| kfox1111 | netstat says its listening... | 00:01 |
|---|---|---|
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Minor fix to install guide for associating k&r to nodes https://review.openstack.org/136187 | 00:01 |
| NobodyCam | kfox1111: the node is attempting to pxe boot. is it getting dhcp address and trying to pull the boot image? | 00:01 |
| kfox1111 | haven't gotten there yet. just trying a tftp client from localhost to see if thats working. | 00:02 |
| NobodyCam | ahh that was the only full example of node create we had | 00:03 |
| *** igordcard has quit IRC | 00:03 | |
| NobodyCam | devananda: ^^^ | 00:03 |
| *** Viswanath has joined #openstack-ironic | 00:04 | |
| NobodyCam | kfox1111: I may have asked already but forgotten what is you os? | 00:04 |
| devananda | NobodyCam: yea, and in the wrong place | 00:05 |
| devananda | I'll add another one | 00:05 |
| NobodyCam | w00t :) hehehehe | 00:05 |
| *** penick has quit IRC | 00:05 | |
| devananda | oh jeez ... the user_guide still references teh Baremetal wiki :( | 00:06 |
| NobodyCam | omg did you remove that like months ago | 00:07 |
| NobodyCam | :( | 00:07 |
| NobodyCam | s/did/didn't/ | 00:07 |
| devananda | yes | 00:08 |
| kfox1111 | can't delete nova instance... {"message": "Error destroying the instance on node 9cb00e2b-c117-427a-9fc8-ab62487579e8. Provision state still 'error' | 00:09 |
| *** Viswanath has quit IRC | 00:09 | |
| kfox1111 | I think I fixed the problem that caused it to get into that state on launch, but I can't delete to retry. | 00:09 |
| devananda | kfox1111: there are probably better ways to fix that, but the simplest one: restart nova-compute | 00:10 |
| kfox1111 | k. | 00:10 |
| devananda | kfox1111: also: ironic node-show NNNN | 00:11 |
| devananda | should give more details on the error | 00:11 |
| kfox1111 | still no good. | 00:11 |
| kfox1111 | I set the state on the node to deleted. it seems happy, but nova wont delete it. | 00:11 |
| devananda | what does ironic node-show show? | 00:12 |
| kfox1111 | looks happy. power_state poer off provision_state None target_provision_state None last_error None | 00:12 |
| devananda | yep - that's good | 00:15 |
| devananda | kfox1111: instance_uuid None too? | 00:15 |
| kfox1111 | ah. no, its 609d56b5-1f0a-4059-a508-e54f9c8b765d | 00:16 |
| devananda | ahh | 00:16 |
| NobodyCam | :) | 00:16 |
| devananda | not sure why nova won't finish cleaning up there | 00:16 |
| devananda | if you unset the instance_uuid, nova should realize the instance is gone and delete it's references | 00:17 |
| NobodyCam | devananda: you know if there is a bug around that | 00:17 |
| devananda | NobodyCam: i thought we squashed those bugs last cycle ... | 00:18 |
| *** ryanpetrello has quit IRC | 00:18 | |
| devananda | it's possible there's another (or a new one) ? | 00:18 |
| NobodyCam | yea... | 00:18 |
| kfox1111 | still won't delete. :/ | 00:19 |
| kfox1111 | the instance_uuid is gone for sure. | 00:19 |
| NobodyCam | whats nova saying | 00:19 |
| kfox1111 | "message": "Error destroying the instance on node 9cb00e2b-c117-427a-9fc8-ab62487579e8. Provision state still 'error'.", "code": 500, "details": " File \"/usr/lib/python2.7/site-packages/nova/compute/manager.py\", line 314, in decorated_function | 00:19 |
| *** r-daneel has quit IRC | 00:20 | |
| devananda | kfox1111: at this point, you should be able to just create a new instance ... | 00:20 |
| kfox1111 | quota doesn't allow it. | 00:21 |
| devananda | hah | 00:21 |
| kfox1111 | :) | 00:21 |
| devananda | welp. there are "can't delete this instance" bugs in Nova that are unrelated to Ironic, too. | 00:21 |
| *** alexpilotti has joined #openstack-ironic | 00:21 | |
| kfox1111 | fair enough. | 00:21 |
| kfox1111 | bumping quota... :) | 00:21 |
| devananda | I dunno what this one is caused by ... restarting n-cpu should cause the compute manager to clean that up | 00:21 |
| devananda | but if it's not ... then I haven't seen that issue before | 00:22 |
| NobodyCam | yea | 00:22 |
| kfox1111 | hmm... back to no valid host was found.... | 00:22 |
| kfox1111 | its like something in nova thinks ironic's node is allocated. | 00:23 |
| *** ryanpetrello has joined #openstack-ironic | 00:24 | |
| NobodyCam | can you paste the output of nova list and ironic node-list | 00:24 |
| kfox1111 | sure. | 00:25 |
| kfox1111 | http://pastebin.com/PJ92GsFk | 00:26 |
| kfox1111 | maybe I need to force the vm into a better state, then retry? | 00:26 |
| kfox1111 | ah. there we go... | 00:27 |
| kfox1111 | nova reset-state seemed to let me delete it. | 00:27 |
| kfox1111 | OSError: [Errno 13] Permission denied: '/tftpboot/9cb00e2b-c117-427a-9fc8-ab62487579e8' rarg.... | 00:28 |
| NobodyCam | ya looks like nova lost something. NOSTATE for power is not a good thing | 00:28 |
| NobodyCam | ahh yep tftp dir need to be owned by the user the conductor is running as | 00:29 |
| kfox1111 | hmm.... chown ironic /tftpboot seems to have done the trick. :) | 00:30 |
| NobodyCam | :) | 00:30 |
| kfox1111 | looks like the node's finally trying to boot. :) | 00:30 |
| devananda | kfox1111: I don't suppose you're recording the bumps you're hitting in the docs, and planning to post a patch? | 00:30 |
| kfox1111 | I'm recording everything in a shell script so I can redeploy. I should be able to get feedback back. :) | 00:31 |
| NobodyCam | lol devananda took the words right off my keyboard | 00:31 |
| kfox1111 | some of the rpm's are missing dependencies too. :) | 00:31 |
| devananda | awesome, ty | 00:31 |
| kfox1111 | sure. :) | 00:31 |
| kfox1111 | hmm... netboot fail... | 00:32 |
| NobodyCam | did it get a dpck address and attemp to pull the boot image? | 00:32 |
| kfox1111 | didn't have the console up at the time. deleting vm and trying again... | 00:33 |
| *** david-lyle is now known as david-lyle_afk | 00:33 | |
| kfox1111 | does setting pxe persistant over ipmi work on all nodes? | 00:35 |
| kfox1111 | hmm... looks like that part might not have worked. | 00:36 |
| *** ryanpetrello_ has joined #openstack-ironic | 00:37 | |
| kfox1111 | hmm... nova's in task_state deleting vm_state building..... for a long time. | 00:37 |
| kfox1111 | do I need to manually go into the bios and set pxe to always be first? | 00:37 |
| *** naohirot has joined #openstack-ironic | 00:38 | |
| *** ryanpetrello has quit IRC | 00:38 | |
| *** ryanpetrello_ is now known as ryanpetrello | 00:38 | |
| naohirot | good morning ironic | 00:38 |
| kfox1111 | hmm... ironic node-get-boot-device is showing boot_device None | 00:39 |
| NobodyCam | kfox1111: how are you controling the node ipmi? | 00:42 |
| NobodyCam | morning naohirot | 00:42 |
| kfox1111 | what do you mean? | 00:42 |
| naohirot | NobodyCam: GM | 00:42 |
| NobodyCam | is this a HW node or vm? | 00:42 |
| kfox1111 | hardware. | 00:42 |
| kfox1111 | real ipmi. | 00:42 |
| NobodyCam | ipmi should be able to set the boot device | 00:43 |
| kfox1111 | I tried this: ironic node-set-boot-device --persistent 9cb00e2b-c117-427a-9fc8-ab62487579e8 pxe | 00:43 |
| kfox1111 | but it didn't seem to stick. | 00:43 |
| kfox1111 | I just did it again, and am double checking... | 00:43 |
| kfox1111 | hmm... so this time its trying to netboot. | 00:43 |
| kfox1111 | hmm.... ok. weak. cobbler took it.... | 00:44 |
| NobodyCam | :-p | 00:44 |
| *** ryanpetrello_ has joined #openstack-ironic | 00:44 | |
| *** ryanpetrello has quit IRC | 00:47 | |
| kfox1111 | the tune to dueling bajo's is in my head now. :/ | 00:49 |
| *** ryanpetrello_ has quit IRC | 00:50 | |
| *** spandhe has joined #openstack-ironic | 00:50 | |
| *** alexpilotti has quit IRC | 00:50 | |
| openstackgerrit | Merged openstack/ironic: Support configdrive in agent driver https://review.openstack.org/128388 | 00:53 |
| kfox1111 | BTW, whats the status of booting local off of ironic managed nodes? | 00:53 |
| kfox1111 | hmm.... yes, the --persistent flag to node-set-boot-device seems to not work on this hardware. :/ | 00:54 |
| jroll | you can do local boot with the agent driver | 00:57 |
| jroll | requires full disk images atm but hey | 00:57 |
| kfox1111 | agent driver? | 00:57 |
| kfox1111 | ok... so cobbler dhcp is leaving it alone now, but no joy from neutron.... | 00:58 |
| kfox1111 | ah.... bridge's wrong. | 00:59 |
| jroll | it's a different deploy driver, ignore me for now | 01:00 |
| kfox1111 | oh. ok. | 01:00 |
| kfox1111 | jroll: in juno or in development? | 01:01 |
| jroll | in juno | 01:01 |
| jroll | also making it better in dev | 01:01 |
| kfox1111 | hmm.. ok. I'll put it on my to look at list. Thanks. :) | 01:02 |
| devananda | kfox1111: wait, cobbler? | 01:02 |
| kfox1111 | devananda: had to deploy my ironic box somehow. ;) | 01:02 |
| devananda | kfox1111: AFAIK, if you have cobbler running on the same network, you'll need to exclude the Node from cobbler's DHCP service | 01:02 |
| devananda | otherwise the two will race ... | 01:02 |
| kfox1111 | yup. found and squashed. | 01:02 |
| devananda | k | 01:02 |
| devananda | oh ya. you said that | 01:03 |
| * devananda actually reads scrolback :) | 01:03 | |
| kfox1111 | just got to figure out how to get the neutron dhcp server onto the network. :) sharing eth0 here is going to make it a bit tricky. | 01:03 |
| kfox1111 | the real trick will be getting it all right, so that I don't drop ssh out from under me. :) | 01:04 |
| devananda | ah, heh, yep | 01:06 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Correct link in user guide https://review.openstack.org/136201 | 01:11 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Add new enrollment and troubleshooting doc sections https://review.openstack.org/136202 | 01:11 |
| devananda | NobodyCam: there's your enrollment example :) | 01:11 |
| devananda | kfox1111: you're probably past those by now, but ^ may still have some bits that will help you | 01:11 |
| * devananda steps away to get ready for dinner | 01:11 | |
| * NobodyCam reads back ... woohoo | 01:13 | |
| kfox1111 | nope.... failure. :) | 01:15 |
| *** Marga_ has joined #openstack-ironic | 01:18 | |
| *** chenglch has joined #openstack-ironic | 01:25 | |
| NobodyCam | kfox1111: :( | 01:33 |
| NobodyCam | devananda: wow I must be tired as I +2'd all three patches | 01:35 |
| NobodyCam | :-p | 01:35 |
| NobodyCam | 'night ya'll :) | 01:36 |
| *** kfox1111 has quit IRC | 01:41 | |
| Haomeng|2 | morning:) | 01:52 |
| jroll | heya Haomeng|2 | 01:53 |
| Haomeng|2 | jroll: :) | 01:54 |
| *** rloo has quit IRC | 01:58 | |
| *** dlaube has quit IRC | 02:04 | |
| *** ryanpetrello has joined #openstack-ironic | 02:04 | |
| *** chenglch|2 has joined #openstack-ironic | 02:09 | |
| *** chenglch has quit IRC | 02:09 | |
| *** penick has joined #openstack-ironic | 02:09 | |
| *** ChuckC has quit IRC | 02:10 | |
| *** penick_ has joined #openstack-ironic | 02:12 | |
| *** nosnos has joined #openstack-ironic | 02:12 | |
| *** penick has quit IRC | 02:14 | |
| *** penick_ is now known as penick | 02:14 | |
| *** Marga_ has quit IRC | 02:16 | |
| *** Marga_ has joined #openstack-ironic | 02:16 | |
| *** Marga__ has joined #openstack-ironic | 02:20 | |
| *** krtaylor has quit IRC | 02:21 | |
| *** Marga_ has quit IRC | 02:21 | |
| *** krtaylor has joined #openstack-ironic | 02:33 | |
| *** ramineni has joined #openstack-ironic | 02:53 | |
| *** ryanpetrello has quit IRC | 03:01 | |
| *** ryanpetrello has joined #openstack-ironic | 03:04 | |
| *** ujuc has joined #openstack-ironic | 03:05 | |
| openstackgerrit | Merged openstack/ironic: Add serial console feature to seamicro driver https://review.openstack.org/132628 | 03:08 |
| openstackgerrit | Merged openstack/python-ironicclient: Add maintenance_reason to node-show output https://review.openstack.org/135999 | 03:09 |
| *** pcrews has quit IRC | 03:13 | |
| *** ryanpetrello has quit IRC | 03:16 | |
| *** ryanpetrello has joined #openstack-ironic | 03:17 | |
| *** nosnos has quit IRC | 03:24 | |
| *** naohirot has quit IRC | 03:26 | |
| *** ryanpetrello has quit IRC | 03:31 | |
| *** Marga__ has quit IRC | 03:33 | |
| *** harlowja is now known as harlowja_away | 03:39 | |
| *** penick has quit IRC | 03:56 | |
| *** ChuckC has joined #openstack-ironic | 03:58 | |
