| jroll | ^^ yessssssss | 00:00 |
|---|---|---|
| jroll | morning naohirot :) | 00:00 |
| naohirot | jroll: good evening :) | 00:00 |
| jroll | so excited about that blog post | 00:01 |
| * JayF <3 open hardware | 00:01 | |
| *** marcoemorais has quit IRC | 00:01 | |
| *** marcoemorais has joined #openstack-ironic | 00:01 | |
| naohirot | jroll: Yeah, I thins so | 00:01 |
| naohirot | JayF: good evening :) | 00:02 |
| *** lucas-dinner has joined #openstack-ironic | 00:07 | |
| *** Marga_ has joined #openstack-ironic | 00:15 | |
| *** Marga_ has quit IRC | 00:15 | |
| *** Marga_ has joined #openstack-ironic | 00:15 | |
| *** romcheg has quit IRC | 00:16 | |
| *** lucas-dinner has quit IRC | 00:16 | |
| *** romcheg1 has joined #openstack-ironic | 00:18 | |
| clif_h | could I get some ironic eyes on this review: https://review.openstack.org/#/c/141188/ | 00:26 |
| clif_h | mrda: I think it is ready for a real review | 00:27 |
| *** romcheg1 has quit IRC | 00:29 | |
| NobodyCam | is mrda around? | 00:31 |
| jroll | no idea | 00:31 |
| JayF | NobodyCam: you should review it too ;) | 00:32 |
| clif_h | I believe he said he had wanted to fix this bug himself, and would review my changes when the time came | 00:32 |
| NobodyCam | JayF: I can take a look. but my question was on: https://review.openstack.org/#/c/141737 | 00:33 |
| NobodyCam | lol | 00:33 |
| JayF | also NobodyCam, would love your thoughts on this: https://review.openstack.org/#/c/141957/ | 00:35 |
| JayF | not specifically on implementation as much as "are folks okay with this existing" | 00:36 |
| jroll | JoshNang: "my$awesome"'string | '"with pipes" | 00:40 |
| *** Masahiro has joined #openstack-ironic | 00:45 | |
| NobodyCam | JayF: initialy I have no objections to something like that. one could make a its also for stand alone testing argument | 00:47 |
| *** jrist is now known as jrist-afk | 00:47 | |
| NobodyCam | clif_h: ValueError: Circular reference detected | 00:48 |
| * NobodyCam notes he has not looked at the code just the error | 00:48 | |
| *** anderbubble has quit IRC | 00:49 | |
| JayF | NobodyCam: fwiw it's legitimately for testing, lol | 00:49 |
| JayF | NobodyCam: we add a new firmware to our hardware manager and want to test it? We commented out the heartbeater and lookup and just told it to flash a firmware. JoshNang thought it was a good enough idea to upstream and I agreed :) | 00:50 |
| NobodyCam | JayF: ++ | 00:51 |
| clif_h | NobodyCam: I think that's an unrelated test error that has been subsequently fixed | 00:51 |
| jroll | ^ | 00:52 |
| jroll | it's been rechecked | 00:52 |
| NobodyCam | oh that still using time.sleep isn't that a not good thing | 00:57 |
| jroll | NobodyCam: whassat? | 00:59 |
| NobodyCam | line 145 https://review.openstack.org/#/c/141188/4/nova/virt/ironic/client_wrapper.py | 00:59 |
| jroll | should that be eventlet.sleep, you're saying? | 01:00 |
| jroll | I wonder if time.sleep is monkeypatched in nova | 01:00 |
| NobodyCam | oh that could be | 01:01 |
| NobodyCam | it's after five | 01:01 |
| NobodyCam | :-p | 01:01 |
| jroll | ^^ | 01:01 |
| *** mjturek has quit IRC | 01:05 | |
| *** anderbubble has joined #openstack-ironic | 01:10 | |
| *** penick has quit IRC | 01:11 | |
| *** penick has joined #openstack-ironic | 01:15 | |
| *** igordcard has quit IRC | 01:18 | |
| *** igordcard has joined #openstack-ironic | 01:20 | |
| *** killer_prince is now known as lazy_prince | 01:20 | |
| openstackgerrit | Naohiro Tamura proposed openstack/python-ironicclient: Removed http proxy environment variable so that httpretty can work https://review.openstack.org/142112 | 01:30 |
| *** penick has quit IRC | 01:39 | |
| naohirot | Haomeng|2: ^^^ thanks for review | 01:41 |
| Haomeng|2 | naohirot: yw:) | 01:41 |
| *** marcoemorais has quit IRC | 01:41 | |
| Haomeng|2 | naohirot: :) | 01:41 |
| naohirot | Haomeng|2: :) | 01:41 |
| openstackgerrit | Merged openstack/ironic-python-agent: Use oslo.utils and oslo.concurrency https://review.openstack.org/138116 | 01:44 |
| *** mitz has quit IRC | 01:47 | |
| *** mitz has joined #openstack-ironic | 01:52 | |
| *** kes has joined #openstack-ironic | 01:56 | |
| *** igordcard has quit IRC | 02:00 | |
| openstackgerrit | Ruby Loo proposed openstack/ironic: Issue with FSM's _target_state https://review.openstack.org/142303 | 02:00 |
| rloo | devananda: I looked at the fsm patches and I still think there's a problem. I added patch 142303 | 02:04 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add standalone mode for IPA https://review.openstack.org/141957 | 02:05 |
| rloo | devananda: some unit tests in patch 142303 fail cuz the fsm's _target_state doesn't get updated when it should (or at least when we want the node's target_pstate to get updated) | 02:05 |
| *** anderbubble has quit IRC | 02:09 | |
| *** yjiang5 is now known as yjiang5_away | 02:09 | |
| *** ChuckC has quit IRC | 02:18 | |
| *** dlaube has quit IRC | 02:34 | |
| *** mitz has quit IRC | 02:37 | |
| *** mitz has joined #openstack-ironic | 02:37 | |
| *** ramineni has joined #openstack-ironic | 02:43 | |
| *** anderbubble has joined #openstack-ironic | 02:49 | |
| *** rloo has quit IRC | 02:49 | |
| *** Marga_ has quit IRC | 02:52 | |
| *** Marga_ has joined #openstack-ironic | 02:53 | |
| *** arif-ali has quit IRC | 02:58 | |
| *** arif-ali has joined #openstack-ironic | 03:01 | |
| *** ChuckC has joined #openstack-ironic | 03:03 | |
| *** lazy_prince is now known as killer_prince | 03:04 | |
| *** kes has quit IRC | 03:06 | |
| *** kes has joined #openstack-ironic | 03:09 | |
| *** Krast has quit IRC | 03:13 | |
| *** chenglch has joined #openstack-ironic | 03:15 | |
| *** datajerk has quit IRC | 03:19 | |
| mrda | jroll, clif_h: I'll review this now. | 03:20 |
| *** kesmellon has joined #openstack-ironic | 03:23 | |
| *** naohirot has quit IRC | 03:24 | |
| *** harlowja is now known as harlowja_away | 03:25 | |
| *** kes has quit IRC | 03:26 | |
| *** PaulCzar has quit IRC | 03:36 | |
| *** ryanpetrello has quit IRC | 03:39 | |
| lintan | rloo:Hi | 03:40 |
| *** chen12 has quit IRC | 03:41 | |
| *** jerryz has joined #openstack-ironic | 03:42 | |
| *** Nisha has joined #openstack-ironic | 03:47 | |
| *** mitz has quit IRC | 03:50 | |
| *** chenglch has quit IRC | 03:51 | |
| *** mitz has joined #openstack-ironic | 03:59 | |
| *** naohirot has joined #openstack-ironic | 04:03 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/142334 | 04:05 |
| *** mitz has quit IRC | 04:15 | |
| *** mitz has joined #openstack-ironic | 04:16 | |
| mrda | JayF: I am looking at it right now :) | 04:18 |
| *** ChuckC_ has joined #openstack-ironic | 04:19 | |
| *** ChuckC has quit IRC | 04:20 | |
| *** datajerk has joined #openstack-ironic | 04:28 | |
| *** ryanpetrello has joined #openstack-ironic | 04:36 | |
| *** Masahiro has quit IRC | 04:39 | |
| *** Masahiro has joined #openstack-ironic | 04:40 | |
| *** ZZYang has joined #openstack-ironic | 04:42 | |
| *** mitz has quit IRC | 04:42 | |
| *** yuanying_ has quit IRC | 04:43 | |
| *** yuanying has joined #openstack-ironic | 04:43 | |
| *** ZZhangYang has quit IRC | 04:45 | |
| *** mitz has joined #openstack-ironic | 04:45 | |
| *** Shrews has quit IRC | 04:59 | |
| *** Marga_ has quit IRC | 05:01 | |
| *** anderbubble has quit IRC | 05:03 | |
| *** pensu has joined #openstack-ironic | 05:10 | |
| *** mitz has quit IRC | 05:12 | |
| *** mitz has joined #openstack-ironic | 05:13 | |
| *** anderbubble has joined #openstack-ironic | 05:14 | |
| *** mitz has quit IRC | 05:14 | |
| *** mitz has joined #openstack-ironic | 05:16 | |
| *** mitz has quit IRC | 05:20 | |
| *** mitz has joined #openstack-ironic | 05:21 | |
| *** ryanpetrello has quit IRC | 05:22 | |
| *** mitz has quit IRC | 05:31 | |
| *** mitz has joined #openstack-ironic | 05:34 | |
| *** anderbubble has quit IRC | 05:39 | |
| *** anderbubble has joined #openstack-ironic | 05:41 | |
| *** mitz has quit IRC | 05:51 | |
| *** mitz has joined #openstack-ironic | 05:52 | |
| *** dlpartain has joined #openstack-ironic | 05:57 | |
| *** mitz has quit IRC | 06:00 | |
| *** mitz has joined #openstack-ironic | 06:00 | |
| *** achanda has joined #openstack-ironic | 06:01 | |
| *** mitz has quit IRC | 06:03 | |
| *** killer_prince is now known as lazy_prince | 06:09 | |
