOYING...don't try to tear down, just try to remove the instance uuid? and get an error because you can't do that? | 20:05 |
lucas-dinner | yeah today deletion is a disaster | 20:05 |
JoshNang | heh | 20:05 |
lucas-dinner | we basically can't delete if the node is spawming | 20:05 |
lucas-dinner | the thing is, we never tried such things | 20:06 |
lucas-dinner | that's what I'm trying to fix here | 20:06 |
lucas-dinner | because downstream we use heat + nova + ironic | 20:06 |
lucas-dinner | so when heat stack fails and it tries to rollback it calls nova deletes | 20:06 |
JoshNang | ++ to fixing it | 20:06 |
lucas-dinner | which tries to delete the things in Ironic even when it's spawning (which is fine on nova POV) | 20:06 |
lucas-dinner | but we end up having nodes stuck | 20:06 |
lucas-dinner | so it's a bit of a can of worm there | 20:07 |
lucas-dinner | downstream == rh installer | 20:07 |
JoshNang | bleh yeah | 20:08 |
lucas-dinner | JoshNang, right I can update the spec about the instance_uuid if you prefer | 20:08 |
jroll | lucas-dinner: JoshNang fwiw this reservation thing solves this instance id problem ;D | 20:08 |
JoshNang | lucas-dinner: please do, i feel like there will be some debate there | 20:09 |
jroll | https://review.openstack.org/#/c/204641/ | 20:09 |
JoshNang | jroll: :D | 20:09 |
lucas-dinner | JoshNang, right I will investigate it and update accordingly tomorrow | 20:09 |
JoshNang | lucas-dinner: thanks! have a good night! | 20:09 |
jroll | since the instance id will be set/removed with the reservation, not as a separate PATCH request or whatever | 20:09 |
lucas-dinner | jroll, ++ I will review that (didn't have time today) | 20:09 |
jroll | lucas-dinner: yeah no rush :) | 20:10 |
lucas-dinner | cool I will be afk! | 20:10 |
jroll | later dude | 20:10 |
lucas-dinner | thanks for the discussion folks | 20:10 |
*** Pradip has joined #openstack-ironic | 20:10 | |
lucas-dinner | later! have a good afternoon there | 20:11 |
JoshNang | o/ | 20:13 |
jroll | devananda: so, I got distracted, but looping back - this is really the best way to deal with this problem, IMO | 20:13 |
jroll | devananda: and it also helps anyone not using nova, so they don't need to re-implement scheduling | 20:13 |
jroll | devananda: if we only provide the API to assist with scheduling, then I just see us building scheduling into the client at some point, which is re-implementing things anyway | 20:14 |
*** pal has joined #openstack-ironic | 20:19 | |
*** pal has quit IRC | 20:24 | |
*** e0ne has quit IRC | 20:26 | |
*** Sukhdev has quit IRC | 20:31 | |
*** davideagnello has joined #openstack-ironic | 20:43 | |
*** absubram has quit IRC | 20:47 | |
*** puranamr has joined #openstack-ironic | 20:50 | |
*** trown is now known as trown|outttypeww | 20:56 | |
jroll | someone pls to https://review.openstack.org/#/c/203628/ | 20:58 |
jroll | before we keep merging things | 20:58 |
*** igordcar1 has joined #openstack-ironic | 21:03 | |
*** igordcar1 has quit IRC | 21:03 | |
*** davideagnello has quit IRC | 21:03 | |
mrda | Morning Ironic | 21:04 |
*** ijw has joined #openstack-ironic | 21:04 | |
*** krtaylor has quit IRC | 21:06 | |
*** igordcar1 has joined #openstack-ironic | 21:08 | |
*** igordcar1 has quit IRC | 21:08 | |
NobodyCam | good morning mrda | 21:09 |
mrda | o/ NobodyCam | 21:09 |