| *** pensu has joined #openstack-ironic | 04:02 | |
| *** Marga_ has joined #openstack-ironic | 04:04 | |
| *** naohirot has joined #openstack-ironic | 04:07 | |
| *** Viswanath has joined #openstack-ironic | 04:09 | |
| *** Marga_ has quit IRC | 04:12 | |
| *** Viswanath has quit IRC | 04:13 | |
| *** spandhe has quit IRC | 04:17 | |
| *** vinbs has joined #openstack-ironic | 04:22 | |
| *** spandhe has joined #openstack-ironic | 04:23 | |
| *** nosnos has joined #openstack-ironic | 04:29 | |
| *** Marga_ has joined #openstack-ironic | 04:39 | |
| *** rushiagr_away is now known as rushiagr | 04:43 | |
| *** david-lyle_afk has quit IRC | 04:43 | |
| *** Marga_ has quit IRC | 04:44 | |
| *** ryanpetrello has joined #openstack-ironic | 04:54 | |
| *** spandhe has quit IRC | 04:57 | |
| openstackgerrit | Sirushti Murugesan proposed openstack/ironic-specs: Whole Disk Image Support for PXE Deploy Driver https://review.openstack.org/97150 | 05:00 |
| *** killer_prince is now known as lazy_prince | 05:01 | |
| vinbs | morning ironic! | 05:09 |
| *** nosnos has quit IRC | 05:14 | |
| *** nosnos has joined #openstack-ironic | 05:16 | |
| *** __mohit__ has joined #openstack-ironic | 05:21 | |
| *** __mohit__ has quit IRC | 05:22 | |
| *** Marga_ has joined #openstack-ironic | 05:40 | |
| *** Marga_ has quit IRC | 05:45 | |
| *** rakesh_hs has joined #openstack-ironic | 05:45 | |
| *** nosnos has quit IRC | 05:59 | |
| *** nosnos has joined #openstack-ironic | 06:00 | |
| *** mrda is now known as mrda_weekend | 06:01 | |
| *** Marga_ has joined #openstack-ironic | 06:02 | |
| *** jcoufal has joined #openstack-ironic | 06:03 | |
| *** vinbs has quit IRC | 06:07 | |
| *** Marga_ has quit IRC | 06:07 | |
| openstackgerrit | Merged openstack/ironic: Minor fix to install guide for associating k&r to nodes https://review.openstack.org/136187 | 06:11 |
| openstackgerrit | Merged openstack/ironic: Correct link in user guide https://review.openstack.org/136201 | 06:11 |
| *** vinbs has joined #openstack-ironic | 06:12 | |
| *** vinbs has quit IRC | 06:17 | |
| *** k4n0 has joined #openstack-ironic | 06:17 | |
| *** vinbs has joined #openstack-ironic | 06:22 | |
| Haomeng|2 | vinbs: morning:) | 06:23 |
| *** Marga_ has joined #openstack-ironic | 06:46 | |
| *** Marga_ has quit IRC | 06:51 | |
| *** lazy_prince has quit IRC | 06:59 | |
| *** ryanpetrello has quit IRC | 07:05 | |
| *** killer_prince has joined #openstack-ironic | 07:11 | |
| *** killer_prince is now known as lazy_prin | 07:12 | |
| *** lazy_prin is now known as lazy_prince | 07:13 | |
| *** lazy_prince has quit IRC | 07:34 | |
| *** lazy_prince has joined #openstack-ironic | 07:34 | |
| *** vinbs has quit IRC | 07:35 | |
| *** achanda has quit IRC | 07:38 | |
| *** bradjones has quit IRC | 07:40 | |
| *** bradjones has joined #openstack-ironic | 07:40 | |
| *** Marga_ has joined #openstack-ironic | 07:47 | |
| *** dlpartain has joined #openstack-ironic | 07:47 | |
| *** Marga_ has quit IRC | 07:51 | |
| *** pensu has quit IRC | 07:56 | |
| *** dlpartain1 has joined #openstack-ironic | 08:04 | |
| *** dlpartain has quit IRC | 08:06 | |
| naohirot | rushiagr: hello, I read your presentation slide http://www.slideshare.net/openstackindia/filesystem-as-a-service-in-openstack | 08:09 |
| naohirot | rushiagr: if you have time right now, can I ask some basic questions? | 08:09 |
| *** pensu has joined #openstack-ironic | 08:11 | |
| *** nosnos has quit IRC | 08:24 | |
| *** nosnos has joined #openstack-ironic | 08:24 | |
| *** jiangfei has quit IRC | 08:30 | |
| *** jiangfei has joined #openstack-ironic | 08:32 | |
| *** achanda has joined #openstack-ironic | 08:38 | |
| *** achanda_ has joined #openstack-ironic | 08:40 | |
| *** achanda has quit IRC | 08:40 | |
| *** achanda_ is now known as achanda | 08:41 | |
| *** Marga_ has joined #openstack-ironic | 08:47 | |
| *** nosnos has quit IRC | 08:51 | |
| *** nosnos has joined #openstack-ironic | 08:51 | |
| *** Marga_ has quit IRC | 08:52 | |
| *** romcheg has joined #openstack-ironic | 08:59 | |
| *** nosnos has quit IRC | 09:06 | |
| *** ujuc has quit IRC | 09:06 | |
| *** ifarkas has joined #openstack-ironic | 09:09 | |
| *** ndipanov_gone is now known as nidipanov | 09:12 | |
| *** lucasagomes has joined #openstack-ironic | 09:24 | |
| *** andreykurilin_ has joined #openstack-ironic | 09:30 | |
| *** achanda has quit IRC | 09:33 | |
| *** nosnos has joined #openstack-ironic | 09:34 | |
| openstackgerrit | Harshada Mangesh Kakad proposed openstack/ironic: Set info = _parse_driver_info(self.node) in setup rather than testcases https://review.openstack.org/136274 | 09:34 |
| *** derekh has joined #openstack-ironic | 09:35 | |
| ramineni | lucasagomes: morning :) | 09:35 |
| *** dtantsur|afk is now known as dtantsur | 09:37 | |
| dtantsur | Morning Ironic, TGIF! | 09:37 |
| ramineni | dtantsur: morning :) | 09:37 |
| *** viktors has joined #openstack-ironic | 09:43 | |
| *** Marga_ has joined #openstack-ironic | 09:48 | |
| naohirot | dtantsur: good morning! | 09:52 |
| dtantsur | naohirot, ramineni o/ | 09:53 |
| *** Marga_ has quit IRC | 09:53 | |
| *** k4n0 has quit IRC | 09:53 | |
| naohirot | dtantsur: can I ask basic question about swift? | 09:53 |
| dtantsur | naohirot, only if very basic, I know very little about swift :( | 09:54 |
| dtantsur | #openstack-swift should be much better place | 09:54 |
| naohirot | dtantsur: Swift export object though HTTP/HTTPS, right? | 09:55 |
| dtantsur | I suppose so | 09:55 |
| *** dlpartain1 has quit IRC | 09:55 | |
| naohirot | dtantsur: and iscsi_ilo mounts ISO from swift object storage. | 09:56 |
| naohirot | dtantsur: I suppose iLO accesses the ISO by HTTP/HTTPS. | 09:56 |
| naohirot | dtantsur: by mounting ISO as virtual media. | 09:57 |
| dtantsur | then it's not iscsi_ilo, iscsi deploy is not using virtual media | 09:57 |
| dtantsur | oh no, you're right | 09:57 |
| dtantsur | sorry, I confused it with pxe_ilo | 09:57 |
| naohirot | dtantsur: when is iscsi protocol used? that's my question. | 09:58 |
| dtantsur | naohirot, virtual media is used to put kernel & ramdisk on the machine. iscsi is used to put a large instance image on a target hard driver | 09:58 |
| dtantsur | naohirot, it's ramdisk responsibility to expose the hard driver via iscsi | 09:59 |
| *** chenglch|2 has quit IRC | 10:00 | |
| naohirot | dtantsur: Aha, ramdisk is iscsi target from conductor's point of view. | 10:00 |
| naohirot | dtantsur: If so, nfs_irmc and cifs_irmc should be also iscsi_irmc, what do you think? | 10:01 |
| *** Masahiroo has joined #openstack-ironic | 10:01 | |
| dtantsur | naohirot, I'm not sure how nfs_irmc and cifs_irmc are going to work :) | 10:02 |
| naohirot | dtantsur: I totally misunderstood iscsi_ilo. | 10:02 |
| naohirot | dtantsur: nfs or cifs is used to mount virtual media in case of irmc. | 10:03 |
| dtantsur | as you see, I also confuse our numerous drivers from time to time :) | 10:03 |
| naohirot | dtantsur: me too :-) | 10:03 |
| dtantsur | naohirot, we have 2 major means of putting an instance image on a node: iSCSI and ironic-python-agent (widely knows is IPA or just agent) | 10:03 |
| dtantsur | * known | 10:03 |
| dtantsur | naohirot, I suppose your drivers should work with one of these | 10:04 |
| naohirot | dtantsur: Okay | 10:04 |
| *** dlpartain has joined #openstack-ironic | 10:10 | |
| ramineni | lucasagomes: hi , need your opinion on seperating boot as seprate resource and keep device and mode as subresource to boot in REST API to expose boot_device and boot_mode respectively | 10:10 |
| ramineni | lucasagomes: currently , 3 resources i could think of correspond to boot. ( boot_device, boot_mode , secureboot (proposed as part of https://review.openstack.org/#/c/135845/1/specs/kilo/uefi-secure-boot-management-interfaces.rst ) . so, thinking exposing boot as seperate resource would be more cleaner way to do this ? what do you think, sounds good to you? | 10:11 |
| openstackgerrit | Imre Farkas proposed openstack/ironic-specs: New driver interface for RAID configuration https://review.openstack.org/135899 | 10:18 |
| *** andreykurilin_ has quit IRC | 10:20 | |
| *** Masahiroo has quit IRC | 10:21 | |
| *** andreykurilin_ has joined #openstack-ironic | 10:22 | |
| *** jcoufal has quit IRC | 10:24 | |
| *** jcoufal has joined #openstack-ironic | 10:25 | |
| *** blinky_ghost has joined #openstack-ironic | 10:26 | |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic-specs: Add get/set boot mode to Management Interface https://review.openstack.org/129529 | 10:31 |
| *** Haomeng has joined #openstack-ironic | 10:38 | |
| *** Haomeng|2 has quit IRC | 10:39 | |
| *** dtantsur is now known as dtantsur|brb | 10:41 | |
| *** igordcard has joined #openstack-ironic | 10:41 | |
| rushiagr | naohirot: hey hi | 10:42 |
| rushiagr | naohirot: I no longer work on it, so I might not be up-to-date on it | 10:42 |
| rushiagr | naohirot: the part I was working on, has been moved to a new openstack project named as Manila | 10:42 |
| rushiagr | naohirot: you can ask more information in it's own channel: #openstack-manila | 10:43 |
| *** Marga_ has joined #openstack-ironic | 10:49 | |
| *** igordcard has quit IRC | 10:50 | |
| *** igordcard has joined #openstack-ironic | 10:51 | |
| *** naohirot has quit IRC | 10:52 | |
| *** foexle has joined #openstack-ironic | 10:53 | |
| *** Marga_ has quit IRC | 10:54 | |
| *** igordcard has quit IRC | 10:55 | |
| *** ramineni has quit IRC | 10:59 | |
| *** pensu has quit IRC | 11:05 | |
| *** igordcard has joined #openstack-ironic | 11:08 | |
| *** Haomeng|2 has joined #openstack-ironic | 11:18 | |
| *** Haomeng has quit IRC | 11:18 | |
| *** pelix has joined #openstack-ironic | 11:19 | |
| *** pensu has joined #openstack-ironic | 11:32 | |
| openstackgerrit | Harshada Mangesh Kakad proposed openstack/ironic: Avoid calling _parse_driver_info in every test https://review.openstack.org/136274 | 11:32 |
| *** Marga_ has joined #openstack-ironic | 11:50 | |
| *** Marga_ has quit IRC | 11:55 | |
| blinky_ghost | Hi all, I'm trying to use autodiscovery in tuskar-ui to provision my node using ironic. The node gets ramdisk discovery but kernel panics: http://picpaste.com/Screenshot_from_2014-11-21_11_53_25-avYqqHX0.png any hint what could be wrong? thanks | 11:55 |
| *** dlpartain has quit IRC | 11:57 | |
| *** jcoufal has quit IRC | 11:58 | |
| *** naohirot has joined #openstack-ironic | 12:00 | |
| lucasagomes | blinky_ghost, it looks like a diff error than u had yesterday, yesterdat the ramdisk booted right? | 12:00 |
| lucasagomes | blinky_ghost, kernel panic can be a lot of different things, missing base drivers, wrong combination of kernel and ramdisk | 12:01 |
| lucasagomes | etc... | 12:01 |
| lucasagomes | what did it change from yesterday to today that it's not booting anymore? | 12:01 |
| blinky_ghost | lucasgomes: yes, different error, I was testing autodiscovery. | 12:01 |
| lucasagomes | blinky_ghost, have u manually updated the ramdisk? or u built it again with DIB? | 12:02 |
| blinky_ghost | lucasgomes now I'm testing without autodiscovery, I build the dib and didn't work so I updated it manually. testing | 12:03 |
| *** dlpartain has joined #openstack-ironic | 12:03 | |
| lucasagomes | right u could have messed permissions when compacting it again and it caused the ramdisk to fail to boot | 12:03 |
| lucasagomes | are u using root to extract and compact the ramdisk before modifying it? | 12:04 |
| lucasagomes | " The node gets ramdisk discovery but kernel panics" So it's the deploy ramdisk!? | 12:04 |
| lucasagomes | s/using/logged as/ | 12:04 |
| blinky_ghost | lucasgomes yes I used root to compact it | 12:06 |
| lucasagomes | right, does the deploy ramdisk works? are they built using the same image? | 12:07 |
| blinky_ghost | lucasgomes I think you're right, I created the ramdisk manually and somehow messed it because I get the error with or without autodiscovery | 12:08 |
| blinky_ghost | lucasgomes I'll try to create a new ramdisk | 12:09 |
| *** alexpilotti has joined #openstack-ironic | 12:13 | |
| *** romcheg has quit IRC | 12:14 | |
| lucasagomes | right | 12:19 |
| lucasagomes | yeah, it's not hard to mess things up doing that :) | 12:19 |