| *** dlpartain has quit IRC | 06:17 | |
| *** achanda has quit IRC | 06:40 | |
| *** achanda has joined #openstack-ironic | 06:41 | |
| *** achanda_ has joined #openstack-ironic | 06:45 | |
| *** achanda has quit IRC | 06:45 | |
| *** achanda_ has quit IRC | 06:48 | |
| *** marcoemorais has joined #openstack-ironic | 06:52 | |
| *** marcoemorais1 has joined #openstack-ironic | 06:54 | |
| *** marcoemorais has quit IRC | 06:56 | |
| *** Shrews has joined #openstack-ironic | 06:59 | |
| *** mrda is now known as mrda-away | 07:00 | |
| *** anderbubble has quit IRC | 07:01 | |
| *** ryanpetrello has joined #openstack-ironic | 07:08 | |
| *** ryanpetrello has quit IRC | 07:13 | |
| *** pcaruana has joined #openstack-ironic | 07:16 | |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Cisco Driver, Update after review changes https://review.openstack.org/139517 | 07:21 |
| *** achanda has joined #openstack-ironic | 07:35 | |
| *** achanda has quit IRC | 07:55 | |
| *** achanda has joined #openstack-ironic | 07:56 | |
| *** dlpartain has joined #openstack-ironic | 07:57 | |
| *** chen has joined #openstack-ironic | 07:58 | |
| *** achanda has quit IRC | 08:01 | |
| *** achanda has joined #openstack-ironic | 08:02 | |
| *** dlpartain has quit IRC | 08:02 | |
| *** dlpartain has joined #openstack-ironic | 08:03 | |
| *** pcrews has quit IRC | 08:04 | |
| *** ifarkas has joined #openstack-ironic | 08:06 | |
| *** Nisha has quit IRC | 08:07 | |
| *** pcrews has joined #openstack-ironic | 08:07 | |
| *** Nisha has joined #openstack-ironic | 08:08 | |
| *** pcrews has quit IRC | 08:13 | |
| *** dlpartain has left #openstack-ironic | 08:26 | |
| *** jcoufal has joined #openstack-ironic | 08:30 | |
| *** erwan_taf has joined #openstack-ironic | 08:47 | |
| *** rameshg87 has joined #openstack-ironic | 08:51 | |
| *** ChuckC_ is now known as ChuckC | 08:52 | |
| *** ryanpetrello has joined #openstack-ironic | 08:57 | |
| *** Masahiro has quit IRC | 09:00 | |
| *** andreykurilin_ has joined #openstack-ironic | 09:00 | |
| *** Masahiro has joined #openstack-ironic | 09:00 | |
| *** ryanpetrello has quit IRC | 09:01 | |
| *** Masahiro has quit IRC | 09:05 | |
| *** kesmellon has quit IRC | 09:06 | |
| naohirot | rameshg87: Hi good morning | 09:12 |
| *** Masahiro has joined #openstack-ironic | 09:12 | |
| rameshg87 | hello naohirot | 09:12 |
| naohirot | rameshg87: did you have a time to look at my gerrit reply? | 09:13 |
| rameshg87 | naohirot, just saw it | 09:13 |
| rameshg87 | naohirot, we did change dib: https://review.openstack.org/#/q/I28ca888b4ead4905f7141b38ed4b9a3614d00d26,n,z | 09:13 |
| * naohirot read url | 09:14 | |
| rameshg87 | naohirot, that code is now refactored, but you can find the code in dib here now: https://github.com/openstack/diskimage-builder/blob/master/elements/ramdisk-base/extra-data.d/scripts/d/init-func#L16-L46 | 09:15 |
| rameshg87 | naohirot, basically we modified the deploy ramdisk to mount the virtual floppy and read the parameters from it (in addition to reading the kernel command line arguments) | 09:15 |
| *** derekh has joined #openstack-ironic | 09:16 | |
| *** marcoemorais1 has quit IRC | 09:16 | |
| naohirot | rameshg87: Okay, basically do we need blue print and spec to modify DIB by working with TripleO team? | 09:16 |
| rameshg87 | naohirot, you might not need a spec | 09:17 |
| rameshg87 | naohirot, for tripleo | 09:17 |
| rameshg87 | naohirot, you can just refer to Ironic change (to tell them why you are changing something) | 09:17 |
| naohirot | rameshg87: I see. | 09:17 |
| rameshg87 | naohirot, but you will need to make the nfs/cifs read/write and be able to upload the generated image there | 09:17 |
| rameshg87 | naohirot, to pass the deploy information | 09:18 |
| naohirot | rameshg87: do you mean that I need to modify DIB? | 09:18 |
| rameshg87 | naohirot, i don't know if you will need to modify dib | 09:19 |
| rameshg87 | naohirot, we find the virtual media floppy device by grepping for 'virtual media' as model in sysfs in linux kernel | 09:19 |
| naohirot | rameshg87: Okay I got your point. I have to set the NFS/CIFS to be read/write mode. | 09:20 |
| rameshg87 | naohirot, that's how virtual media in our proliant machines are visible to the linux kernel | 09:20 |
| rameshg87 | naohirot, may be you might just need to grep for something else - the name which your servers give the model name | 09:20 |
| *** romcheg has joined #openstack-ironic | 09:22 | |
| rameshg87 | naohirot, https://github.com/openstack/diskimage-builder/blob/master/elements/ramdisk-base/extra-data.d/scripts/d/init-func#L21-L22 | 09:22 |
| naohirot | rameshg87: who is grepping for 'virtual media', the shell script in DIB? I believe | 09:22 |
| rameshg87 | naohirot, ^^ the code which does that | 09:23 |
| naohirot | rameshg87: Okey, but why can't iLO driver create the boot image in case of UEFI? | 09:24 |
| rameshg87 | naohirot, we couldn't complete everything in juno timeframe | 09:24 |
| rameshg87 | naohirot, we are planning to do that in kilo | 09:24 |
| rameshg87 | naohirot, there is nothing technically blocking | 09:25 |
| rameshg87 | naohirot, Nisha from my team is working on that: https://review.openstack.org/#/c/134022/ | 09:25 |
| naohirot | rameshg87: in order to create the boot image for UEFI by the command "disk-image-create ubuntu baremetal iso" | 09:26 |
| naohirot | rameshg87: why isn't there no information about UEFI in command line? | 09:27 |
| naohirot | rameshg87: I'm referring to the bottom of http://docs.openstack.org/developer/ironic/drivers/ilo.html | 09:28 |
| *** achanda has quit IRC | 09:28 | |
| naohirot | rameshg87: If I created a boot image for Legacy BIOS by the command, is the command line different? | 09:29 |
| * naohirot reading https://review.openstack.org/#/c/134022/4/specs/kilo/automate-uefi-bios-iso-creation.rst | 09:30 | |
| *** achanda has joined #openstack-ironic | 09:34 | |
| rameshg87 | naohirot, if you create boot image by "disk-image-create ubuntu baremetal iso" it's bootable on both bios and uefi machines | 09:35 |
| rameshg87 | naohirot, since it works on both, it's not mentioned explicitly :) | 09:35 |
| rameshg87 | naohirot, brb | 09:35 |
| *** rameshg87 is now known as rameshg87-brb | 09:35 | |
| naohirot | rameshg87-brb: I see, Okay. I learned that what brb means :-) | 09:37 |
| *** MattMan has quit IRC | 09:41 | |
| *** MattMan has joined #openstack-ironic | 09:42 | |
| naohirot | rameshg87-brb: It seems that I have to investigate what the device name of iRMC virtual floppy under sys/fs. | 09:47 |
| *** rameshg87-brb is now known as rameshg87 | 09:52 | |
| rameshg87 | naohirot, yes | 09:52 |
| rameshg87 | naohirot, and you will need to copy the generated floppy images to your nfs/cifs and keep them read/write | 09:52 |
| naohirot | rameshg87: yes | 09:53 |
| naohirot | rameshg87: I typed "ls /sys/class/block/*/device/model" on Ubuntu | 09:53 |
| naohirot | rameshg87: then I got "/sys/class/block/sda/device/model /sys/class/block/sr0/device/model | 09:54 |
| naohirot | " | 09:54 |
| naohirot | rameshg87: In case of iLO, what is the value of * part? | 09:54 |
| naohirot | rameshg87: virtual media? | 09:55 |
| rameshg87 | naohirot, it's about finding the virtual media device file | 09:57 |
| rameshg87 | naohirot, it can be any of sdx in case of proliant | 09:57 |
| naohirot | rameshg87: is 'x' number? | 09:58 |
| rameshg87 | naohirot, /dev/sda, /dev/sdb, etc | 09:58 |
| naohirot | rameshg87: Aha, so the difference come form PC model, not vendor, right? | 09:59 |
| rameshg87 | naohirot, i guess both :) | 09:59 |
| naohirot | rameshg87: s/come form/comes from/ | 09:59 |
| naohirot | rameshg87: If so, I think I can use the script as is, but I have to try. | 10:00 |
| rameshg87 | naohirot, yeah | 10:01 |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic-specs: Support Zapping for iLO Nodes https://review.openstack.org/142401 | 10:01 |
| rameshg87 | naohirot, but you will might need to grep for a different model i guess | 10:01 |
| *** achanda has quit IRC | 10:01 | |
| naohirot | rameshg87: Yes, I'll do that. | 10:02 |
| *** achanda has joined #openstack-ironic | 10:02 | |
| naohirot | rameshg87: So I believe that the virtual media deploy spec needs to change from read-only NFS/CIFS to read/write. This is the most important. | 10:03 |