*** stendulker has joined #openstack-ironic | 21:10 | |
TheJulia | good morning mrda | 21:11 |
mrda | hi TheJulia | 21:12 |
*** tab____ has quit IRC | 21:12 | |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Add CLEANWAIT state https://review.openstack.org/200152 | 21:14 |
*** david-lyle has quit IRC | 21:17 | |
*** jjohnson2 has quit IRC | 21:22 | |
*** marzif_ has quit IRC | 21:25 | |
*** stendulker has quit IRC | 21:26 | |
*** e0ne has joined #openstack-ironic | 21:30 | |
*** ijw__ has joined #openstack-ironic | 21:32 | |
*** ijw has quit IRC | 21:35 | |
openstackgerrit | John L. Villalovos proposed openstack/ironic: Add CLEANWAIT state https://review.openstack.org/200152 | 21:36 |
NobodyCam | jlvillal: you back? | 21:37 |
jlvillal | NobodyCam, No. Just happened to update a commit message :) | 21:37 |
jlvillal | Still on vacation! | 21:37 |
NobodyCam | lol well then keep enjoying | 21:37 |
jlvillal | NobodyCam, About two more weeks :) | 21:37 |
jlvillal | NobodyCam, Thanks! | 21:37 |
NobodyCam | thats what I thought | 21:38 |
NobodyCam | :-p | 21:38 |
*** Sukhdev has joined #openstack-ironic | 21:38 | |
*** achanda has quit IRC | 21:43 | |
*** kkoski has joined #openstack-ironic | 21:46 | |
*** thiagop has quit IRC | 21:47 | |
*** penick has quit IRC | 21:51 | |
*** cing has joined #openstack-ironic | 21:53 | |
*** e0ne has quit IRC | 21:58 | |
*** igordcard_ has joined #openstack-ironic | 22:00 | |
*** igordcard has quit IRC | 22:01 | |
*** [1]cdearborn has quit IRC | 22:02 | |
*** lucas-dinner has quit IRC | 22:06 | |
*** cing has quit IRC | 22:06 | |
*** cing has joined #openstack-ironic | 22:06 | |
*** bnemec has quit IRC | 22:09 | |
*** achanda has joined #openstack-ironic | 22:15 | |
*** lucas-dinner has joined #openstack-ironic | 22:18 | |
*** cing has quit IRC | 22:22 | |
*** cing has joined #openstack-ironic | 22:28 | |
*** dprince has quit IRC | 22:31 | |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Generalize the dib builder role https://review.openstack.org/199683 | 22:35 |
NobodyCam | jroll: quick question on the ml2 stuff.. we ARE planning both vlan and vxlan support? | 22:47 |
*** kkoski has quit IRC | 22:58 | |
*** kkoski has joined #openstack-ironic | 22:58 | |
*** bizarroc_ has quit IRC | 22:58 | |
*** igordcard has joined #openstack-ironic | 23:00 | |
*** krtaylor has joined #openstack-ironic | 23:04 | |
*** lucas-dinner has quit IRC | 23:08 | |
*** thrash|brb is now known as thrash|g0ne | 23:08 | |
*** davideagnello has joined #openstack-ironic | 23:09 | |
*** kkoski has quit IRC | 23:11 | |
*** karimb has quit IRC | 23:11 | |
*** cing has quit IRC | 23:13 | |
*** igordcard_ has quit IRC | 23:17 | |
*** igordcard_ has joined #openstack-ironic | 23:18 | |
*** igordcard_ has quit IRC | 23:30 | |
*** igordcard_ has joined #openstack-ironic | 23:30 | |
*** praneshp has joined #openstack-ironic | 23:31 | |
*** Sukhdev has quit IRC | 23:38 | |
*** igordcard_ has quit IRC | 23:40 | |
*** igordcard_ has joined #openstack-ironic | 23:41 | |
openstackgerrit | Chris Krelle proposed openstack/bifrost: Vagrantfile and vagrant.yml for testing https://review.openstack.org/191110 | 23:42 |
*** naohirot has joined #openstack-ironic | 23:47 | |
*** achanda has quit IRC | 23:49 | |
*** davideagnello has quit IRC | 23:53 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!