| naohirot | lucasagomes: hi | 12:22 |
| naohirot | blinky_ghost: hi | 12:22 |
| blinky_ghost | hi | 12:22 |
| naohirot | blinky_ghost: may I talk to lucasagomes , are you still discussing? | 12:23 |
| blinky_ghost | naohirot: sure :) | 12:24 |
| lucasagomes | naohirot, hi | 12:24 |
| naohirot | blinky_ghost: thanks! | 12:24 |
| naohirot | lucasagomes: I'd like to hear your opinion how to implement nfs_irmc and cifs_irmc. | 12:24 |
| naohirot | lucasagomes: I totally have been misunderstood iscsi_ilo. | 12:25 |
| lucasagomes | naohirot, right, is it one of the specs proposed? | 12:25 |
| naohirot | lucasagomes: I though that iscsi is used to mount virtual media. | 12:25 |
| naohirot | lucasagomes: yes, it is one on spec, irmc virtual media deploy driver. | 12:26 |
| lucasagomes | naohirot, oh... right... not really that's how the ramdisk exposes the disk to the conductor, not really to do with mounting a iscsi volume using the BMC or anything like that | 12:26 |
| naohirot | lucasagomes: Yes, but I thought that iLO uses iscsi to mount virtual media :-) | 12:27 |
| *** romcheg has joined #openstack-ironic | 12:27 | |
| lucasagomes | i c :) | 12:27 |
| naohirot | lucasagomes: that's the reason I named nfs_irmc and cifs_irmc, because iRMC use NFS/CIFS to mount virtual media. | 12:28 |
| naohirot | lucasagomes: and by reading iLO4 manual, iLO seems to use HTTP/HTTPS to mount virtual media. | 12:29 |
| naohirot | lucasagomes: that's the reason iscsi_ilo uses swift object storage. | 12:29 |
| lucasagomes | naohirot, right, AFAIUI. they put it on swift and get a temp url from there | 12:30 |
| naohirot | lucasagomes: what I'd like to hear your opinion is that whether I should use cinder instead of swift or not. | 12:30 |
| lucasagomes | naohirot, not 100% sure, you may want to talk to ramesh about it, he was directly involved into writting the driver | 12:30 |
| lucasagomes | but I believe that | 12:31 |
| lucasagomes | by using swift and temp urls u avoid the need of having a authentication key to fetch the image | 12:31 |
| lucasagomes | just like it would be required if they used only glance for example | 12:31 |
| lucasagomes | naohirot, for what u want to do, attaching external block devices | 12:31 |
| lucasagomes | I believe cinder is the right place | 12:32 |
| naohirot | lucasagomes: but iRMC cannot mount virtual media by HTTP/HTTPS. | 12:32 |
| lucasagomes | and time to time we see folks wanting the cinder integration | 12:32 |
| lucasagomes | naohirot, also, jroll was looking at having diskless nodes in Ironic | 12:33 |
| lucasagomes | idk if he's planning to use cinder or what, but you guys may want to talk | 12:33 |
| naohirot | lucasagomes: Okay I'll talk to jroll, I understood that cinder is right place for iRMC. | 12:34 |
| lucasagomes | yeah it seems to be | 12:35 |
| lucasagomes | I think there's even a old blueprint (not spec) about cinder boot volumes | 12:35 |
| * lucasagomes checks | 12:35 | |
| lucasagomes | https://blueprints.launchpad.net/ironic/+spec/cinder-integration | 12:35 |
| lucasagomes | well not many info there | 12:35 |
| naohirot | lucasagomes: regarding iscsi, is the conductor initiator of iscsi? | 12:35 |
| naohirot | lucasagomes: or is nova initiator of iscsi to install the User OS? | 12:37 |
| lucasagomes | naohirot, the ramdisk will create the target and pass it to the ironic conductor | 12:38 |
| lucasagomes | the conductor will them mount it format and copy the iamge | 12:38 |
| lucasagomes | image* | 12:38 |
| lucasagomes | the deploy ramdisk* | 12:38 |
| naohirot | lucasagomes: Okay, ironic conductor is initiator of iscsi. | 12:38 |
| *** pensu has quit IRC | 12:39 | |
| lucasagomes | yup | 12:39 |
| naohirot | lucasagomes: If so, is all driver, pxe_*, iscsi_ilo, agent_*, use iscsi to install user os, right? | 12:39 |
| * lucasagomes rechecks | 12:40 | |
| lucasagomes | naohirot, no the agent_ does it differently | 12:40 |
| lucasagomes | they fetch the image directly from glance (I believe) convert and write it locally | 12:40 |
| lucasagomes | it's something that the normal ramdisk could also do, but right now it doesn't | 12:40 |
| naohirot | lucasagomes: do you mean agent_* by "they"? | 12:41 |
| lucasagomes | yeah... I mean drivers using the Agent deploy ramdisk | 12:42 |
| lucasagomes | fetches the image directly and write it to the disk locally | 12:42 |
| lucasagomes | without exposing the local disk via iscsi, passing to the conductor to wrote it etc | 12:42 |
| lucasagomes | (maybe it sounded confusing... ) | 12:43 |
| lucasagomes | I mean the Agent ramdisk does fetch the image and write it locally | 12:43 |
| lucasagomes | on the machine it's booted on | 12:43 |
| naohirot | lucasagomes: don't you know what kind of protocol is that? | 12:43 |
| naohirot | lucasagomes: user os is large size | 12:44 |
| *** andreykurilin_ has quit IRC | 12:44 | |
| lucasagomes | naohirot, I believe they use HTTP to fetch it | 12:44 |
| * lucasagomes gotta dig into the code | 12:45 | |
| openstackgerrit | Harshada Mangesh Kakad proposed openstack/ironic: Add documentation for SeaMicro driver https://review.openstack.org/136324 | 12:45 |
| naohirot | lucasagomes: I see. | 12:45 |
| *** Haomeng has joined #openstack-ironic | 12:46 | |
| *** Haomeng|2 has quit IRC | 12:46 | |
| naohirot | lucasagomes: I case of agent_*, deply ramdisk becomes client to conductor or glance. | 12:46 |
| *** rakesh_hs has quit IRC | 12:47 | |
| *** Marga_ has joined #openstack-ironic | 12:47 | |
| *** rakesh_hs has joined #openstack-ironic | 12:48 | |
| *** dprince has joined #openstack-ironic | 12:50 | |
| naohirot | rushiagr: thanks for replying me. I'm new to openstack. I just wanted to make sure the difference b/w swift and cinder | 12:56 |
| *** sambetts has joined #openstack-ironic | 12:58 | |
| Shrews | morning ironicers | 13:03 |
| lucasagomes | Shrews, morning | 13:06 |
| *** athomas has joined #openstack-ironic | 13:06 | |
| *** lucasagomes is now known as lucas-hungry | 13:15 | |
| *** nosnos has quit IRC | 13:16 | |
| *** Marga__ has joined #openstack-ironic | 13:18 | |
| *** Marga_ has quit IRC | 13:21 | |
| *** dtantsur|brb is now known as dtantsur | 13:27 | |
| dtantsur | Shrews, morning | 13:28 |
| *** linggao has joined #openstack-ironic | 13:37 | |
| *** linggao_ has joined #openstack-ironic | 13:38 | |
| *** rushiagr is now known as rushiagr_away | 13:38 | |
| sambetts | Afternoon ironic | 13:41 |
| *** lazy_prince is now known as killer_prince | 13:48 | |
| *** dlpartain has left #openstack-ironic | 13:49 | |
| openstackgerrit | Naohiro Tamura proposed openstack/ironic-specs: iRMC Power Driver for Ironic https://review.openstack.org/134487 | 13:53 |
| *** Marga__ has quit IRC | 13:54 | |
| *** Marga_ has joined #openstack-ironic | 13:55 | |
| *** mjturek has joined #openstack-ironic | 13:59 | |
| *** jistr has joined #openstack-ironic | 14:03 | |
| *** ryanpetrello has joined #openstack-ironic | 14:03 | |
| *** jistr is now known as jistr|mtg | 14:04 | |
| *** ryanpetrello_ has joined #openstack-ironic | 14:05 | |
| *** ryanpetrello has quit IRC | 14:08 | |
| *** ryanpetrello_ is now known as ryanpetrello | 14:08 | |
| *** dlpartain has joined #openstack-ironic | 14:09 | |
| *** lucas-hungry is now known as lucasagomes | 14:13 | |
| *** rainya has quit IRC | 14:15 | |
| *** rainya has joined #openstack-ironic | 14:16 | |
| *** cohn has quit IRC | 14:16 | |
| *** russell_h has quit IRC | 14:16 | |
| *** cohn has joined #openstack-ironic | 14:16 | |
| *** cohn is now known as Guest82711 | 14:16 | |
| *** russell_h has joined #openstack-ironic | 14:17 | |
| *** rakesh_hs has quit IRC | 14:17 | |
| *** rushiagr_away is now known as rushiagr | 14:27 | |
| *** dlpartain has left #openstack-ironic | 14:27 | |
| *** ronald has quit IRC | 14:29 | |
| NobodyCam | Good Morning Ironic .... TGIF! | 14:31 |
| dtantsur | NobodyCam, oh really TGIF :) | 14:34 |
| NobodyCam | morning dtantsur :) ya | 14:34 |
| dtantsur | sambetts, NobodyCam, morning/afternoon | 14:34 |
| NobodyCam | dtantsur: are you planing on posting a status report today ? | 14:39 |
| dtantsur | NobodyCam, you mean bugs? we agreed on creating a report on Mon, but as I may be absent on Mon, I'll probably create it now | 14:40 |
| NobodyCam | :) you are correct... /me goes back to sipping coffee | 14:41 |
| NobodyCam | :-p | 14:41 |
| Shrews | NobodyCam: no sip. gulp | 14:41 |
| dtantsur | I'll post it today and update on Mon, if I'm available | 14:42 |
| NobodyCam | morning Shrews :) | 14:51 |
| NobodyCam | thank you dtantsur | 14:51 |
| NobodyCam | Shrews: if I gulp I'll be out of coffee sooner | 14:51 |
| Shrews | NobodyCam: well we can't have that. best just sip then :) | 14:52 |
| *** Guest82711 is now known as cohn | 14:53 | |
| *** cohn has quit IRC | 14:53 | |
| *** cohn has joined #openstack-ironic | 14:53 | |
| *** r-daneel has joined #openstack-ironic | 14:54 | |
| NobodyCam | lol | 14:56 |
| NobodyCam | I love some of these names. | 15:02 |
| NobodyCam | I want to combine them like: Pixie Boots | 15:03 |
| dtantsur | lol | 15:03 |
| NobodyCam | lucasagomes: is that allowed ^^^ | 15:03 |
| dtantsur | that could be a winner actually :D | 15:04 |
| lucasagomes | hah | 15:04 |
| lucasagomes | well yeah why not | 15:04 |
| lucasagomes | add to the list :D | 15:04 |
| * NobodyCam replys | 15:04 | |
| lucasagomes | \o/ | 15:05 |
| NobodyCam | :) lucasagomes I also added a section to the meeting agenda, please feel free to remove if its not needed | 15:10 |
| lucasagomes | NobodyCam, right | 15:10 |
| lucasagomes | the only thing is I might not be present on next meeting ;( | 15:10 |
| lucasagomes | I'll be going to paris (again) on monday >.< | 15:11 |
| NobodyCam | oh nice :) | 15:11 |
| NobodyCam | dang that city is expensive ... wow | 15:11 |
| lucasagomes | yeah | 15:11 |
| *** Marga_ has quit IRC | 15:21 | |
| blinky_ghost | lucasgomes: after recreating the image it works :) | 15:23 |
| NobodyCam | blinky_ghost: great to hear! | 15:25 |
| *** Marga_ has joined #openstack-ironic | 15:25 | |
| blinky_ghost | a big thank you to everyone been 2 weeks struggling with this :) | 15:26 |
| lucasagomes | blinky_ghost, :) yvw | 15:28 |
| dtantsur | ifarkas, lucasagomes, I have something for you to review ;) https://review.openstack.org/#/c/136291/ | 15:30 |
| lucasagomes | o/ | 15:31 |
| dtantsur | ... and anyone suddenly feeling like reviewing ironic-discoverd patches too :D | 15:32 |
| *** zz_jgrimm is now known as jgrimm | 15:34 | |
| *** andreykurilin_ has joined #openstack-ironic | 15:36 | |
| *** Marga_ has quit IRC | 15:37 | |
| foexle | hey guys, short question: i get every time node is locked by xxx if i try to boot a new node. compute => Error contacting Ironic server for 'node.update' ####### api => "PATCH /v1/nodes/f1ddbd1a-4758-4ba5-bce6-822435e3da0b HTTP/1.1" 409 node is locked -.- anyone a short hint ? | 15:37 |
| lucasagomes | *sights* | 15:39 |
| NobodyCam | foexle: what state is the node in. | 15:39 |
| lucasagomes | foexle, yeah that node may have a lock stuck :/ | 15:39 |
| foexle | | f1ddbd1a-4758-4ba5-bce6-822435e3da0b | None | power off | None | False | | 15:40 |
| NobodyCam | I've two cases where node did not remove the instance-uuid | 15:40 |
| foexle | lucasagomes: yeah .... i had delete (in db) and recreate again .... but i doesn't work .... | 15:40 |
| lucasagomes | foexle, delete the node? or the lock? | 15:40 |
| *** Haomeng has quit IRC | 15:41 | |
| foexle | after a while if 60 attemps reached by ironic conductor i get the node (nova list) in ERROR state | 15:41 |
| foexle | and i can delete | 15:41 |
| *** naohirot has quit IRC | 15:43 | |
| foexle | lucasagomes: i've read thats a bug ... so not a bug but it's not possible to solve a deadlock right ? | 15:43 |
| foexle | i need to delete the whole node in my ironic db and recreate it | 15:43 |
| foexle | but how does this deadlocks occurs ? | 15:44 |
| lucasagomes | foexle, u can delete the lock in the db only | 15:44 |
| lucasagomes | don't need to delete the whole node | 15:44 |
| foexle | lucasagomes: how ? | 15:44 |
| lucasagomes | at the node tables, there's a (forgot the name of the field) | 15:44 |