| *** lifeless_ is now known as lifeless | 10:04 | |
| rameshg87 | naohirot, yeah | 10:04 |
| naohirot | rameshg87: I believe rest of the spec remain if I follows same implementation as iLO. | 10:04 |
| rameshg87 | naohirot, and how will you copy the files to this nfs/cifs ? | 10:05 |
| naohirot | rameshg87: s/follows/follow/ | 10:05 |
| naohirot | rameshg87: That's important point too. I think Ironic conductor mounts the NFS/CIFS. | 10:06 |
| *** achanda has quit IRC | 10:07 | |
| rameshg87 | naohirot, yeah may be you can just copy the files then | 10:07 |
| rameshg87 | naohirot, from within the conductor node | 10:07 |
| naohirot | rameshg87: Okay, that is the story we discussed before, ironic conductor become NFS server. right? | 10:08 |
| rameshg87 | naohirot, hmm yeah. but in-short conductor doesn't need to worry about this i guess | 10:08 |
| naohirot | rameshg87: What do you mean by "you can just copy the file" from where to where? | 10:10 |
| rameshg87 | naohirot, we generate the vfat image in the ironic conductor which is to be mounted on the server as virtual floppy | 10:11 |
| rameshg87 | naohirot, now the conductor needs to expose this vfat image as virtual floppy | 10:11 |
| naohirot | rameshg87: Yes, so the conductor becomes NFS server from iRMC's point of view, right? | 10:12 |
| rameshg87 | naohirot, yes | 10:12 |
| rameshg87 | naohirot, there can be a config nfs_root which is the directory to which conductor should copy these files to make it availble to the the bare metal | 10:12 |
| rameshg87 | naohirot, just like the current tftp_root in pxe driver or http_root in ipxe | 10:13 |
| naohirot | rameshg87: I see | 10:14 |
| *** Nisha_away has joined #openstack-ironic | 10:14 | |
| *** cuihaozhi has quit IRC | 10:15 | |
| *** cuihaozhi has joined #openstack-ironic | 10:15 | |
| naohirot | rameshg87: If I copy floppy image or boot image into nfs_root in the conductor | 10:17 |
| *** Nisha has quit IRC | 10:17 | |
| naohirot | rameshg87: you mentioned before about file based lock is necessary, but not memory based. Am I correct? | 10:18 |
| rameshg87 | naohirot, i guess it was devananda who mentioned it | 10:18 |
| rameshg87 | naohirot, but that was if different conductors shared it | 10:18 |
| rameshg87 | naohirot, but in your case only per-conductor, right ? | 10:19 |
| naohirot | rameshg87: right now I'm not sure because I didn't fully understand what kind of mechanize the ring provides. | 10:20 |
| naohirot | rameshg87: Is the ring some kind of P2P among conductors? | 10:21 |
| rameshg87 | naohirot, do you mean hash ring ? | 10:21 |
| *** Nisha_away has quit IRC | 10:23 | |
| naohirot | rameshg87: Yes, I thinks so, plurals of conductors create hash ring right? | 10:23 |
| rameshg87 | naohirot, yes, but i don't think it affects hash ring | 10:23 |
| rameshg87 | naohirot, you could have one nfs share on one conductor and avoid issues with multiple-conductor | 10:24 |
| rameshg87 | naohirot, just like tftp_root is managed per conductor currently | 10:24 |
| naohirot | rameshg87: I see | 10:24 |
| naohirot | rameshg87: If deployer type "ironic create node" in multiple-conductor configuration, is there any way to know where the node created? | 10:27 |
| rameshg87 | naohirot, the node is actually created in db :) | 10:28 |
| rameshg87 | naohirot, the node is just being managed by a particular conductor and that is defined by hash ring | 10:28 |
| rameshg87 | naohirot, if the conductor goes down, the node is taken over by some other conductor | 10:28 |
| naohirot | rameshg87: that's what I thought, distributed hash ring typically does, I believe. | 10:30 |
| *** Masahiro has quit IRC | 10:32 | |
| naohirot | rameshg87: So that means node A is assigned to conductor B today, the node A might be assigned to conductor B tomorrow, right? | 10:34 |
| naohirot | rameshg87: So that means node A is assigned to conductor B today, the node A might be assigned to conductor *C* tomorrow, right? | 10:35 |
| * naohirot I'm really confused | 10:36 | |
| * naohirot I believe the hash ring is created automatically , but not defined by human manually. | 10:38 | |
| *** athomas has quit IRC | 10:39 | |
| *** alexpilotti has joined #openstack-ironic | 10:41 | |
| *** rakesh_hs has joined #openstack-ironic | 10:42 | |
| rameshg87 | naohirot, yes | 10:45 |
| rameshg87 | naohirot, that's why we have to take_over method in deploy which is for a conductor to do stuffs while taking over a node from another conductor | 10:46 |
| *** ryanpetrello has joined #openstack-ironic | 10:46 | |
| naohirot | rameshg87: Okay, you mean that there is a hook if taking over starts. right? | 10:47 |
| rameshg87 | naohirot, yes | 10:47 |
| rameshg87 | naohirot, https://github.com/openstack/ironic/blob/master/ironic/drivers/base.py#L199-L216 | 10:48 |
| *** ryanpetrello has quit IRC | 10:51 | |
| naohirot | rameshg87: did ironic team made a consensus in Juno time frame that implementing of the take_over function is not necessary as the first step or something? | 10:52 |
| naohirot | rameshg87: s/made/make/ | 10:52 |
| rameshg87 | naohirot, well that depends | 10:53 |
| naohirot | rameshg87: Is iLO team implementing it in Kilo? | 10:53 |
| rameshg87 | naohirot, there is a *no* to things that doesn't take multi-conductor environment into account | 10:54 |
| rameshg87 | naohirot, but may be you can do without it as a first-step, and enhance it later on | 10:54 |
| rameshg87 | naohirot, for iLO drivers, take_over is not required as we do http with Swift | 10:54 |
| rameshg87 | naohirot, and there is nothing that is conductor-dependent with the deploy mechanism | 10:54 |
| rameshg87 | naohirot, you could start the deploy with an iLO driver on conductor A and continue deploy from conductor B | 10:55 |
| *** lucasagomes has joined #openstack-ironic | 10:55 | |
| naohirot | rameshg87: I see, in order to avoid complicated thing in the first step, I'd like to put all images into NFS/CIFS which deployer prepared, but not into the conductor. | 10:56 |
| rameshg87 | naohirot, okay. so what are these *images* ? | 10:58 |
| *** ramineni has quit IRC | 10:58 | |
| naohirot | rameshg87: I meant the deploy iso image, floppy image, and the boot iso image. | 10:58 |
| rameshg87 | naohirot, okay | 10:59 |
| rameshg87 | naohirot, deploy iso image = prepared by deployer and deployer keeps it in nfs/cifs share, right ? | 10:59 |
| *** dlpartain has joined #openstack-ironic | 10:59 | |
| naohirot | rameshg87: Yes, exactly. | 11:00 |
| rameshg87 | naohirot, floppy image = prepared by conductor during deploy. how does conductor *put* image in nfs/cifs ? | 11:00 |
| naohirot | rameshg87: I thinking the conductor mount the NIF/CFFS which deployer prepared, so conductor just can copy it. | 11:01 |
| naohirot | rameshg87: then iRMC mounts the floppy image. | 11:02 |
| rameshg87 | naohirot, sounds good | 11:02 |
| rameshg87 | naohirot, same for generated *boot iso image* as well, correct ? | 11:03 |
| naohirot | rameshg87: yes. | 11:03 |
| rameshg87 | naohirot, it should be upto deployer to mount nfs/cifs and conductor needn't manage it | 11:03 |
| rameshg87 | naohirot, deployer just need to let conductor know where it is mounted in the conductor node | 11:03 |
| rameshg87 | naohirot, something like nfs_root = /path/to/where/nfs/is/mounted | 11:04 |
| rameshg87 | naohirot, correct ? | 11:04 |
| *** Jatin360 has joined #openstack-ironic | 11:04 | |
| *** athomas has joined #openstack-ironic | 11:05 | |
| naohirot | rameshg87: yes, those information for NIF/CIFS client is stored in driver_info, so the conductor knows about it. | 11:05 |
| rameshg87 | naohirot, it can be in ironic.conf i guess | 11:06 |
| rameshg87 | naohirot, driver_info contain node-related settings | 11:06 |
| rameshg87 | naohirot, and nfs_root is a conductor-related setting | 11:06 |
| rameshg87 | naohirot, i will need to leave for home now. shall we continue this discussion later | 11:06 |
| naohirot | rameshg87: Yes, part of information can be found in the conf file, but not all of them. | 11:06 |
| rameshg87 | naohirot, may be you can add these information as well into spec and i will have a look at it | 11:07 |
| naohirot | rameshg87: thank you for sparing your time, I'll update the spec first, and please let me know if the updated one is okay or not. | 11:09 |
| rameshg87 | naohirot, sure .. | 11:09 |
| rameshg87 | naohirot, see you later, bye | 11:09 |
| *** rameshg87 has quit IRC | 11:09 | |
| naohirot | see you, bye! | 11:09 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic-specs: Root device hints https://review.openstack.org/138729 | 11:14 |
| openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic-specs: Add a new driver for Fuel Agent https://review.openstack.org/138115 | 11:14 |
| *** andreykurilin_ has quit IRC | 11:18 | |
| *** Jatin360 has quit IRC | 11:26 | |
| *** Masahiro has joined #openstack-ironic | 11:33 | |
| *** Masahiro has quit IRC | 11:38 | |
| *** naohirot has quit IRC | 11:43 | |
| afaranha | Helo, I'm followed this http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html#deploying-ironic-with-devstack to test ironic, and worked fine. Now, I have two available machines, and I would like to use them to launch servers as bare metal nodes, how can I do this? | 11:56 |
| afaranha | If I change this variables in the local.conf file: NETWORK_GATEWAY, FIXED_RANGE, FIXED_NETWORK; to stays in the same network as the machines, will it work? | 11:57 |
| *** dlpartain has quit IRC | 12:04 | |
| *** dlpartain has joined #openstack-ironic | 12:04 | |
| *** jiangfei has quit IRC | 12:29 | |
| *** jiangfei has joined #openstack-ironic | 12:29 | |
| *** ryanpetrello has joined #openstack-ironic | 12:35 | |
| *** rakesh_hs has quit IRC | 12:37 | |
| *** ryanpetrello has quit IRC | 12:39 | |
| *** ZZYang has quit IRC | 12:48 | |
| *** lazy_prince is now known as killer_prince | 12:49 | |
| *** Masahiro has joined #openstack-ironic | 12:49 | |
| *** ZhangYang has joined #openstack-ironic | 12:49 | |
| openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic: Improve Agent deploy driver validation https://review.openstack.org/141766 | 12:51 |
| afaranha | devananda, I use the configuration for PXE on the guide, but I'm not sure how to do it work with physical machines, where do I config the pxe_ipmitool driver ? | 12:51 |
| *** dprince has joined #openstack-ironic | 12:52 | |
| *** Masahiro has quit IRC | 12:54 | |
| *** vipul has quit IRC | 13:05 | |
| Shrews | lucasagomes: morning! re: 139374... i'm pretty sure that was just a global search/replace by infra to update links in all projects. We'll probably want to assist Jeremy and make a patch set to make it look the way we want. I'll take care of that for us today. | 13:06 |
| lucasagomes | Shrews, oh I see | 13:07 |
| Shrews | jeremy == fungi in #-infra | 13:07 |
| lucasagomes | Shrews, hmm, alright so let's merged that and put a following patch | 13:07 |
| Shrews | lucasagomes: yeah. i can fix that up quick | 13:08 |
| lucasagomes | Shrews, sounds good, well let's ping fungi first see if he will fix that up | 13:08 |
| lucasagomes | Shrews, thanks for letting me know | 13:08 |
| lucasagomes | usually I identify -infra stuff via the openstack-bot | 13:09 |
| Shrews | lucasagomes: i'll try to get in touch with him. a bit early for him probably | 13:10 |
| *** vipul has joined #openstack-ironic | 13:10 | |
| lucasagomes | Shrews, cool yeah I just pinged him at -infra | 13:10 |
| lucasagomes | but I'm OK on merging it and fixing with a following patch if we can't get in touch | 13:11 |
| Shrews | let's give him an hour or so | 13:11 |
| lucasagomes | sounds good :) | 13:12 |
| lucasagomes | afaranha, if they are in the same network I believe it will work. You can configure pxe_ipmitool driver via ironic.conf when it's a more global config, or you can pass parameters to node's driver_info attribute when it's node specific configuration | 13:14 |
| lintan | lucasagomes:Hi | 13:17 |
| lucasagomes | lintan, hi there | 13:18 |
| lintan | lucasagomes:I wonder do you have some review bandwidth to help me on amt driver, it's still target for kilo-1 as we discuss in meeting | 13:19 |
| lucasagomes | lintan, oh def, last week I was on holidays so i haven't reviewed much | 13:19 |
| lucasagomes | I will fo sure | 13:19 |
| lucasagomes | lintan, the spec is merged and all? | 13:19 |
| lintan | lucasagomes:Thanks, My code is ready but rloo disagree with some part. So I have a discussion with and he suggest I update the spec. | 13:21 |
| lucasagomes | lintan, right I will take a look soonish | 13:21 |
| lucasagomes | just finishing a quick thing here first | 13:22 |
| lintan | lucasagomes:Thanks here is the spec https://review.openstack.org/#/c/141269/ | 13:22 |
| *** kevinbenton has quit IRC | 13:23 | |
| *** kevinbenton has joined #openstack-ironic | 13:25 | |
| *** rameshg87 has joined #openstack-ironic | 13:25 | |
| rameshg87 | ifarkas, hi | 13:26 |
| ifarkas | rameshg87, hi | 13:26 |
| rameshg87 | ifarkas, got some 10 mins to discuss the comments on the raid spec ? | 13:26 |
| ifarkas | rameshg87, sure | 13:26 |
| ifarkas | rameshg87, did you see my comments? | 13:27 |
| rameshg87 | ifarkas, yeah, i had replied to them actually | 13:27 |
| ifarkas | rameshg87, hah, then I am going to check them! | 13:27 |
| rameshg87 | ifarkas, :) | 13:27 |
| rameshg87 | ifarkas, regarding the wwid, it's actually wwn that hp's tool returns | 13:28 |
| rameshg87 | ifarkas, it's seen as /dev/disk/by-id/wwn-0x600508b1001ce4acf473ee9c826230ff by the linux kernel | 13:28 |
| ifarkas | rameshg87, right. that still doesn't really help because drac doesn't know anything about that as well | 13:28 |
| rameshg87 | ifarkas, do you see the same for FQDD as well ? | 13:28 |
| ifarkas | rameshg87, no | 13:29 |
| ifarkas | rameshg87, let me paste you the output of a virtual disk query | 13:29 |
| rameshg87 | ifarkas, here is mine for a logical disk: http://paste.openstack.org/show/152267/ | 13:29 |
| *** naohirot has joined #openstack-ironic | 13:30 | |
| ifarkas | rameshg87, here's drac: http://paste.openstack.org/show/152268/ | 13:30 |
| rameshg87 | ifarkas, the "Unique Identifier" is what is seen as wwn in /dev/disk/by-id/wwn-xxxxxx | 13:30 |
| *** pensu has quit IRC | 13:31 | |
| rameshg87 | ifarkas, hmm i see | 13:31 |
| rameshg87 | ifarkas, then how do identify this disk in linux ? | 13:31 |
| ifarkas | rameshg87, I was thinking about identifying via it's physical disks | 13:32 |
| ifarkas | rameshg87, although I haven't tried it yet | 13:32 |
| ifarkas | rameshg87, on the paste you can see, it lists the associated physical disks | 13:32 |
| rameshg87 | ifarkas, i mean we should in some way ask Ironic to deploy to the disk marked by the operator as "boot_volume" | 13:32 |
| rameshg87 | ifarkas, physical disks aren't directly visible to the linux kernel because they are behind raid controller, right ? | 13:34 |
| ifarkas | rameshg87, hmm, not sure | 13:35 |
| ifarkas | rameshg87, I haven't actually tried it yet, what can be queried this way | 13:35 |
| ifarkas | rameshg87, but maybe you are right | 13:35 |
| ifarkas | rameshg87, the problem is that the drac only gives back this much information | 13:35 |
| rameshg87 | ifarkas, i see name in your case | 13:36 |
| ifarkas | rameshg87, yes, which can also be used internally by the driver | 13:36 |
| ifarkas | rameshg87, but still, I don't know if that's exposed to the linux kernel, or it is something internal to drac | 13:36 |
| *** wanghua has quit IRC | 13:36 | |
| rameshg87 | ifarkas, yeah i don't think it is exposed to linux kernel | 13:37 |
| *** mjturek has joined #openstack-ironic | 13:37 | |
| rameshg87 | ifarkas, spec by lucasagomes covers these: https://review.openstack.org/#/c/138729/7/specs/kilo/root-device-hints.rst | 13:37 |
| rameshg87 | ifarkas, uuid, model, vendor, serial, wwn, hctl, size | 13:38 |
| rameshg87 | ifarkas, we can fall back to size of the disk as the last alternative | 13:38 |
| ifarkas | rameshg87, that would be quite error prone, but yeah | 13:39 |
| rameshg87 | ifarkas, but if the operator wanted 200 gb raid 1 and 200 gb raid 5 - then size becomes a problem | 13:39 |
| rameshg87 | ifarkas, yeah | 13:39 |
| rameshg87 | ifarkas, may be hwinfo command in linux can help you as well | 13:40 |
| rameshg87 | lucasagomes, are you there ? | 13:40 |
| ifarkas | rameshg87, yes, I need to figure that out | 13:40 |
| rameshg87 | ifarkas, okay for the moment shall i keep wwn in the spec ? | 13:41 |
| lucasagomes | rameshg87, hi there yes | 13:41 |