| lucasagomes | lemme check | 15:44 |
| foexle | reservation ? | 15:44 |
| lucasagomes | reservation column | 15:45 |
| lucasagomes | exactly | 15:45 |
| foexle | i need to clean ? so update nodes set reservation=''; ? | 15:46 |
| lucasagomes | foexle, so it occurs if a conductor that is managing that node dies mid-operation | 15:46 |
| lucasagomes | foexle, it won't work... we talked in Paris about exposing it in the API | 15:46 |
| foexle | hmmm dies? ... ok i'll test :) | 15:46 |
| lucasagomes | I mean exposing a way in the API to break the lock | 15:46 |
| foexle | lucasagomes: yeah i know; i saw it in launchpad | 15:46 |
| foexle | like the state changer in nova | 15:47 |
| lucasagomes | foexle, people also wants to make it pluggable to be able to use zookeeper for e.g | 15:47 |
| lucasagomes | yup | 15:47 |
| foexle | so i need only to clean this column and try again ?! | 15:47 |
| foexle | let me check :) | 15:47 |
| lucasagomes | but right now, yeah I know it kinda sucks to not have a way to break it from the api | 15:47 |
| lucasagomes | foexle, yeah | 15:47 |
| *** Haomeng has joined #openstack-ironic | 15:50 | |
| foexle | lucasagomes: nope, doesn't has an effect :( to clean the column and try again isn't working | 15:51 |
| foexle | same error ... is there other relationship maybe ? ... i found nothing i've checked all tables | 15:51 |
| *** nidipanov has quit IRC | 15:51 | |
| lucasagomes | hmm checks | 15:52 |
| lucasagomes | wondering if it changed after the conductor_affinity was introduced | 15:52 |
| lucasagomes | foexle, will check the code 1 sec | 15:52 |
| lucasagomes | foexle, ur getting a NodeLocked exception right? | 15:53 |
| *** pcrews has joined #openstack-ironic | 15:54 | |
| *** anderbubble has joined #openstack-ironic | 15:54 | |
| devananda | mornin, all | 15:55 |
| * anderbubble waves | 15:55 | |
| *** achanda has joined #openstack-ironic | 15:56 | |
| lucasagomes | devananda, morning | 15:56 |
| NobodyCam | mornign devananda | 15:57 |
| foexle | lucasagomes: yep :) | 15:58 |
| sambetts | o/ devananda | 15:59 |
| lucasagomes | foexle, odd, looks setting reservation to NULL should do the work | 16:00 |
| *** achanda has quit IRC | 16:02 | |
| foexle | lucasagomes: nope -.- .... it's ok i'll delete the node completely | 16:03 |
| lucasagomes | devananda, when u get some time, wanna talk about the ENROLL and INIT states of the new state machine? | 16:03 |
| foexle | and recreate again | 16:03 |
| *** andreykurilin_ has quit IRC | 16:03 | |
| *** andreykurilin__ has joined #openstack-ironic | 16:03 | |
| lucasagomes | foexle, :( | 16:03 |
| * lucasagomes tries to figure out what's going on | 16:03 | |
| devananda | lucasagomes: I have ~15 minutes, then need to run -- have to go to the DoT office and get a replacement title for my old car | 16:04 |
| devananda | that should be .. .fun ... | 16:04 |
| lucasagomes | devananda, I see... have u checked the spec recently? | 16:04 |
| lucasagomes | I added some more thoughts there | 16:05 |
| foexle | lucasagomes: maybe i'm wrong ... so i try my first ironic setup from scratch (juno) ... but i runs everytime in lock errors :D ... so maybe anything wrong. Save your time | 16:05 |
| foexle | lucasagomes: but thanks for your help ! | 16:05 |
| devananda | lucasagomes: ack. will look now | 16:05 |
| lucasagomes | foexle, ack, if u get really stuck lemme know | 16:05 |
| foexle | all right thanks | 16:06 |
| *** anderbubble has quit IRC | 16:07 | |
| *** Viswanath has joined #openstack-ironic | 16:11 | |
| *** Viswanath has quit IRC | 16:14 | |
| devananda | lucasagomes: "both ENROLL and INIT means the same, it means that the node is registered in Ironic but may not contain all the information to be managed by it yet" | 16:19 |
| devananda | lucasagomes: that's where I'm thinking of it differently | 16:19 |
| devananda | lucasagomes: to me, ENROLL means "may or may not have all the information necessary to manage it" | 16:20 |
| devananda | lucasagomes: whereas INIT means "ironic has confirmed it has all the information it needs, node is -initialized-" | 16:20 |
| devananda | example | 16:20 |
| *** Marga_ has joined #openstack-ironic | 16:21 | |
| devananda | I create a node and don't pass in ipmi creds. node is in ENROLL state, and driver validation fails | 16:21 |
| devananda | i update node to add ipmi creds, but no properties | 16:21 |
| devananda | node is in ENROLL state still, but driver validation succeeds | 16:21 |
| lucasagomes | a-ha, right | 16:21 |
| devananda | I now have a choice | 16:21 |
| lucasagomes | I see I understand it better with the validation checks | 16:21 |
| devananda | 1. PUT DISCOVERING /provision_state | 16:21 |
| devananda | and ironic tries to find the properties | 16:22 |
| devananda | or 2. PUT {some data} /node/properties | 16:22 |
| devananda | at the end of either of those, ironic checks again -- does it have enough info? if so, automatically transition to INIT | 16:22 |
| devananda | because now the node is "initialized" | 16:22 |
| devananda | still requiers an operator to move it from INIT to AVAILABLE | 16:23 |
| lucasagomes | one thing, can a node in INIT triggers discover again? Or should operator but it back to ENROLL and then trigger discover | 16:23 |
| lucasagomes | I see | 16:23 |
| devananda | INIT can go back to DISCOVERY by manual action (not automatic) | 16:23 |
| devananda | oooh | 16:23 |
| devananda | what if all the *FAIL states dump a node to INIT | 16:23 |
| devananda | i mean, try to transition it *to* INIT | 16:24 |
| devananda | so it re-runs the same validation that taht step does? -- just thinking outloud here | 16:24 |
| lucasagomes | heh it shouldn't | 16:24 |
| devananda | but anyway -- do you see the difference that I'm thinking of between ENROLL and INIT? is that useful? | 16:24 |
| lucasagomes | yeah | 16:24 |
| lucasagomes | I see what it means now | 16:24 |
| lucasagomes | I will think over, I'm thinking about jumping with both feet on that state machine | 16:25 |
| lucasagomes | it's kinda strategic for us, so we need to get it done asap | 16:25 |
| devananda | yup | 16:26 |
| devananda | lots of things depend on this | 16:26 |
| devananda | I thinkw e should prioritize getting this done by k-1 | 16:26 |
| lucasagomes | +1 | 16:26 |
| devananda | i'll try to spend some time over the holidays working on it, assuming it's going to take at least a few weeks to get ready to land it | 16:26 |
| devananda | ooh, also, what I said above is not what victor_lowther had in mind | 16:27 |
| devananda | "INIT: Ironic has verified that it can manage the node, but the node may require further configuration before it can be made available." | 16:27 |
| devananda | his distinction may be even more useful there | 16:27 |
| devananda | ENROLL: there is a node, but Ironic is not able to even contact it | 16:27 |
| * lucasagomes doesn't remember that part | 16:27 | |
| lucasagomes | oh he just commented on it | 16:28 |
| * lucasagomes refreshs | 16:28 | |
| devananda | INIT: Ironic has confirmed it can control the node | 16:28 |
| lucasagomes | yeah, I see the distinction more clear now | 16:28 |
| devananda | AVAILABLE: Ironic has enough information about the node to make it available AND the operator has requested that it is made available | 16:28 |
| lucasagomes | the validation to check whether it's manageble or not makes sense | 16:28 |
| lucasagomes | sounds good | 16:29 |
| devananda | ok - I gotta run .. .hope that's helpful | 16:29 |
| devananda | will bbl | 16:29 |
| lucasagomes | devananda, good luck there | 16:29 |
| lucasagomes | I will read his comments and think more about it | 16:29 |
| *** Marga_ has quit IRC | 16:32 | |
| *** foexle has quit IRC | 16:33 | |
| victor_lowther | lucasagomes: please do. My irc access at work is spotty, so I would prefer that discussion on the spec happen in the comments. | 16:33 |
| lucasagomes | victor_lowther, sure, +1 for having discussions on the spec | 16:34 |
| lucasagomes | I just needed a high-throughput here to try to understand the differences better | 16:35 |
| *** Marga_ has joined #openstack-ironic | 16:36 | |
| victor_lowther | Yeah, if I did not have to do irc over my phone at work for misguided security reasons I would pay more attention to it. | 16:38 |
| lucasagomes | fair enuff | 16:39 |
| jroll | morning everybody :) | 16:49 |
| jroll | victor_lowther: :((( | 16:50 |
| victor_lowther | Although irccloud let's me know when my name is said... | 16:51 |
| lucasagomes | jroll, moring | 16:59 |
| lucasagomes | morning* | 16:59 |
| NobodyCam | mornign jroll :) | 17:01 |
| *** Marga_ has quit IRC | 17:02 | |
| *** Marga_ has joined #openstack-ironic | 17:03 | |
| dtantsur | morning jroll :) | 17:08 |
| dtantsur | and g'night to everyone else! have a nice weekend | 17:09 |
| *** dtantsur is now known as dtantsur|afk | 17:10 | |
| *** Marga_ has quit IRC | 17:10 | |
| *** Marga_ has joined #openstack-ironic | 17:11 | |
| NobodyCam | night dtantsur|afk :) | 17:15 |
| NobodyCam | have a good weekend your self | 17:15 |
| *** anderbubble has joined #openstack-ironic | 17:20 | |
| *** dprince has quit IRC | 17:26 | |
| *** dprince has joined #openstack-ironic | 17:26 | |
| *** Marga_ has quit IRC | 17:30 | |
| *** Marga_ has joined #openstack-ironic | 17:31 | |
| *** Marga_ has quit IRC | 17:31 | |
| *** Marga_ has joined #openstack-ironic | 17:32 | |
| *** Marga_ has quit IRC | 17:32 | |
| *** Marga_ has joined #openstack-ironic | 17:33 | |
| *** Marga_ has quit IRC | 17:33 | |
| *** Marga_ has joined #openstack-ironic | 17:34 | |
| *** Marga_ has quit IRC | 17:35 | |
| *** Marga_ has joined #openstack-ironic | 17:36 | |
| *** Marga_ has quit IRC | 17:36 | |
| *** Marga_ has joined #openstack-ironic | 17:37 | |
| *** kfox1111 has joined #openstack-ironic | 17:43 | |
| kfox1111 | Almost got ironic working. pxe booting is working up to the deploy image, then deploying is failing. | 17:44 |
| kfox1111 | http://pastebin.com/851N1jEM | 17:44 |
| kfox1111 | Any idea what might cause that? | 17:44 |
| jroll | Nov 21 09:38:07 ironic.osg.pnl.gov ironic-conductor[27312]: Stderr: 'Error: Partition(s) 1, 2, 3 on /dev/sdc have been written, but we have been unable to inform the kernel of the change, probably because it/they are in use. As a result, the old partition(s) will remain in use. You should reboot now before making further changes.\n' | 17:45 |
| jroll | that's weird | 17:45 |
| *** Marga_ has quit IRC | 17:46 | |
| kfox1111 | yeah. | 17:47 |
| kfox1111 | I followed the install docs and used the ubuntu image instructions. | 17:48 |
| kfox1111 | Note, the image on disk from before did have md arrays on it. Maybe that had an effect? | 17:48 |
| jroll | oh, maybe? | 17:49 |
| jroll | we do clear the very beginning and end | 17:49 |
| jroll | seems like we could let that error through, though | 17:49 |
| kfox1111 | so if you clear it, maybe redeploying it a second time will work? | 17:49 |
| jroll | maybe? | 17:50 |
| jroll | worth a shot | 17:50 |
| kfox1111 | let me try it. if so, I'll submit a bug. | 17:50 |
| jroll | cool | 17:50 |
| kfox1111 | Is there any way to get the deploy image to use the ipmi console? its very... quiet during that part. | 17:50 |
| jroll | I have no idea, I don't really use the pxe driver | 17:51 |
| kfox1111 | hmm.. now Stderr: 'Error: The location 1024001 is outside of the device /dev/sdd.\n' | 17:51 |
| *** harlowja_away is now known as harlowja | 17:52 | |
| *** sambetts has quit IRC | 17:52 | |
| kfox1111 | what does -i pxe_root_gb=<root partition size> \ mean? | 17:56 |
| jroll | in the client? | 17:57 |
| kfox1111 | in the flavor | 17:57 |
| *** derekh has quit IRC | 17:57 | |
| jroll | oh, that's the size for the root partition | 17:57 |
| kfox1111 | sorry. in the node-create | 17:57 |
| jroll | there's also (optionally) ephemeral/swap | 17:57 |
| jroll | um, that shouldn't be in the node-create call, nova sends that | 17:58 |
| kfox1111 | there's pxe_root_gb and there is -p local_gb=$DISK_GB | 17:58 |
| kfox1111 | http://docs.openstack.org/developer/ironic/deploy/install-guide.html#configure-identity-service-for-bare-metal | 17:58 |
| JayF | kfox1111: absolutely nuke the whole disk if it had a raid on it | 17:58 |