| lucasagomes | sorry a bit afk | 13:41 |
| rameshg87 | ifarkas, may be i will include that the driver will post one of the root_device hint as the raid configuration | 13:41 |
| ifarkas | rameshg87, no, the spec shouldn't contain wwn for now if it's meant to be generic | 13:41 |
| rameshg87 | lucasagomes, are you planning to use hwinfo for the root device hint related work ? | 13:42 |
| rameshg87 | lucasagomes, or is it by what is available already in linux kernel ? | 13:42 |
| rameshg87 | ifarkas, okay, i will keep it wwn for now | 13:42 |
| rameshg87 | ifarkas, regarding sync vs async in create_raid | 13:42 |
| ifarkas | rameshg87, I just proposed the other way :-) | 13:42 |
| lucasagomes | rameshg87, for gathering the information ? I use a combinatio of sysfs and lsblk | 13:43 |
| lucasagomes | but yeah I could use hwinfo if there's some information needed from that not available on the other tools | 13:43 |
| rameshg87 | lucasagomes, okay | 13:43 |
| lucasagomes | rameshg87, the spec covers more the ironic case, it's up to the ramdisk implement how to get the data | 13:43 |
| lucasagomes | ironic just pass the hints to it | 13:43 |
| rameshg87 | lucasagomes, ah okay | 13:43 |
| lucasagomes | rameshg87, I got something implemented for the default ramdisk, lemme give you the link | 13:44 |
| * lucasagomes searchs | 13:44 | |
| rameshg87 | lucasagomes, ifarkas and myself are planning to use it after raid configuration | 13:44 |
| lucasagomes | rameshg87, +1! | 13:44 |
| lucasagomes | sounds good | 13:44 |
| lucasagomes | rameshg87, https://review.openstack.org/#/c/142178/ | 13:44 |
| lucasagomes | rameshg87, if there's anything needed for the RAID, like a specific hint you need to pass feel free to comment on the spec and I will add it | 13:45 |
| rameshg87 | lucasagomes, great .. | 13:45 |
| rameshg87 | ifarkas, may be you can check the above as well if you need something for drac hardware | 13:45 |
| rameshg87 | lucasagomes, ifarkas, i can use wwn for hp hardware | 13:45 |
| *** jcoufal_ has joined #openstack-ironic | 13:47 | |
| ifarkas | rameshg87, I will | 13:47 |
| rameshg87 | lucasagomes, thanks, will let you know if we need some help :) | 13:48 |
| rameshg87 | ifarkas, and regarding the sync vs async | 13:48 |
| rameshg87 | ifarkas, how about adding a method to the base class RaidInterface something like update_raid_info() | 13:48 |
| rameshg87 | ifarkas, which the create can call once it is done creating the raid configuration | 13:49 |
| rameshg87 | ifarkas, does that sound good ? | 13:49 |
| rameshg87 | ifarkas, i could use the same for inband raid configuration too | 13:49 |
| lucasagomes | cool thank you! | 13:49 |
| ifarkas | rameshg87, yeah, sound good to me | 13:49 |
| rameshg87 | ifarkas, okay i will update the spec and then post for review | 13:49 |
| *** viktors|afk is now known as viktors | 13:50 | |
| ifarkas | rameshg87, since it's completely async in case of drac, the driver needs to pool the job status | 13:50 |
| rameshg87 | ifarkas, okay | 13:50 |
| *** jcoufal has quit IRC | 13:50 | |
| ifarkas | rameshg87, when it's done, it can call the update_raid_info method | 13:50 |
| rameshg87 | ifarkas, and for me i need to wait till ramdisk is booted :) | 13:50 |
| rameshg87 | ifarkas, i mean till ipa heartbeats back | 13:50 |
| *** ZhangYang has quit IRC | 13:51 | |
| *** ZhangYang has joined #openstack-ironic | 13:51 | |
| rameshg87 | ifarkas, okay sounds good | 13:51 |
| rameshg87 | ifarkas, i will update the spec | 13:51 |
| openstackgerrit | Tan Lin proposed openstack/ironic: Add AMT-PXE-Driver to deploy cloud on PC https://review.openstack.org/135184 | 13:52 |
| ifarkas | rameshg87, thanks | 13:52 |
| *** datajerk has quit IRC | 14:02 | |
| *** rloo has joined #openstack-ironic | 14:07 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-specs: New driver interface for RAID configuration https://review.openstack.org/135899 | 14:13 |
| *** lintan_ has joined #openstack-ironic | 14:13 | |
| rameshg87 | ifarkas, ^^^ | 14:13 |
| lintan | rloo:Hi morning rloo | 14:14 |
| rameshg87 | ifarkas, will come back later | 14:14 |
| *** rameshg87 has quit IRC | 14:14 | |
| rloo | hi (evening?) lintan | 14:14 |
| *** lintan_ has quit IRC | 14:14 | |
| lintan | rloo: :) yes | 14:14 |
| lintan | rloo: do you have some review bandwidth today for amt? | 14:15 |
| rloo | lintan: I'll try. There are a bunch of reviews I should look at, but I've been focussing on the fsm patches. | 14:15 |
| lintan | rloo: OK, thanks. I reply your comment to address your concern in the spec, tell me if you have any questions | 14:18 |
| *** datajerk has joined #openstack-ironic | 14:18 | |
| rloo | lintan: I think I saw that. Did you update the spec, or just reply? | 14:19 |
| lintan | rloo: just reply | 14:19 |
| lintan | rloo: Not update yet, if we agree with use instance_info instead driver_info, I will update it | 14:20 |
| Shrews | lucasagomes: what if we just add this link to CONTRIBUTING.rst rather than duplicate the info you suggested? http://docs.openstack.org/developer/python-ironicclient/contributing.html | 14:20 |
| lucasagomes | shardy, touché, sounds pretty good | 14:20 |
| lucasagomes | and have all the IRC, maillisting etc.. | 14:21 |
| Shrews | hi rloo. morning :) | 14:21 |
| lucasagomes | Shrews, * | 14:21 |
| lucasagomes | shardy, sorry | 14:21 |
| rloo | lintan: I'll try to get to it today | 14:21 |
| lucasagomes | rloo, morning | 14:21 |
| rloo | morning Shrews, hi lucasagomes | 14:21 |
| rloo | I wondered if 'shardy' was some slang I didn't know ;) | 14:21 |
| *** dlpartain has quit IRC | 14:25 | |
| openstackgerrit | David Shrewsbury proposed openstack/python-ironicclient: Workflow documentation is now in infra-manual https://review.openstack.org/139374 | 14:26 |
| lucasagomes | rloo, nop heh sh<tab> | 14:26 |
| Shrews | rloo, lucasagomes: see if that addresses your concerns ^^^^ | 14:26 |
| rloo | Shrews: in a few min, want to finish up a review | 14:27 |
| Shrews | rloo: no hurry :) | 14:28 |
| openstackgerrit | AKAHIRA Junya proposed openstack/ironic: Update command options in the Installation Guide https://review.openstack.org/142107 | 14:33 |
| rloo | Shrews: any idea where (other than the file itself) the CONTRIBUTING.rst stuff shows up? | 14:38 |
| *** Masahiro has joined #openstack-ironic | 14:38 | |
| lucasagomes | Shrews, will see in a min | 14:38 |
| Shrews | rloo: i don't think it shows up anywhere :) | 14:39 |
| rloo | Shrews: ok, I wondered... | 14:39 |
| Shrews | unless it's in docs directory, it isn't published | 14:40 |
| openstackgerrit | Vladyslav Drok proposed openstack/ironic: Support for non-Glance image references https://review.openstack.org/136741 | 14:42 |
| *** Masahiro has quit IRC | 14:43 | |
| rloo | hey lintan, instance_info is cleared when we tear down a node. Is the boot_device info only needed for that one instance? | 14:43 |
| *** killer_prince is now known as lazy_prince | 14:44 | |
| lintan | rloo: yes, each instance will have their own boot_device | 14:45 |
| Shrews | why is 142204 a "Partial-Bug" fix? doesn't it close the bug? | 14:46 |
| rloo | lintan: ok. so if eg someone issues a PUT /v1/nodes/../management/boot_device when the node is active, that info will be lost when the node is torn down. | 14:47 |
| lucasagomes | rloo, I really don't know a place to put that :/ I was thinking even whether extra would do it | 14:48 |
| lucasagomes | it seems to be an exception case | 14:48 |
| rloo | Shrews: the bug associated with that also has an oslo-incubator component. Should it be partial or fixes in that case? (I don't remember) | 14:49 |
| rloo | lucasagomes: I was going to ask you, since all the other set_boot_device stuff doesn't save it but issues a request to the node directly | 14:49 |
| Shrews | rloo: ah, hrm, yeah. partial is correct then | 14:50 |
| lucasagomes | rloo, yeah, because the persistent vs non-persistent ting is complicated to handle if we have a db field | 14:50 |
| *** pcaruana has quit IRC | 14:51 | |
| rloo | lucasagomes: yeah. anyway, that's why I'm holding up lintan :-(, so would be good if we could suggest something for him. | 14:51 |
| lucasagomes | perhaps maybe if we have a db field only for the persistent :/ (still we need to re validate it time to time so the info doesn't get out-dated, e.g someone changed the boot device via ipmitool directly instead of using our API) | 14:51 |