| kfox1111 | the local_gb makes sense. there is no explination on what pxe_root_gb means. | 17:58 |
| JayF | kfox1111: this is one of the reasons I've been advocating for decom/zapping as a default; you can have weird things happen with raid superblocks left over :) | 17:58 |
| jroll | I don't see pxe_root_gb there? | 17:59 |
| kfox1111 | JayF. a couple of tries should have cleared it by now? | 17:59 |
| kfox1111 | jroll: search in the document for "ironic node-create -d pxe_ipmitool" | 17:59 |
| JayF | kfox1111: yeah, it's likely other things were/are wrong too; but generally Ironic doesn't clear your whole disk, only beginning/end then writes an image out ... | 18:00 |
| jroll | I don't even see node-create on that page :| | 18:00 |
| kfox1111 | Section "Flavor Creation" | 18:00 |
| kfox1111 | step 3. | 18:00 |
| kfox1111 | Under Juno | 18:01 |
| jroll | I see flavor-create | 18:01 |
| kfox1111 | second box. | 18:01 |
| *** imtiaz has joined #openstack-ironic | 18:01 | |
| jroll | and node-update | 18:01 |
| jroll | but I don't see pxe_root_gb | 18:01 |
| jroll | ironic node-update $NODE_UUID add \ | 18:01 |
| jroll | driver_info/pxe_deploy_kernel=$DEPLOY_VMLINUZ_UUID \ | 18:01 |
| jroll | driver_info/pxe_deploy_ramdisk=$DEPLOY_INITRD_UUID \ | 18:01 |
| kfox1111 | what?!? | 18:02 |
| kfox1111 | heh. I just refreshed the page, and its different now. :/ | 18:02 |
| kfox1111 | yeah. no node-create anywhere. | 18:02 |
| jroll | lol :( | 18:02 |
| kfox1111 | not a useful update to the doc. :/ | 18:02 |
| kfox1111 | so whats the proper way to node-create? | 18:03 |
| jroll | odd | 18:03 |
| kfox1111 | the old doc said: http://pastebin.com/pegSdXJP | 18:04 |
| jroll | devananda: uh | 18:04 |
| jroll | devananda: you removed the only node-create from our deployer docs :P | 18:04 |
| jroll | https://review.openstack.org/#/c/136187/1/doc/source/deploy/install-guide.rst,cm | 18:04 |
| *** ChuckC has quit IRC | 18:04 | |
| kfox1111 | its also missing the port create. though that was never there. | 18:05 |
| *** romcheg has quit IRC | 18:05 | |
| *** romcheg has joined #openstack-ironic | 18:05 | |
| jroll | kfox1111: yeah, that's correct, just skip the pxe_root_gb bit | 18:05 |
| kfox1111 | it should do the right thing automatically? | 18:05 |
| jroll | yes, nova sets that | 18:05 |
| kfox1111 | k. | 18:05 |
| *** romcheg has quit IRC | 18:06 | |
| mjturek | not sure this is the best place to ask but I'm getting the following error when running diskimage-builder I'm getting "dib-run-parts: command not found". Anyone seen this before? Already ran the dependency install script from the triple-o guys | 18:06 |
| jroll | mjturek: I'd ask in #tripleo | 18:07 |
| mjturek | jroll thanks will do! | 18:07 |
| jroll | :) | 18:07 |
| kfox1111 | could it also be that my disk sizes are wrong? I have 1 TB disks, but set flavor and node to have 1000GB disk sizes. The partitioning complaint was that it couldnt do 1024001 which doesn't count the way hd manufacturers count. :/ | 18:09 |
| jroll | hmm, maybe | 18:10 |
| jroll | that's strange | 18:10 |
| jroll | well, what's your flavor look like | 18:10 |
| jroll | in terms of root/ephemeral/swap | 18:10 |
| kfox1111 | cbf47302-9ac3-4f03-99b2-a04c6f6320eb | my-baremetal-flavor | 65536 | 1000 | 0 | | 32 | 1.0 | True | 18:10 |
| jroll | what's that 32? | 18:10 |
| kfox1111 | vcpu's | 18:10 |
| jroll | oh, so no swap or ephemeral? | 18:11 |
| kfox1111 | correct. the doc didn't say to create any | 18:11 |
| jroll | right | 18:11 |
| kfox1111 | should I just drop both of them to 900GB and see if it works better? | 18:12 |
| jroll | I have no idea, then | 18:12 |
| jroll | yeah, that will probably work | 18:12 |
| NobodyCam | gah corp exp report filed... brain only slightly damaged as a result | 18:12 |
| JayF | NobodyCam: I did mine first day back | 18:12 |
| jroll | lol | 18:12 |
| JayF | NobodyCam: made that a habit early on because otherwise I'd end up with a drawer of reciepts that had faded 3 months later | 18:13 |
| JayF | lol | 18:13 |
| NobodyCam | I wait for all expences to show up. else I end up filing like two or three reports for the same trip | 18:13 |
| JayF | Aha, corporate card or such? I just keep reciepts | 18:14 |
| NobodyCam | yep corp card :-p | 18:15 |
| *** achanda has joined #openstack-ironic | 18:18 | |
| kfox1111 | hmm.. yeah. dropping it to 900GB seems to be working better. | 18:18 |
| jroll | neat | 18:18 |
| jroll | that should still be filed as a bug | 18:18 |
| kfox1111 | stupid hard drive vendors and their winning supreme court case. :/ | 18:19 |
| JayF | and probably clarified in the docs | 18:19 |
| jroll | though I can't math right now, so | 18:19 |
| jroll | it might be correct | 18:19 |
| JayF | wanna become an openstack ATC? fix the doc? :) | 18:19 |
| jroll | kfox1111: oh, is the drive actually 1,000,000,000,000 bytes? | 18:19 |
| *** achanda has quit IRC | 18:19 | |
| jroll | or whatever? | 18:19 |
| *** achanda has joined #openstack-ironic | 18:19 | |
| kfox1111 | jroll: yes. | 18:19 |
| JayF | iirc there was a lot of discussion about GiB vs GB when that code waas refactored | 18:19 |
| * jroll cries | 18:19 | |
| JayF | I think it was left the way it was because it had been that way and backwards compat, etc | 18:20 |
| kfox1111 | but ironic thinks it should be 1024000 | 18:20 |
| JayF | that's really awkward | 18:20 |
| NobodyCam | JayF: you see deva's doc refactoring patches from lastnight | 18:20 |
| JayF | NobodyCam: no | 18:20 |
| * JayF was ETO yesterday | 18:20 | |
| NobodyCam | like three or four of them | 18:20 |
| *** spandhe has joined #openstack-ironic | 18:21 | |
| jroll | right, 1000GB * 1024 | 18:21 |
| jroll | or something | 18:21 |
| *** Marga_ has joined #openstack-ironic | 18:21 | |
| jroll | GiB vs GB always confuses me | 18:21 |
| kfox1111 | yeah. but hd vendors never count that way anymore. | 18:21 |
| jroll | NobodyCam: he removed the only node-create command | 18:21 |
| jroll | right | 18:21 |
| kfox1111 | they always are base 10. | 18:21 |
| jroll | always? or just consumer drivers? | 18:21 |
| jroll | drives? | 18:21 |
| JayF | it's pretty much always jroll | 18:22 |
| kfox1111 | so if they say they are 1TB, it means 1,000,000,000,000 | 18:22 |
| jroll | ugh | 18:22 |
| kfox1111 | always. | 18:22 |
| JayF | that's why our "32 GB" satadoms show up as less via df | 18:22 |
| JayF | OS tools tell the troof, marketeers and box-writing folk do not | 18:22 |
| jroll | yeah | 18:22 |
| kfox1111 | Some customers sued them a while back for not giving them base2 numbers. | 18:22 |
| kfox1111 | the court upheld what the vendors were doing. | 18:22 |
| jroll | yeah, I remember that | 18:22 |
| kfox1111 | so its in the hd vendors best interest to always use base 10 so they can inflate their numbers. | 18:23 |
| kfox1111 | at least they aren't as bad as the old game cartridge convention, giving you everything in megabits. ;) | 18:24 |
| jroll | lol | 18:24 |
| *** dlaube has joined #openstack-ironic | 18:24 | |
| *** Marga__ has joined #openstack-ironic | 18:24 | |
| *** igordcard has quit IRC | 18:25 | |
| *** Marga_ has quit IRC | 18:25 | |
| *** achanda_ has joined #openstack-ironic | 18:26 | |
| kfox1111 | arg. I gota figure out how to get ironic to specify the ipmi serial console... not seeing whats going on sucks. :) | 18:27 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic: Updated from global requirements https://review.openstack.org/135963 | 18:27 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/135964 | 18:27 |
| kfox1111 | hmm... looks like the node imaged... its pingable now.. | 18:27 |
| jroll | kfox1111: there's a config option for extra kernel params, if you're just trying to set the console | 18:28 |
| kfox1111 | yeah. in the flavor or node? | 18:28 |
| jroll | it's an ironic-conductor config | 18:28 |
| * jroll looks | 18:28 | |
| kfox1111 | ah. ok. | 18:28 |
| kfox1111 | just gota poke: console=ttyS1,115200 in there somehow. | 18:29 |
| jroll | pxe.pxe_append_params | 18:29 |
| jroll | "pxe." meaning in the [pxe] section | 18:30 |
| kfox1111 | perfect. thanks. | 18:30 |
| *** achanda has quit IRC | 18:30 | |
| jroll | np | 18:30 |
| kfox1111 | gota love the crudini tool. :) | 18:31 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/135985 | 18:33 |
| NobodyCam | brb | 18:33 |
| *** achanda_ is now known as achanda | 18:34 | |
| *** pelix has quit IRC | 18:35 | |
| lucasagomes | I will call it a day folks | 18:37 |
| lucasagomes | have a great weekend | 18:37 |
| lucasagomes | see ya | 18:37 |
| JayF | later lucas | 18:38 |
| *** lucasagomes is now known as lucas-beer | 18:39 | |
| NobodyCam | have a great weekend lucas-beer | 18:39 |
| kfox1111 | ahhh. there we go. serial console. thanks. :) | 18:40 |
| jroll | night lucas | 18:40 |
| *** imtiaz has quit IRC | 18:40 | |
| JayF | kfox1111: Mind talking about what you're using Ironic for? /me curious | 18:40 |
| JayF | also wondering if I should evangalize the agent driver to you :P | 18:40 |
| jroll | "deploying bare metal" | 18:41 |
| *** imtiaz has joined #openstack-ironic | 18:41 | |
| JayF | jroll: "Deploying openstack clouds" vs "deploying instances on demand for customers" | 18:41 |
| kfox1111 | right now, just kicking the tires. I work for the pacific northwest national lab, in their super computer devision. | 18:41 |
| kfox1111 | We're using openstack right now internally to provide support services for our big hpc cluster. | 18:42 |
| jroll | oh, hpc stuff? I think you mentioned that yesterday | 18:42 |
| *** andreykurilin__ has quit IRC | 18:42 | |
| *** andreykurilin_ has joined #openstack-ironic | 18:42 | |
| kfox1111 | We use cobbler for most deployments and are interested in ironic as a possible replacement for the provisioning piece. | 18:42 |
| *** Marga__ has quit IRC | 18:42 | |
| kfox1111 | I think we'll still need it for its awesome repo mirring capability. | 18:43 |
| *** Marga_ has joined #openstack-ironic | 18:43 | |
| jroll | neat | 18:43 |
| kfox1111 | hmm.... Calling 'http://169.254.169.254/2009-04-04/meta-data/instance-id' failed [25/120s]: request error [HTTPConnectionPool(host='169.254.169.254', port=80): Max retries exceeded with url: /2009-04-04/meta-data/instance-id (Caused by <class 'socket.error'>: [Errno 113] No route to host)] | 18:43 |
| kfox1111 | is that expected? | 18:43 |
| JayF | That's from cloud-init inside the image? | 18:43 |
| jroll | are you running nova's metadata service? | 18:43 |
| kfox1111 | yeah. hmmm.. let me double check on its status. | 18:44 |
| jroll | so that's probably a networking thing | 18:44 |
| jroll | or configs maybe? | 18:44 |
| jroll | idk how that works | 18:44 |
| jroll | the 169.254.169.254 looks sketchy to me :P | 18:44 |
| NobodyCam | thats metadta service | 18:44 |
| *** ChuckC has joined #openstack-ironic | 18:44 | |
| kfox1111 | k. I've debugged the metadata server before. just wan't sure it was expected to work with bare metal or not. | 18:44 |
| kfox1111 | 169.254.169.254 is a link local address. | 18:45 |
| jroll | right | 18:45 |
| jroll | your metadata service is probably not accessible via that IP | 18:45 |
| JayF | kfox1111: I suspect you need physical networking to cooperate to make metadata service work | 18:45 |
| jroll | I would assume | 18:45 |
| *** imtiaz has quit IRC | 18:45 | |
| JayF | kfox1111: we're using ConfigDrive downstream in what we do, and patches are up to support it already (I think?) but only for agent driver | 18:45 |
| kfox1111 | hmm.. | 18:45 |
| jroll | JayF: ironic patches are merged :) | 18:46 |
| JayF | kfox1111: jroll would know as he's implementing configdrive for upstream :) | 18:46 |
| JayF | jroll: oh wow, that was fast | 18:46 |
| jroll | nova patches aren't quite up yet as I need to build a swift wrapper etc | 18:46 |
| jroll | I'm good like that :P | 18:46 |
| kfox1111 | trying to remember how this works... with neutron, I think the router namespace puts in a metadata proxy. | 18:46 |
| kfox1111 | but since I'm using a flat network, I didn't build a router... | 18:46 |
| jroll | right | 18:47 |
| jroll | can you set that IP in cloud-init configs, maybe? | 18:47 |
| jroll | or, the correct IP, I should say | 18:47 |