| lucasagomes | rloo, yeah, I think extra, while not perfect | 14:51 |
| lucasagomes | it continues after tear_down | 14:51 |
| lucasagomes | and isn't necessarily a field that only operators should update | 14:52 |
| lucasagomes | could be both | 14:52 |
| lucasagomes | both == operators and drivers | 14:52 |
| rloo | lucasagomes: It isn't quite clear to me what info goes into which fields, driver_info, instance_info, extra, and the capabilities are in properties | 14:54 |
| lucasagomes | rloo, the way I understand it is: driver_info = driver configuration, instance_info = information used to deploy the instance, extra = anything, properties = characteristics of the node | 14:55 |
| rloo | lucasagomes: so boot_device could be a property of a node? | 14:58 |
| * lucasagomes thinks | 14:59 | |
| rloo | lucasagomes: hmm, 'supported_boot_devices' could be a property of a node | 14:59 |
| lucasagomes | because the way I think about properties is something like physical characterists of it | 14:59 |
| rloo | oh wait, we have 'supported_boot_devices' as a property of a driver. | 15:00 |
| rloo | (sort of) | 15:00 |
| lucasagomes | 1 sec I'm on a quick call | 15:00 |
| *** erwan_taf has quit IRC | 15:00 | |
| *** erwan_taf has joined #openstack-ironic | 15:05 | |
| jroll | morning all :) | 15:08 |
| rloo | morning jroll | 15:08 |
| openstackgerrit | Naohiro Tamura proposed openstack/ironic-specs: iRMC Virtual Media Deploy Driver for Ironic https://review.openstack.org/134865 | 15:09 |
| lucasagomes | rloo, back | 15:11 |
| lucasagomes | jroll, morning | 15:11 |
| rloo | lucasagomes: I was looking around, does anything even use node.extra? | 15:11 |
| lucasagomes | rloo, so... hmm I'm not completly sure where it should go. I haven't thought about the case where the manager controller actually doesn't save that option | 15:11 |
| lucasagomes | rloo, not that I know | 15:11 |
| lucasagomes | not upstream at least | 15:11 |
| rloo | lucasagomes: I think for now, maybe use extra. we can always move it later. | 15:12 |
| lucasagomes | imo that's like a fallback attribute, that's why I was thinking about using it to unblock lintan | 15:12 |
| lucasagomes | rloo, yeah, sounds good | 15:12 |
| rloo | lucasagomes: yeah | 15:12 |
| rloo | lucasagomes: thx | 15:12 |
| rloo | lintan: ^^^ use node.extra | 15:12 |
| lintan | OK | 15:12 |
| lintan | thanks | 15:13 |
| lucasagomes | lintan, rloo I'm going for lunch and later I will finish the review and add a comment about it as well | 15:13 |
| lucasagomes | brb | 15:13 |
| naohirot | good night all | 15:13 |
| *** naohirot has quit IRC | 15:13 | |
| lucasagomes | rloo, ah one more thing | 15:13 |
| lucasagomes | naohirot night | 15:13 |
| jroll | lintan: rloo: lucasagomes: I think we've said before that ironic shouldn't touch extra | 15:13 |
| jroll | we've/deva | 15:13 |
| jroll | but idk | 15:14 |
| lintan | :( | 15:14 |
| lucasagomes | jroll, oh, why? I thought extra was like anything | 15:14 |
| lucasagomes | when we have a info that doesn't fit anywhere | 15:14 |
| jroll | I think it's "anything the operator wants to stuff there" | 15:14 |
| jroll | what are we putting here, current boot device? | 15:14 |
| lucasagomes | jroll, yeah | 15:14 |
| jroll | seems like driver_info, but I'm sure someone has pointed out that the operator shouldn't touch it | 15:15 |
| jroll | like for agent driver, we put agent_url in driver_info, operator isn't supposed to touch it | 15:15 |
| jroll | idk | 15:15 |
| jroll | seems like similar thing | 15:15 |
| jroll | "information the driver needs to know to work" | 15:15 |
| lucasagomes | yeah it does... hmm | 15:15 |
| lucasagomes | driver_info indeed is not bad | 15:16 |
| rloo | driver_info properties are things the user can specify, but lintan's code deletes it. | 15:16 |
| lucasagomes | rloo, re ur commnet https://review.openstack.org/#/c/135184/11/ironic/drivers/modules/amt/common.py at L191 | 15:16 |
| NobodyCam | morning ya'll | 15:17 |
| jroll | rloo: that's ok, just document that this key shouldn't be touched by the user | 15:17 |
| lucasagomes | rloo, I want add support for multiple types on those fields, so we could have dicts, list, bool etc... everything that can be JSON serializable | 15:17 |
| jroll | heya NobodyCam :) | 15:17 |
| lucasagomes | rloo, see: https://review.openstack.org/#/c/137762/ when you get some time | 15:17 |
| rloo | jroll: it just seems odd to me that the user can/could specify boot_device via driver_properties AND via the PUT.../management/boot_device API | 15:18 |
| jroll | clif_h: grats on your first patch landing :) https://review.openstack.org/#/c/142204/ | 15:18 |
| jroll | rloo: the user isn't supposed to specify boot device through driver_info in this case, right? | 15:19 |
| rloo | lucasagomes: yeah, will take a look sometime (I still want to focus on getting the fsm stuff first) | 15:19 |
| NobodyCam | morning jroll | 15:19 |
| lucasagomes | jroll, yeah no because there's no validation whether it's a valid/supported boot device or not | 15:19 |
| rloo | jroll: the user isn't, but if they see it in driver_info... I'm worried that it will set a precedent | 15:19 |
| lucasagomes | jroll, but in this case, I mean it's an exception | 15:19 |
| lucasagomes | so we could document it to unblock the work and then think about something better later | 15:19 |
| rloo | why not use extra for now until we can figure it out? | 15:19 |
| jroll | rloo: we have a precedent with agent_url already, the operator could touch that and break it as well | 15:20 |
| rloo | jroll: I should have -1'd that! | 15:20 |
| lucasagomes | I also prefer extra here because, for me extra sounds like "extra information" | 15:20 |
| lucasagomes | I mean I can't define which type of information goes into extra, I consider that everything needed | 15:20 |
| lucasagomes | jroll, is it documented somewhere? | 15:21 |
| jroll | lucasagomes: honestly, I think extra should only be touched by the operator, I think that's the intent | 15:21 |
| jroll | I'm grepping irc logs :P | 15:21 |
| jroll | lucasagomes: is which documented? | 15:21 |
| lucasagomes | jroll, the agent_url thing | 15:21 |
| jroll | not sure | 15:22 |
| jroll | our docs are horrible tbh | 15:22 |
| lucasagomes | jroll, is the agent_url a node-specific thing? I was wondering why you guys don't add it to the ironic.config instead | 15:22 |
| lucasagomes | under [agent] | 15:22 |
| jroll | yeah, it's the url to the agent running on that node | 15:22 |
| clif_h | jroll: thanks, it will be merged automatically now right? | 15:22 |
| lucasagomes | sounds like all nodes using the agent will talk to the same url | 15:22 |
| jroll | clif_h: yep! | 15:22 |
| lucasagomes | oh | 15:22 |
| lucasagomes | I see | 15:22 |
| jroll | lucasagomes: yeah, it's for ironic to talk to the agent | 15:23 |
| lucasagomes | gotcha | 15:23 |
| lucasagomes | I will have to step out a bit to have lunch with my fam | 15:24 |
| lucasagomes | brb | 15:24 |
| *** lucasagomes is now known as lucas-hungry | 15:24 | |
| jroll | later | 15:25 |
| lintan | jroll: So can we put amt_boot_device in extra at first? | 15:26 |
| jroll | I think it should go in driver_info | 15:26 |
| jroll | I am not the only voice here :) | 15:26 |
| jroll | other folks think extra is better | 15:26 |
| rloo | jroll: your objection to extra is cuz we shouldn't be using it? | 15:26 |
| jroll | we'll need to find swords and whatnot | 15:26 |
| jroll | rloo: basically, and that sounds like a bad reason when you put it that way | 15:27 |
| rloo | jroll: if others are ok with driver_info, that's fine. I just think that it doesn't make sense to have two places where a user could specify the boot device. | 15:27 |
| jroll | rloo: then let's make the key boot_device_do_not_touch | 15:27 |
| jroll | or current_boot_device | 15:27 |
| jroll | or something | 15:27 |
| jroll | idk, I'd want to ask devananda | 15:27 |
| rloo | jroll: that looks messy | 15:28 |
| jroll | because I swear he's the one that said extra is ops-only | 15:28 |
| jroll | but I might just be making things up | 15:28 |
| jroll | (the boot_device_do_not_touch thing was mostly a joke) | 15:28 |
| rloo | jroll: ok, wait for the man then. We should document this stuff... | 15:28 |
| lintan | That's fine for me | 15:29 |
| jroll | yeah, focus on the rest for now :) | 15:30 |
| yuriyz | morning Ironic | 15:37 |
| NobodyCam | morning yuriyz | 15:37 |
| NobodyCam | morning rloo | 15:37 |
| lintan | morning yuriyz | 15:37 |
| NobodyCam | morning lintan | 15:37 |
| lintan | morning NobodyCam | 15:37 |
| rloo | morning NobodyCam, afternoon yuriyz | 15:37 |
| BadCub_ | morning Ironic | 15:38 |
| yuriyz | lucasagomes, please look at new version of Fuel Agent spec if you have some time | 15:39 |
| NobodyCam | morning BadCub_ | 15:40 |
| lintan | Hi yuriyz: | 15:42 |
| *** achanda has joined #openstack-ironic | 15:44 | |
| Shrews | We need to prioritize 141598, if another core has a moment to take a peek. | 15:47 |
| *** anderbubble has joined #openstack-ironic | 15:48 | |
| Shrews | http://lists.openstack.org/pipermail/openstack-dev/2014-December/053179.html | 15:49 |
| * jroll looks | 15:50 | |
| NobodyCam | yuriyz: I know I said I would review 1st thing. but I forgot about our corp all hand call... I will take a look after that | 15:51 |
| NobodyCam | :-p | 15:51 |
| rloo | Shrews: since when did we get +2s on stable branches? | 15:55 |
| rloo | Shrews: +2 powers? | 15:55 |
| Shrews | rloo: i was surprised, too | 15:55 |
| * Shrews brb | 15:55 | |
| rloo | Shrews: does that mean we can approve too? I thought there was a long thread about us not approving... | 15:56 |
| jroll | there was a mail about this | 15:58 |
| jroll | I forget what it said :) | 15:58 |
| *** pcrews has joined #openstack-ironic | 16:00 | |
| rloo | jroll: ha ha. I think I skimmed/skipped it too. I +2'd but too chicken to approve ;) | 16:00 |
| jroll | :P | 16:01 |
| jroll | oops, still haven't sent subteam status thing :| | 16:01 |
| * jroll does that | 16:01 | |
| *** erwan_taf has quit IRC | 16:11 | |
| *** jerryz has quit IRC | 16:12 | |
| *** lucas-hungry is now known as lucasagomes | 16:23 | |
| *** erwan_taf has joined #openstack-ironic | 16:24 | |
| *** Masahiro has joined #openstack-ironic | 16:27 | |
| *** jrist-afk is now known as jrist | 16:30 | |
| *** jcoufal_ is now known as jcoufal | 16:31 | |
| *** Masahiro has quit IRC | 16:31 | |
| *** christopheraedo is now known as docaedo | 16:38 | |
| devananda | projecs are now responsible for their own stable maint. part of big-tent-ification of openstack | 16:40 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/python-ironicclient: Support non-string setting non-string fields https://review.openstack.org/142492 | 16:40 |
| lucasagomes | devananda, morning | 16:40 |
| lucasagomes | devananda, even cores have no power to merge thing on stable branches, they already changed that? | 16:40 |
| *** achanda has quit IRC | 16:41 | |
| NobodyCam | morning devananda | 16:41 |
| *** dwalleck has joined #openstack-ironic | 16:41 | |
| lucasagomes | NobodyCam, yo morning | 16:41 |
| *** achanda has joined #openstack-ironic | 16:41 | |
| NobodyCam | morning lucasagomes hehehehe | 16:42 |
| lucasagomes | yuriyz, I have to catch up with the whole discussion on the fuel, I was reading the ML but haven't finished yet | 16:43 |
| lucasagomes | I will try to catch up later on | 16:43 |
| lucasagomes | s/fuel/fuel agent/ | 16:43 |
| yuriyz | thanks lucasagomes | 16:43 |
| lucasagomes | I'm impressed of how many can things happens in 1 week | 16:44 |
| devananda | lucasagomes: I believe so, yes | 16:44 |
| lucasagomes | devananda, cool! | 16:44 |
| devananda | lucasagomes: we should update that list, though. it's separate from ironic-core (but copied from it initially) | 16:44 |
| lucasagomes | devananda, btw, I know it's early morning there. But when you find a time mind scrolling up a bit and see the extra vs driver_info discussion to unblock the AMT driver works? | 16:45 |
| devananda | I don't think all our core team need to or are interested ins table maint | 16:45 |
| lucasagomes | devananda, right, yeah we have diff lists for specs as well | 16:45 |
| *** achanda has quit IRC | 16:46 | |
| devananda | lucasagomes, rloo: why is boot device a one-time setting? | 16:46 |
| devananda | i mean, I undersatnd that the AMT hardware doesn't save that info between power cycles | 16:47 |
| lucasagomes | devananda, seems AMT doesn't support persistent boot devices | 16:47 |
| devananda | it needs to be applied to the hardware each boot | 16:47 |
| devananda | but it doesn't need to be a different *vaklue* each time | 16:47 |
| devananda | value, even | 16:47 |
| devananda | so it is clearly driver_info to me | 16:47 |
| devananda | what am I missing? | 16:47 |
| lucasagomes | I think that someone pointed that driver_info should be set by the operator configuring the driver and not the driver setting stuff there | 16:48 |
| devananda | instance_info is for things that are changed for each instance. this is a constant value which needs to be applied prior to each power-on | 16:48 |
| lucasagomes | tho I'm not sure about those assumptions | 16:48 |
| devananda | sure | 16:48 |
| lucasagomes | I would rather put it in driver_info or extra... | 16:48 |
| devananda | oh. so why does it even need to be stored in the db? | 16:49 |
| lucasagomes | there's also an assuptiom that extra shouldn't be edited by drivers | 16:49 |
| devananda | when booting a deploy ramdisk, it always needs to be "network" | 16:49 |
| devananda | when booting the user image via (i)pxe, it still needs to be network | 16:49 |
| * lucasagomes checks if code sets it everytime | 16:49 | |
| devananda | when booting from local media, it should be soemthing else | 16:49 |
| devananda | the driver knows this | 16:49 |
| devananda | but it doens't need to be stored anywhere | 16:49 |
| devananda | or settable by nova, or a user | 16:49 |
| lucasagomes | the only case I see it's needed to be store somewhere is to adhere to the get_boot_device() from management interface | 16:50 |
| lucasagomes | tho it could raise not implemented if not supported | 16:51 |
| devananda | yup | 16:51 |
| *** achanda has joined #openstack-ironic | 16:51 | |
| lucasagomes | so seems PXE_ ony sets the boot_device as part of the deploy() (set as persistent) | 16:52 |
| lucasagomes | iLO when used with virtualmedia does set it on every power on/off | 16:53 |
| lucasagomes | I believe it's because the swift temp url may expire so it needs to reset that | 16:54 |
| * lucasagomes diggs more | 16:54 | |
| NobodyCam | brb | 16:55 |
| *** dwalleck has quit IRC | 16:55 | |
| *** viktors is now known as viktors|afk | 16:55 | |
| rloo | lucasagomes, devananda: I didn't have time to look into detail at the amt driver patch, but it was doing something odd wrt having boot-device in driver_info, and then deleting that info. and the way it was coded, looked like the user could specify boot-device in driver-info. | 16:56 |
| rloo | lucasagomes, devananda: but the user can specify boot-device via mgt interface. | 16:56 |
| lucasagomes | rloo, right, yeah I think that once kept in driver_info user will be able to modify it :/ | 16:56 |
| rloo | lucasagomes, devananda: and if you allow the user to specify boot-device via driver-info, should that info be deleted w/o them knowing | 16:56 |
| lucasagomes | same for extra etc | 16:56 |
| * lucasagomes will finish the review on that patch | 16:57 | |
| *** igordcard has joined #openstack-ironic | 16:58 | |
| *** anderbubble has quit IRC | 16:59 | |
| *** Marga_ has joined #openstack-ironic | 17:01 | |
| *** harlowja_away is now known as harlowja | 17:01 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Extend API multivalue fields https://review.openstack.org/137762 | 17:04 |
| *** pensu has joined #openstack-ironic | 17:04 | |
| devananda | rloo: that sounds like poor form. good raeson to block it for now | 17:05 |
| devananda | rloo: I'm also concerned that it imports pywsman library, then shells out to wsman CLI | 17:05 |
| devananda | I dont understand why it needs both, and doesn't mention in the spec that it will use the wsman CLI | 17:05 |
| devananda | driving into the office, bbiab, though i'm likely to get pulled into meetings today | 17:06 |
| rloo | devananda: hmm, I didn't think about that wsman stuff. That's a good question. | 17:06 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Extend API multivalue fields https://review.openstack.org/137762 | 17:06 |