| kfox1111 | would have to build images with it hardcoded I think. easier just to get the service on the right ip. | 18:47 |
| jroll | right, you need image cooperation | 18:48 |
| jroll | on a link local IP? | 18:48 |
| kfox1111 | should be able to build a dummyr router connected to the network with no gateway and it should then work I think. | 18:49 |
| JayF | jroll: IDK if metadata service URL is configurable, that's a damn good idea | 18:49 |
| jroll | well, I would think normally the OVS or whatever on the hypervisor would be what handles that | 18:49 |
| jroll | I have no idea, this is way out of my knowledge range | 18:49 |
| jroll | JayF: yes | 18:49 |
| kfox1111 | JayF: chicken and the egg problem. how does an image built by a 3rd party know what url to contact? | 18:50 |
| kfox1111 | Solution, always the same link local one. :/ | 18:51 |
| kfox1111 | Amazon garantee's it, so everyone uses it. | 18:51 |
| jroll | sadface | 18:51 |
| NobodyCam | tripleo folks have been known to do somehting like: https://github.com/openstack/tripleo-image-elements/blob/master/elements/devstack/install.d/98-baremetal-network#L16 | 18:52 |
| *** imtiaz has joined #openstack-ironic | 18:52 | |
| JayF | kfox1111: ++ I agree, but you can build your own images | 18:52 |
| JayF | kfox1111: the images we use for Rackspace OnMetal have a bit of cloud-init configuration in them too, to bootstrap it up | 18:52 |
| kfox1111 | I know. but am lazy. I'd rather all images work unchanged rather then tweak every one. ;) | 18:53 |
| kfox1111 | ok. have a router... | 18:53 |
| *** alexpilotti has quit IRC | 18:53 | |
| *** alexpilotti has joined #openstack-ironic | 18:53 | |
| kfox1111 | hmm... no I don't... strange... | 18:53 |
| *** Marga_ has quit IRC | 18:54 | |
| *** Marga_ has joined #openstack-ironic | 18:54 | |
| kfox1111 | ah... l3-agent is not working... probably because of the inital packstack config.. just a sec. | 18:54 |
| *** Marga_ has quit IRC | 18:57 | |
| *** Marga_ has joined #openstack-ironic | 18:58 | |
| kfox1111 | hmm.. still no metadata agent rule... | 18:58 |
| *** Marga_ has quit IRC | 18:58 | |
| *** Marga_ has joined #openstack-ironic | 18:59 | |
| *** JackieHurst has joined #openstack-ironic | 19:00 | |
| *** igordcard has joined #openstack-ironic | 19:01 | |
| *** alexpilotti has quit IRC | 19:03 | |
| *** killer_prince has quit IRC | 19:03 | |
| *** jistr|mtg has quit IRC | 19:04 | |
| *** killer_prince has joined #openstack-ironic | 19:05 | |
| *** killer_prince is now known as lazy_prince | 19:05 | |
| kfox1111 | so, anyway, we deploy a lot of our systems using xcat, cobbler or rocks. It would be interesting if we could use ironic, and let ourselves more easily switch nodes between hpc and more traditional cloud usecases. | 19:08 |
| kfox1111 | looking at the messages in the log, the deploy image exports an iscsi volume and conductor copies the bits to it? | 19:10 |
| *** ifarkas has quit IRC | 19:13 | |
| jroll | yes, that's how the pxe drivers work | 19:15 |
| jroll | this is all the deploy ramdisk does: https://github.com/openstack/diskimage-builder/blob/master/elements/deploy-ironic/init.d/80-deploy-ironic | 19:16 |
| jroll | the agent drivers run a python agent in a ramdisk, which has a rest api that ironic can call to tell it to do things, like download an image over http and write it to disk | 19:16 |
| kfox1111 | so the agent driver version should scale better? | 19:17 |
| jroll | yes, that's one of the reasons we built it :) | 19:18 |
| kfox1111 | ok. cool. | 19:18 |
| *** lazy_prince has quit IRC | 19:18 | |
| kfox1111 | we usually have a few hundred machines per associate director machine, so they could potentially overwhelm it. I'd guess we'd do one ironic conductor per associate director. | 19:19 |
| *** killer_prince has joined #openstack-ironic | 19:19 | |
| *** killer_prince is now known as lazy_prince | 19:19 | |
| *** andreykurilin_ has quit IRC | 19:19 | |
| jroll | sounds about right | 19:20 |
| JayF | I think we have something around 200 nodes per conductor in our env? | 19:20 |
| jroll | we (rackspace) are currently running just over 200 nodes per conductor | 19:20 |
| jroll | using the agent driver | 19:20 |
| *** andreykurilin_ has joined #openstack-ironic | 19:20 | |
| kfox1111 | ok. cool. | 19:21 |
| *** lazy_prince has quit IRC | 19:24 | |
| kfox1111 | hmm... the metadata agent/proxy is up. just doesn't seem to be working... lets see... | 19:24 |
| *** Marga__ has joined #openstack-ironic | 19:25 | |
| *** Marga_ has quit IRC | 19:25 | |
| *** JackieHurst is now known as GiancarloRizzi | 19:27 | |
| *** killer_prince has joined #openstack-ironic | 19:29 | |
| *** GiancarloRizzi has quit IRC | 19:30 | |
| *** JackieHurst has joined #openstack-ironic | 19:30 | |
| *** killer_prince is now known as lazy_prince | 19:30 | |
| *** romcheg has joined #openstack-ironic | 19:32 | |
| openstackgerrit | Chris Krelle proposed openstack/python-ironicclient: Update README https://review.openstack.org/134541 | 19:33 |
| *** romcheg has quit IRC | 19:33 | |
| *** romcheg has joined #openstack-ironic | 19:33 | |
| *** blinky_ghost has quit IRC | 19:34 | |
| *** Marga__ has quit IRC | 19:34 | |
| *** Marga_ has joined #openstack-ironic | 19:35 | |
| *** Marga_ has quit IRC | 19:36 | |
| *** yjiang5_away is now known as yjiang5 | 19:36 | |
| *** Marga_ has joined #openstack-ironic | 19:36 | |
| *** Marga_ has quit IRC | 19:37 | |
| *** Marga_ has joined #openstack-ironic | 19:38 | |
| *** Marga_ has quit IRC | 19:39 | |
| *** Marga_ has joined #openstack-ironic | 19:40 | |
| *** igordcard has quit IRC | 19:40 | |
| *** JackieHurst has quit IRC | 19:41 | |
| *** lazy_prince has quit IRC | 19:41 | |
| *** bahazc has joined #openstack-ironic | 19:41 | |
| *** killer_prince has joined #openstack-ironic | 19:43 | |
| *** Marga_ has quit IRC | 19:43 | |
| *** killer_prince is now known as lazy_prince | 19:43 | |
| *** Marga_ has joined #openstack-ironic | 19:44 | |
| *** bahazc has quit IRC | 19:45 | |
| *** penick has joined #openstack-ironic | 19:46 | |
| *** Marga_ has quit IRC | 19:48 | |
| *** Marga_ has joined #openstack-ironic | 19:49 | |
| *** romcheg has quit IRC | 19:50 | |
| *** romcheg has joined #openstack-ironic | 19:51 | |
| *** spandhe has quit IRC | 19:52 | |
| *** bahazc has joined #openstack-ironic | 19:53 | |
| *** spandhe has joined #openstack-ironic | 19:54 | |
| *** Marga_ has quit IRC | 19:54 | |
| *** Marga_ has joined #openstack-ironic | 19:55 | |
| *** bahazc has quit IRC | 19:55 | |
| *** Marga_ has quit IRC | 20:00 | |
| *** Marga_ has joined #openstack-ironic | 20:01 | |
| *** Marga_ has quit IRC | 20:01 | |
| *** Marga_ has joined #openstack-ironic | 20:02 | |
| *** lazy_prince has quit IRC | 20:03 | |
| *** killer_prince has joined #openstack-ironic | 20:04 | |
| *** killer_prince is now known as lazy_prince | 20:05 | |
| *** Marga_ has quit IRC | 20:06 | |
| *** datajerk has joined #openstack-ironic | 20:07 | |
| kfox1111 | yay. passed my Mirantis certification. :) | 20:09 |
| dlaube | nice! | 20:11 |
| dlaube | congrats kfox1111 | 20:11 |
| dlaube | how many questions? or is it lab based or something? | 20:12 |
| kfox1111 | lab based. | 20:17 |
| *** r-daneel has quit IRC | 20:17 | |
| dlaube | any ironic material in there yet? | 20:17 |
| kfox1111 | hehe. no, not yet. | 20:18 |
| dlaube | darn | 20:18 |
| kfox1111 | now, back to figuring out why the metadata server isn't working quite right... | 20:18 |
| *** lucas-beer has quit IRC | 20:19 | |
| kfox1111 | OH.... I think I know why.... its being routed via the neutron router... | 20:20 |
| *** dprince has quit IRC | 20:22 | |
| NobodyCam | brb quick run to grab a some smokes, maybe some starbuxks | 20:31 |
| *** andreykurilin_ has quit IRC | 20:37 | |
| *** achanda has quit IRC | 20:41 | |
| kfox1111 | YAY! | 20:42 |
| kfox1111 | Metadata server's working. | 20:42 |
| kfox1111 | so nova boot -> ironic, ssh -> instance = joy. :) | 20:42 |
| *** jjohnson2 has joined #openstack-ironic | 20:42 | |
| jroll | \o/ | 20:43 |
| jjohnson2 | fyi, I just pushed 1.0 RC to pypi for confluent | 20:43 |
| kfox1111 | ok, so to get the metadata server working, I setup a router on a different address, on the same network, then | 20:44 |
| kfox1111 | neutron subnet-update osg-subnet --host-route 'destination=169.254.169.254/32,nexthop=192.168.122.248' | 20:44 |
| kfox1111 | alittle hackish, but it worked. :) | 20:45 |
| kfox1111 | now to really kick the tires. :) | 20:45 |
| kfox1111 | hmm... so ironic deploy only uses the first drive it finds? | 20:46 |
| jroll | yes, it uses /dev/sda | 20:47 |
| jroll | another benefit of the agent | 20:47 |
| jroll | standard agent uses smallest disk larger than 4GB, you can write a few lines of python to customize that | 20:47 |
| * kfox1111 thinks he needs to start looking closer at the agent... | 20:47 | |
| kfox1111 | these nodes have two identical 1T disks. | 20:48 |
| kfox1111 | usually just raid0 them. | 20:48 |
| JayF | We don't support configuring raid via ironic today | 20:48 |
| JayF | however it would absolutely be possible to hack into the agent | 20:49 |
| kfox1111 | any pointers to playing with the agent? | 20:49 |
| JayF | driver: s/pxe/agent/ | 20:50 |
| *** rushiagr is now known as rushiagr_away | 20:50 | |
| JayF | you can use a ramdisk of master ipa | 20:50 |
| JayF | http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 20:50 |
| JayF | or you can build your own (it's simple, and documented in openstack/ironic-python-agent) | 20:50 |
| JayF | and IPA only handles full disk images | 20:50 |
| JayF | so you'll have to use a full disk image instead of a partition image | 20:51 |
| *** Marga_ has joined #openstack-ironic | 20:51 | |
| JayF | jroll: ^ forgetting anything? | 20:51 |
| kfox1111 | cool. I'll give it a shot. :) | 20:52 |
| JayF | kfox1111: you can embed your own hardware managers into the agent as wel | 20:52 |
| JayF | kfox1111: for instance; this is what we use: https://github.com/rackerlabs/onmetal-ironic-hardware-manager | 20:52 |
| *** alexpilotti has joined #openstack-ironic | 20:53 | |
| * NobodyCam is back | 20:53 | |
| kfox1111 | so, that lets you configure additional hardware or something? | 20:54 |
| JayF | kfox1111: so today; in upstream Ironic, there's not much interesting hardware stuff that the agent does | 20:54 |
| JayF | kfox1111: at Rackspace, we're running an early version of "zapping" which has specs up todya | 20:54 |
| JayF | which does things like erase disks, setup bios settings, flash firmwares, etc etc (anything you'd want to happen between workloads) | 20:55 |
| kfox1111 | ah. ok. | 20:55 |
| JayF | That's obviously very hardware specific; so it's important to be able to make your own code to do it | 20:55 |
| kfox1111 | yeah. makes sense. | 20:55 |
| *** athomas has quit IRC | 20:56 | |
| *** Marga_ has quit IRC | 20:56 | |
| kfox1111 | I can't remmber if I asked or not. can you use the agent driver to let you pxeboot once, then boot off of the image until the instance is deleted? | 20:57 |
| jroll | JayF: dunno, don't think so | 20:59 |
| jroll | kfox1111: the agent driver today only supports whole disk images, and booting the instance from disk | 20:59 |
| jroll | it never pxe boots the instance today | 20:59 |
| kfox1111 | cool. thats awesome. :) | 20:59 |
| kfox1111 | seperating the kernel/initrd out is kind of a pain, | 21:00 |
| kfox1111 | and I've wondered about the trippleo type undercloud stuff. if you must pxeboot your controllers, that means you must keep your seed around. If they can local boot after deployed, the seed can go away. | 21:00 |
| jroll | yeah, indeed | 21:01 |
| kfox1111 | it lets you yum upgrade the instance nicely too. | 21:01 |
| jroll | right | 21:01 |
| jroll | tripleo doesn't believe in that | 21:01 |
| kfox1111 | yeah. Ideally you wouldn't have to. but sometimes its nice to have the option. | 21:02 |
| jroll | definitely | 21:03 |
| kfox1111 | is there plans on making the agent the default at some point? | 21:03 |
| JayF | kfox1111: a lot of the tripleo people like the behavior of none of the instances coming up until the control plane is up | 21:04 |
| jroll | that's a bit of a goal, yes | 21:04 |
| JayF | kfox1111: idgi but it's considered by some people to be a feature not a bug | 21:04 |
| kfox1111 | having to keep one seed vm up and running so your "fault tolerent" cluster can boot seems counter intuitive. ;) | 21:05 |
| *** Marga_ has joined #openstack-ironic | 21:07 | |
| *** romcheg has quit IRC | 21:08 | |
| kfox1111 | is there any cinder support in ironic yet? | 21:10 |
| *** Marga_ has quit IRC | 21:10 | |
| *** Marga_ has joined #openstack-ironic | 21:10 | |
| jroll | not yet, folks are working on it | 21:11 |
| kfox1111 | k. | 21:12 |
| jroll | (theoretically, I haven't seen code) | 21:12 |
| kfox1111 | I'm guessing security groups are off the table too? The docs had you configure them, but I'm guessing they don't work at all? | 21:12 |
| jroll | probably not? | 21:13 |
| jroll | I haven't tried, personally | 21:13 |
| jroll | or heard of folks trying | 21:13 |
| jroll | but neutron doesn't have support for hardware switches yet | 21:13 |
| kfox1111 | do you need to match the ironic-python-agent image build to the openstack release, or will trunk always work with the latest stable release? | 21:14 |
| jroll | trunk will always work | 21:14 |
| kfox1111 | yeah. that was what I was thinking. just wondered if somethign happened that I had missed. | 21:14 |
| kfox1111 | k. thx. | 21:14 |
| * NobodyCam notes date and time : 21:14 | jroll > trunk will always work | 21:14 | |
| kfox1111 | hehe. | 21:15 |
| jroll | nope, we do our best to keep trunk working with last stable | 21:15 |
| jroll | shhhhh | 21:15 |
| NobodyCam | :-p | 21:15 |
| kfox1111 | Oh. btw, thanks for posting official prebuilt images. something that TripleO REALLY needs to do. | 21:16 |
| jroll | :) | 21:17 |
| *** imtiaz has quit IRC | 21:18 | |
| *** imtiaz has joined #openstack-ironic | 21:18 | |
| NobodyCam | "the J crew" are a awesome group of guys | 21:19 |
| jroll | d'aww | 21:19 |
| JayF | kfox1111: it's built by the CI, so those will change on each commit to IPA | 21:20 |
| kfox1111 | even better. :) | 21:20 |
| JayF | kfox1111: but I strongly, strongly reccomend you end up building your own agent. Ramdisk agent is super useful for fleet management, just gotta drop an ssh key inside :) | 21:20 |
| * JayF stuffs all kinds of tools into his deploy ramdisk | 21:20 | |
| kfox1111 | yeah. I don't mind building everyting. We have a jenkins setup for that. We end up customizing all of our images. | 21:21 |
| kfox1111 | Its more the, I want to kick the tires to see if something is worth spending time on. | 21:21 |
| kfox1111 | You have to implement all of TripleO before you can ensure its a good fit. :/ | 21:21 |
| jroll | ha | 21:21 |
| kfox1111 | I'd rather try it out at scale with prebuilt images, make sure its what we want, then customize the bits we care about. | 21:22 |
| jroll | +1 | 21:22 |
| *** david-lyle has joined #openstack-ironic | 21:22 | |
| *** Marga_ has quit IRC | 21:22 | |
| *** Marga_ has joined #openstack-ironic | 21:23 | |
| * devananda waves | 21:23 | |
| jroll | ohai | 21:23 |
| devananda | hi! | 21:23 |
| devananda | in the office today, which means talking with people and not getting any work done | 21:23 |
| jroll | woohoo | 21:23 |
| kfox1111 | yeah, thats the way of it isnt it... | 21:24 |
| jroll | sounds like a blast :P | 21:24 |
| *** Marga_ has quit IRC | 21:26 | |
| *** Marga_ has joined #openstack-ironic | 21:27 | |
| NobodyCam | wb devananda | 21:29 |
| NobodyCam | devananda: talking to folks IS getting work done! | 21:30 |
| jroll | I like to pretend it i | 21:30 |
| jroll | s | 21:30 |
| devananda | jroll: I just call it like it is | 21:30 |
| *** Marga_ has quit IRC | 21:30 | |
| *** Marga_ has joined #openstack-ironic | 21:31 | |
| NobodyCam | humm -c has stoped working on my devtest... gota be somehting with my setup | 21:33 |
| *** Marga_ has quit IRC | 21:34 | |
| *** achanda has joined #openstack-ironic | 21:35 | |
| *** Marga_ has joined #openstack-ironic | 21:35 | |
| *** Marga_ has quit IRC | 21:37 | |
| *** Marga_ has joined #openstack-ironic | 21:38 | |
| *** romcheg has joined #openstack-ironic | 21:38 | |
| *** Marga_ has quit IRC | 21:38 | |
| *** Marga_ has joined #openstack-ironic | 21:39 | |
| *** Marga__ has joined #openstack-ironic | 21:43 | |
| *** Marga_ has quit IRC | 21:43 | |
| *** Marga__ has quit IRC | 21:44 | |
| dlaube | hey guys, just deployed an ubuntu image with ironic and I'm sitting at a login prompt | 21:44 |
| *** Marga_ has joined #openstack-ironic | 21:44 | |
| NobodyCam | nice!!! | 21:44 |
| jroll | \o/ | 21:44 |
| dlaube | :D | 21:44 |
| *** Marga_ has quit IRC | 21:44 | |
| dlaube | I wish I could say it was all of my doing but I certainly had some help!!! | 21:44 |
| -openstackstatus- NOTICE: gating is going offline while we deal with a broken block device, eta unknown | 21:44 | |
| *** ChanServ changes topic to "gating is going offline while we deal with a broken block device, eta unknown" | 21:44 | |
| dlaube | BUT.... | 21:44 |
| *** Marga_ has joined #openstack-ironic | 21:45 | |
| dlaube | can anyone tell me the username associated with the adminpass that nova boot spits out? | 21:45 |
| NobodyCam | how did you build your image | 21:45 |
| dlaube | image-builder | 21:45 |
| jroll | if it's the standard ubuntu image, should be root or ubuntu | 21:45 |
| NobodyCam | did you add the stackuser element | 21:45 |
| dlaube | ahhh | 21:45 |
| dlaube | looks like an image rebuild and glance create is in order | 21:46 |
| dlaube | heh | 21:46 |
| NobodyCam | also the root password should be listed on the nova boot output | 21:46 |
| NobodyCam | but ya I use stackuser element alot :-p | 21:47 |
| jroll | yeah, always try root first | 21:47 |
| dlaube | root user was a no-go | 21:47 |
| NobodyCam | with password from nova boot? | 21:47 |
| NobodyCam | I think it listed as adminPass | 21:49 |
| dlaube | yep, with the adminPass that nova boot spits out | 21:49 |
| NobodyCam | :( | 21:49 |
| NobodyCam | ubuntu / ubuntu? | 21:49 |
| NobodyCam | google seems to say Try 'ubuntu' with an empty password | 21:50 |
| dlaube | nope | 21:51 |
| dlaube | neither of those two work | 21:51 |
| jroll | I mean | 21:51 |
| jroll | do you have the metadata service set up properly? | 21:51 |
| NobodyCam | did cloud-init fire off | 21:51 |
| jroll | if not, the passowrd probably isn't getting set | 21:51 |
| dlaube | nope | 21:51 |
| jroll | yeah that | 21:51 |
| NobodyCam | there ya go | 21:52 |
| dlaube | saw some warnings at provision time about metadata failure | 21:52 |
| dlaube | aha! | 21:52 |
| NobodyCam | stackuser element will allow you log in to the instance | 21:52 |
| dlaube | I thought it was only needed for cloud-init stuff | 21:52 |
| NobodyCam | but I'd suggest setting up meta data | 21:52 |
| jroll | cloud-init is what sets the root password | 21:52 |
| NobodyCam | and installs ssh keys | 21:53 |
| JayF | jroll: Are you sure cloud-init will set a root password? | 21:53 |
| jroll | I'm not | 21:53 |
| JayF | jroll: Pretty sure it only does $distroname users + sudo and/or keys for any user | 21:53 |
| jroll | but what else would | 21:53 |
| NobodyCam | JayF: it should yes | 21:53 |
| JayF | but imbw | 21:53 |
| NobodyCam | JayF: how else would nova set the admin password | 21:54 |
| JayF | NobodyCam: nova-agent | 21:54 |
| jroll | there's nova-agent | 21:54 |
| NobodyCam | agent? | 21:54 |
| NobodyCam | icky | 21:54 |
| jroll | is that another rackspace thing | 21:54 |
| * jroll hides | 21:54 | |
| NobodyCam | lol | 21:54 |
| dlaube | heh | 21:55 |
| NobodyCam | humm why is my neutron only listing dhcp agent... what happened to my Open vSwitch agent. | 21:56 |
| NobodyCam | :( | 21:56 |
| dlaube | ok, well thanks for the info gents | 21:56 |
| dlaube | going to get back to it and see what we can do | 21:56 |
| *** Marga_ has quit IRC | 21:56 | |
| NobodyCam | :) | 21:56 |
| JayF | jroll: I'm somewhat sure that's upstream? | 21:56 |
| JayF | NobodyCam: I think agents are icky too; that's why onmetal uses only cloud-init | 21:56 |
| NobodyCam | :) +++++ | 21:57 |
| *** Marga_ has joined #openstack-ironic | 21:57 | |
| dlaube | seriously, going to head up into SF one of these days and buy you guys some coffees or beers | 21:57 |
| NobodyCam | I believe we (ironic and to that extent nova) should be hands off a user image | 21:57 |
| jroll | JayF: I have no idea, man | 21:57 |
| jroll | dlaube: and a cider for JayF :P | 21:58 |
| jroll | NobodyCam: ++ | 21:58 |
| jroll | I don't want my provider running any code in my instance | 21:58 |
| JayF | I won't make a value judgement about doing that generally, becuase the company I work for manages servers and clouds and I'd rather them use a python agent than a human agent | 21:59 |
| JayF | but if people just want compute, they should just get that, without anyone mucking in their instance/image | 21:59 |
| dlaube | :P | 21:59 |
| dlaube | true that | 21:59 |
| NobodyCam | yes there are cases when I ask for the provider to do somehting for me... (ie I have asked for backup's with past providers) thou I installed their software my self. | 22:01 |
| *** Marga_ has quit IRC | 22:03 | |
| *** Marga_ has joined #openstack-ironic | 22:05 | |
| dlaube | hmm… any way to specify the metadata agent IP at provision time? | 22:05 |
| *** Marga_ has quit IRC | 22:05 | |
| dlaube | or is that something that comes out of the nova.conf ? | 22:05 |
| *** Marga_ has joined #openstack-ironic | 22:06 | |
| NobodyCam | oh looks like I have to stop installing Boinc updates now | 22:07 |
| NobodyCam | the ip I believe is aways 169.254.169.254 | 22:07 |
| NobodyCam | ? | 22:07 |
| dlaube | interesting | 22:09 |
| PaulCzar | I think I'm making some progress here . I've got the virtualbox power stuff working with the ssh agent ... now I need to figure out how to plumb networking over | 22:13 |
| kfox1111 | dlaube: metadata ip always should be 169.254.169.254. | 22:14 |
| dlaube | ahh ok thanks NobodyCam and kfox1111 | 22:14 |
| PaulCzar | the neutron dhcp agent runs in a network namespce ..so I assume I have to do something to allow a the external node to dhcp off it ? | 22:15 |
| *** ryanpetrello has quit IRC | 22:17 | |
| kfox1111 | you need to ensure the network you create is bound to the flat provider network exposed to the external host. | 22:19 |
| PaulCzar | yeah | 22:19 |
| kfox1111 | is there a document describing how to set a node to use the agent? | 22:21 |
| *** imtiaz has quit IRC | 22:23 | |
| JayF | kfox1111: IDK if you can change the driver of a node post-adding it to Ironic | 22:23 |
| JayF | kfox1111: might wanna delete that node and recreate it, just specifying the agent driver | 22:24 |
| *** Marga_ has quit IRC | 22:24 | |
| kfox1111 | JayF: Thats ok. | 22:25 |
| kfox1111 | I'm not seeing it in the driver-list. I'm assuming I need to add it to enabled_drivers | 22:25 |
| JayF | aha, yes | 22:26 |
| kfox1111 | know offhand what the line needs to be? | 22:26 |
| jroll | agent_*, where * is ssh/ipmitool/pyghmi | 22:29 |
| *** Marga_ has joined #openstack-ironic | 22:29 | |
| kfox1111 | k. | 22:30 |
| *** jjohnson2 has quit IRC | 22:30 | |
| kfox1111 | There we go. | 22:32 |
| kfox1111 | crudini --set /etc/ironic/ironic.conf DEFAULT enabled_drivers pxe_ipmitool,agent_ipmitool,agent_ssh | 22:32 |
| jroll | nice | 22:32 |
| jroll | you don't need agent_ssh unless you want the ssh power driver, btw | 22:32 |
| kfox1111 | yeah. I figured it was easy to enable, so.. why not... | 22:33 |
| kfox1111 | hmm.. and then I need to update agent_pxe_append_params too... | 22:34 |
| jroll | yeah | 22:36 |
| kfox1111 | so, since kvm images are usually partitioned, I should be able to use a stock ubuntu cloud image with the agent? | 22:36 |
| jroll | should | 22:37 |
| jroll | but I've noticed this before: https://github.com/openstack/diskimage-builder/blob/master/elements/ubuntu/install.d/10-support-physical-hardware#L6 | 22:37 |
| kfox1111 | ah. | 22:38 |
| kfox1111 | hmm... is there any way to tag nodes with a name so they can be tracked easier? | 22:41 |
| *** imtiaz has joined #openstack-ironic | 22:46 | |
| jroll | kfox1111: soon https://review.openstack.org/#/c/134439/ | 22:49 |
| kfox1111 | cool. thx. | 22:50 |
| kfox1111 | hehe. better reason would be, I don't want to print out, then affix a uuid to a physical node for harware replacement. ;) | 22:50 |
| kfox1111 | and have to update the physical label when I delete and re-add the node. ;) | 22:51 |
| jroll | so, node.extra is a dict | 22:52 |
| jroll | and super nice for stuffing random data | 22:52 |
| jroll | (like server name) | 22:52 |
| kfox1111 | cool. I'll give that a shot. | 22:53 |
| dlaube | couldn't you also just add that extra data to the node properties or extra ? | 22:53 |
| *** anderbubble has quit IRC | 22:53 | |
| *** achanda has quit IRC | 22:53 | |
| NobodyCam | dlaube: I'd use extra for that data | 22:53 |
| *** romcheg has quit IRC | 22:54 | |
| *** achanda has joined #openstack-ironic | 22:54 | |
| JayF | We totally add lots of IDs in node.extra | 22:54 |
| kfox1111 | would be nice if it showed up in node-list though. | 22:55 |
| kfox1111 | yeah. I could see extra being useful for stuff like, rack, U, slot in chassis, etc. | 22:56 |
| jroll | node-list --detail might be a thing | 22:56 |
| jroll | or /nodes/detail definitely is | 22:56 |
| kfox1111 | its totally ignored by things? | 22:56 |
| jroll | extra? yeah | 22:57 |
| kfox1111 | k. | 22:57 |
| kfox1111 | strange... disk-image-create -u ubuntu -o my-image doesn't partition things? | 22:58 |
| *** romcheg has joined #openstack-ironic | 22:58 | |
| *** imtiaz_ has joined #openstack-ironic | 22:58 | |
| NobodyCam | partition things? | 22:58 |
| jroll | what do you mean "partition things" | 22:58 |
| *** Marga_ has quit IRC | 22:58 | |
| kfox1111 | will the agent work with that image, or do I need to get it partitioned nicely? | 22:58 |
| kfox1111 | root is in /dev/sda. | 22:59 |
| jroll | the agent takes a qcow and dd's it | 22:59 |
| jroll | that's it | 22:59 |
| kfox1111 | not in /dev/sda1 | 22:59 |
| jroll | oh, so no bootloader in the image? | 22:59 |
| NobodyCam | kfox1111: you want to add vm element for agent | 22:59 |
| kfox1111 | ah. ok. | 22:59 |
| kfox1111 | disk-image-create -u ubuntu -o my-bare-image vm | 22:59 |
| NobodyCam | sure | 23:00 |
| kfox1111 | k. thx. | 23:00 |
| kfox1111 | so if you use disk image builder on ubuntu images, it always installs the physical hardware bits into the image too? | 23:01 |
| devananda | "physical hardware bits" ? | 23:01 |
| NobodyCam | same question here? | 23:01 |
| NobodyCam | hi again devananda :) | 23:01 |
| kfox1111 | < jroll> but I've noticed this before: https://github.com/openstack/diskimage-builder/blob/master/elements/ubuntu/install.d/10-support-physical-hardware#L6 | 23:02 |
| *** imtiaz has quit IRC | 23:02 | |
| NobodyCam | kfox1111: yep thats part of the ubuntu element so it will get included | 23:02 |
| kfox1111 | ok. | 23:03 |
| Haomeng | morning | 23:03 |
| *** mjturek has quit IRC | 23:03 | |
| NobodyCam | morning Haomeng | 23:03 |
| *** jgrimm is now known as zz_jgrimm | 23:05 | |
| kfox1111 | ? http://tarballs.openstack.org/ironic-python-agent/coreos/files/ no longer works. :/ | 23:05 |
| JayF | note /topic | 23:05 |
| Haomeng | NobodyCam: :) | 23:05 |
| kfox1111 | ah. | 23:05 |
| JayF | kfox1111: I'd presume tarballs.o.o is impacted by that | 23:05 |
| NobodyCam | d'oh | 23:06 |
| NobodyCam | lol | 23:06 |
| kfox1111 | yeah... | 23:06 |
| JayF | confirmed by infra, it's impacted | 23:06 |
| *** jeblair is now known as corvus | 23:06 | |
| JayF | kfox1111: really building the image is a cakewalk if you have a trusty vm | 23:07 |
| JayF | not trusty as in trustworthy | 23:07 |
| JayF | trusty as in ubuntu 14.04 :) | 23:07 |
| *** corvus is now known as jeblair | 23:07 | |
| kfox1111 | arg. k. | 23:07 |
| JayF | kfox1111: https://github.com/openstack/ironic-python-agent/tree/master/imagebuild/coreos | 23:08 |
| JayF | kfox1111: clone the thing, run full_trusty_build.sh | 23:08 |
| JayF | kfox1111: if you wanna know what's actually happening; use the README | 23:08 |
| JayF | but that .sh file is exactly what the gate runs against a bare-trusty node to generate the images you were using | 23:08 |
| adam_g | JayF, what are the current memory requirements for DIB built agent ramdisks? | 23:10 |
| JayF | adam_g: I think jroll got it working with 3GB ? | 23:11 |
| NobodyCam | adam_g: at least 3072 | 23:11 |
| adam_g | ah | 23:11 |
| NobodyCam | but I understand testing was done on 8gb systems | 23:11 |
| kfox1111 | I've got an sl7 box I'm building all this on. I should be able to tweak the script to build on sl7? | 23:11 |
| adam_g | i just got multinode ironic devstack working with a second node just hosting libvirt + conductor | 23:11 |
| kfox1111 | its just doing docker stuff? | 23:11 |
| adam_g | we may be able to get concurrency working with 3GB VMs across both nodes | 23:11 |
| JayF | adam_g: lets use the CoreOS image for that now then? until the DIB image goes on a diet and gets better vetted? | 23:14 |
| adam_g | JayF, yeah--this should give us plenty of resources if we end up using it upstream | 23:14 |
| JayF | that's killer man, great job | 23:15 |
| JayF | seriously | 23:15 |
| JayF | that'll also be nice if we can get it working locally as well | 23:15 |
| kfox1111 | copying the full_trusty_build.sh and changing it to yum install docker instead seems to be working. | 23:16 |
| adam_g | i just finished hacking it together based on some multi node stuff others are working on it.. it should be easy with some devstack updates | 23:16 |
| jroll | adam_g: yeah, coreos ramdisk only needs 1gb, let's just do that | 23:18 |
| adam_g | jroll, word | 23:19 |
| jroll | kfox1111: just curious, what's sl7? | 23:19 |
| kfox1111 | oh. scientific linux 7. almost identical to centos7. | 23:20 |
| jroll | ah, neat | 23:21 |
| *** andreykurilin_ has joined #openstack-ironic | 23:21 | |
| JayF | kfox1111: yeah; the build itself is driven totally by docker, if you have recentish docker you're gtg | 23:21 |
| *** alexpilotti has quit IRC | 23:22 | |
| JayF | kfox1111: The container IPA runs in is still ubuntu; although I wouldn't be opposed to s/coreos/coreos-ubuntu/ and adding a coreos-fedora | 23:22 |
| JayF | but I think I want to make the builder more split-out before doing that | 23:22 |
| kfox1111 | not recent enough for my tastes, but usually does work. :) | 23:22 |
| kfox1111 | doesn't matter to me. We're just much more familior with rhel. | 23:23 |
| JayF | I don't care much about distributions ... I care about init systems, and I <3 me some systemd | 23:23 |
| JayF | so I've been running a lot of fedora lately :) | 23:23 |
| kfox1111 | yeah. I really dont get the flamewar over systemd. | 23:23 |
| kfox1111 | I think its a big misunderstanding. | 23:24 |
| kfox1111 | sure systemd doesn't follow unix philosophy. | 23:24 |
| kfox1111 | but neither does the linux kernel. | 23:24 |
| jroll | modular monolith | 23:25 |
| jroll | is like my new favorite phrase | 23:25 |
| kfox1111 | If you were to s/systemd|lenart/linux|linus/ they suddently would be on the wrong side of the argument. | 23:25 |
| JayF | I think people see systemd-* and think it's all in their init system | 23:25 |
| JayF | when really the project just believes in namespacing | 23:26 |
| kfox1111 | seriously, you really want unix philosophy, its plan9 or the hurd for you. :) | 23:26 |
| kfox1111 | yeah. exactly. | 23:26 |
| JayF | I want to never write an initscript ever again :) | 23:26 |
| kfox1111 | otherwise, you want the linux philosophy. :) | 23:26 |
| JayF | kfox1111: if you're a systemd guy; you should appreciate the IPA image being in coreos; that's driven entirely by systemd (see imagebuild/coreos/oem/cloud-config.yml) | 23:26 |
| kfox1111 | JayF: yeah, I hear you. proper init scripts are a PAIN. | 23:26 |
| * devananda posts more spec reviews | 23:27 | |
| JayF | and the image we use downstream for Rackspace OnMetal has probably about a 3x as large cloud-config.yml | 23:27 |
| * devananda goes AFK | 23:27 | |
| kfox1111 | yeah. coreos is on my list of things to study closer once I have some time. | 23:27 |
| NobodyCam | have a good weekend devananda | 23:27 |
| JayF | well, you are about to be using it. Congratulations :P | 23:27 |
| kfox1111 | wow... its really working on building this image... | 23:28 |
| *** imtiaz_ has quit IRC | 23:29 | |
| *** imtiaz has joined #openstack-ironic | 23:29 | |
| *** imtiaz has quit IRC | 23:33 | |
| kfox1111 | there we go. | 23:35 |
| jroll | yeah, it caches docker layers so it will be faster after the first build | 23:36 |
| kfox1111 | so building on el7 seems to work. | 23:36 |
| JayF | I'd expect so; if the build broke across different OS versions that'd either be my bug or docker's bug | 23:37 |
| JayF | it should just need docker, around 1.0 or newer | 23:38 |
| JayF | and the various basic linux utilities that are on "every" distribution | 23:38 |
| JayF | kfox1111: they fixed http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 23:38 |
| kfox1111 | should be able to modify the shell script to check for ubuntu/redhat and yum install -y docker; systemctl start docker and then the rest of the script is the same. then rename it to something more generic. | 23:38 |
| JayF | kfox1111: JIT for you to be done building your own :) | 23:38 |
| kfox1111 | heh. nice. | 23:39 |
| JayF | kfox1111: in that case; I'd probably ask it be made a separate build file, because I don't wanna make what runs in the gate more complex | 23:39 |
| kfox1111 | stupid murphy. | 23:39 |
| kfox1111 | :) | 23:39 |
| jroll | it's interesting to me that tarballs.o.o isn't ssl | 23:39 |
| kfox1111 | or that. just fork the file and change the two lines. | 23:39 |
| *** imtiaz has joined #openstack-ironic | 23:40 | |
| kfox1111 | jroll: http should be jsut fine if there are relyable sha's somewhere? | 23:40 |
| jroll | full-fedora/centos/sl7-build | 23:40 |
| jroll | kfox1111: I'm not sure there is | 23:40 |
| kfox1111 | full-el7? | 23:41 |
| kfox1111 | el is usually the shortened term for rhel/centos/sl. | 23:41 |
| *** Marga_ has joined #openstack-ironic | 23:45 | |
| jroll | got it | 23:45 |
| kfox1111 | ok.. so I got an image up... I have the driver enabled... lets see... | 23:46 |
| kfox1111 | ok. gota redo the node-create... | 23:46 |
| jroll | should be able to node-update I would think? | 23:46 |
| *** imtiaz_ has joined #openstack-ironic | 23:46 | |
| kfox1111 | are all the params the same? | 23:47 |
| kfox1111 | hmm.... pxe_deploy_ramdisk/kernel or agent_deploy_ramdisk? | 23:48 |
| *** achanda has quit IRC | 23:48 | |
| jroll | just deploy_ramdisk, deploy_kernel | 23:48 |
| jroll | afaik | 23:48 |
| jroll | yep, double checked | 23:48 |
| kfox1111 | k. thx. | 23:49 |
| *** imtiaz has quit IRC | 23:51 | |
| kfox1111 | k... moment of truth.. | 23:53 |
| *** achanda has joined #openstack-ironic | 23:53 | |
| NobodyCam | totaly unreleated but /me wants this: http://www.fasthemis.com/procharger-supercharger-kit-dodge-challenger-6.1l-srt8-2008-2009-2010.aspx | 23:54 |
| jroll | om nom | 23:54 |
| kfox1111 | hmm? ironic conductor's mentioning swift. why? | 23:55 |
| NobodyCam | its a agent thing | 23:56 |
| NobodyCam | agent use swift temp urls | 23:56 |
| kfox1111 | hmm... I never setup swift. | 23:56 |
| kfox1111 | we haven't had much need of it. | 23:56 |
| kfox1111 | arg. | 23:56 |
| *** imtiaz_ has quit IRC | 23:57 | |
| jroll | oh | 23:57 |
| jroll | ohhhh. | 23:57 |
| jroll | JayF: there's the difference you missed | 23:57 |
| jroll | kfox1111: basically, it requires swift to be your glance backend | 23:57 |
| kfox1111 | yeah. :/ | 23:58 |
| kfox1111 | would a ceph gateway work? | 23:58 |
| jroll | not out of the box, idk much about ceph | 23:58 |
| kfox1111 | hmm.. probably not, since glance stores directly in ceph, not through the ceph gateway.. | 23:58 |
| jroll | basically, using the swift backend allows you to make a signed url for an image | 23:59 |
| kfox1111 | this is a major setback. | 23:59 |
| JayF | :-C | 23:59 |
| jroll | which doesn't require an auth token | 23:59 |
| jroll | yeah, sorry :( | 23:59 |
| JayF | sorry kfox1111 :(( | 23:59 |
| JayF | the agent itself supports downloading from any http endpoint | 23:59 |
| JayF | but ironic only knows how to tell it about swift temp urls | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