| rloo | devananda: so i'm not sure how to unblock the amt driver. Maybe lucasagomes will have more ideas after reviewing. | 17:07 |
| devananda | rloo: also, thanks for the unit test patch. I'm trying to find time to get back to working on that | 17:07 |
| devananda | NobodyCam: any updates on the node logical names patch? | 17:07 |
| NobodyCam | nope | 17:07 |
| NobodyCam | :( | 17:07 |
| rloo | devananda: i am hoping to find more time today to add more unit tests for the fsm stuff. | 17:07 |
| devananda | rloo: awesome, tyvm | 17:07 |
| rloo | devananda, NobodyCam: on my list to look at the node logical names stuff today too. | 17:07 |
| devananda | ok, realy gotta run now -- bbiab | 17:08 |
| NobodyCam | yuriyz: question on hte fuel spec | 17:11 |
| NobodyCam | The driver gets a callback when ramdisk boot(s) is that via the conductors api? or another method? | 17:12 |
| *** Marga_ has quit IRC | 17:15 | |
| *** lazy_prince is now known as killer_prince | 17:18 | |
| *** Marga_ has joined #openstack-ironic | 17:19 | |
| *** athomas has quit IRC | 17:19 | |
| *** achanda has quit IRC | 17:20 | |
| *** penick has joined #openstack-ironic | 17:22 | |
| *** anderbubble has joined #openstack-ironic | 17:22 | |
| *** romcheg has quit IRC | 17:25 | |
| *** jcoufal has quit IRC | 17:25 | |
| *** erwan_taf has quit IRC | 17:26 | |
| afaranha | lucasagomes, I'm trying this, but again I got some erros. At least it's an advance | 17:27 |
| afaranha | I have: console | False | Missing 'ipmi_terminal_port' parameter in node's driver_info. | 17:27 |
| afaranha | and: deploy | False | Cannot validate iSCSI deploy. Some parameters were missing in node's instance_info. Missing are: ['root_gb', 'image_source'] | 17:27 |
| afaranha | I tried to add the ipmi_terminal_port as 22, but I got a conflict error: | 17:29 |
| afaranha | ironic node-update a33d58e1-945d-4b6d-a291-c3f8b9cddd33 add driver_info/ipmi_terminal_port=22 | 17:29 |
| lucasagomes | afaranha, right, root_gb and image_source will be fullfield by nova. (perhaps it's a bug that we are validating that before the node is even deployed) | 17:29 |
| lucasagomes | afaranha, it always returns conflict? | 17:29 |
| jroll | yeah, sounds like a bug | 17:29 |
| afaranha | I tried ironic node-update add, and node-update replace | 17:31 |
| afaranha | but same error | 17:31 |
| afaranha | Bad Request (HTTP 400) | 17:32 |
| *** achanda has joined #openstack-ironic | 17:32 | |
| lucasagomes | afaranha, the cmdline looks correct | 17:32 |
| lucasagomes | afaranha, mind passing --debug to it pasting ? | 17:33 |
| afaranha | lucasagomes, jroll management and power are True, does this mean the node was recognized? | 17:33 |
| afaranha | lucasagomes, http://paste.openstack.org/show/152354/ | 17:34 |
| *** thiagop has joined #openstack-ironic | 17:35 | |
| lucasagomes | afaranha, you said you've tried add and replace right? | 17:36 |
| lucasagomes | do add return 409? cause I see why replace may return 400 | 17:36 |
| lucasagomes | because it's trying to replace something that doesn't exist | 17:36 |
| lucasagomes | so it's a bad request | 17:36 |
| afaranha | lucasagomes, sorry this is for username and password | 17:36 |
| afaranha | the correct one: http://paste.openstack.org/show/152355/ | 17:36 |
| *** penick has quit IRC | 17:36 | |
| NobodyCam | brb | 17:37 |
| afaranha | lucasagomes, and now as replace http://paste.openstack.org/show/152356/ | 17:37 |
| lucasagomes | right I think the 409 is the error here... lemme check something in the code | 17:37 |
| lucasagomes | afaranha, in a node-show, the reservation field is None? | 17:38 |
| *** Nisha has joined #openstack-ironic | 17:39 | |
| afaranha | lucasagomes, it's localhost | 17:39 |
| lucasagomes | afaranha, a-ha, so I think that's the cause of the conflict | 17:39 |
| afaranha | http://paste.openstack.org/show/152361/ | 17:39 |
| lucasagomes | the reservation field indicates if a conductor is working on that node | 17:39 |
| *** marcoemorais has joined #openstack-ironic | 17:39 | |
| lucasagomes | mid-operation... the bad news is that | 17:40 |
| *** anderbubble has quit IRC | 17:40 | |
| afaranha | lucasagomes, ah, so... Can I just set it None? | 17:40 |
| lucasagomes | that works as a lock, and if conductor dies mid-operation we can end up with that lock stuck there | 17:40 |
| afaranha | lucasagomes, Actually there is nothing running on the machine | 17:40 |
| lucasagomes | and we currently have no API calls that could "break/reset" that to None | 17:40 |
| lucasagomes | afaranha, yea, but time to time conductors lock the node to do things like sync'ing power state etc | 17:41 |
| lucasagomes | I'm wondering if the conductor died or was killing when doing that | 17:41 |
| lucasagomes | afaranha, you can access the db ? can you set that reservation to NULL and try updating the node? | 17:41 |
| afaranha | yes I can | 17:42 |
| afaranha | So, I still have to inform the ipmi port? | 17:42 |
| lucasagomes | afaranha, I don't think you need it, unless you want to have console access | 17:43 |
| lucasagomes | but not necessary to deploy a node | 17:43 |
| lucasagomes | the root_gb and image_source will be filled by nova as part of the deploy process | 17:43 |
| lucasagomes | jroll, any news from last week about breaking the lock via API? | 17:43 |
| lucasagomes | the more I think, the more I believe we need an API for it | 17:44 |
| *** marcoemorais has quit IRC | 17:44 | |
| *** marcoemorais has joined #openstack-ironic | 17:44 | |
| *** marcoemorais1 has joined #openstack-ironic | 17:45 | |
| afaranha | lucasagomes, done! Let me try active the node again :) | 17:45 |
| lucasagomes | cool | 17:45 |
| *** marcoemorais2 has joined #openstack-ironic | 17:45 | |
| *** marcoemorais2 has quit IRC | 17:45 | |
| lucasagomes | check if u can update it first | 17:45 |
| lucasagomes | just to see if that solves the update problem | 17:45 |
| *** marcoemorais1 has quit IRC | 17:46 | |
| *** marcoemorais2 has joined #openstack-ironic | 17:46 | |
| *** marcoemorais2 has quit IRC | 17:46 | |
| *** marcoemorais1 has joined #openstack-ironic | 17:46 | |
| lucasagomes | also as u'r using baremetal nodes (ipmi driver) you can try to power on/off it before dpeloying just to see if ironic can talk to the BMC correctly and all | 17:46 |
| lucasagomes | ironic node-set-power-state on/off | 17:46 |
| afaranha | lucasagomes, Yes, I'm able now to add the ipmi_port :) | 17:47 |
| lucasagomes | afaranha, cool | 17:48 |
| afaranha | lucasagomes, sure, this is what I wanna know | 17:48 |
| *** PaulCzar has joined #openstack-ironic | 17:48 | |
| afaranha | If ironic can communicate with the server | 17:48 |
| lucasagomes | yeah, power it on and off is easier to identify that instead of trying to deploy it directly | 17:48 |
| lucasagomes | easier to debug as well | 17:48 |
| *** marcoemorais has quit IRC | 17:49 | |
| afaranha | lucasagomes, I think I have a problem in this point, I still have power_state as None | 17:50 |
| lucasagomes | I see | 17:50 |
| lucasagomes | when you try to change the power state via CLI/API do you see anything odd in the conductor logs? | 17:50 |
| afaranha | lucasagomes, let me check. Now I got conflict again when I try to power on/off the node | 17:52 |
| lucasagomes | sounds like the conductor is trying to sync the power state and is getting stuck | 17:52 |
| lucasagomes | you can disable that via config | 17:52 |
| lucasagomes | lemme find the option for ya | 17:52 |
| *** andreykurilin_ has joined #openstack-ironic | 17:53 | |
| lucasagomes | afaranha, #force_power_state_during_sync=true | 17:53 |
| lucasagomes | set it to false | 17:53 |
| afaranha | okay | 17:54 |
| JayF | That might keep the conductor from trying to change the power status but that doesn't disable the status loop itself | 17:55 |
| JayF | so conductor can still get "stuck" if there are lots of nodes with invalid ipmi creds | 17:55 |
| afaranha | I didn't see anything strange in the conduct, no errors | 17:55 |
| *** ryanpetrello has joined #openstack-ironic | 17:56 | |
| afaranha | JayF, Let me try something else, I'll try to put the username and password | 17:56 |
| *** penick has joined #openstack-ironic | 17:56 | |
| *** derekh has quit IRC | 18:00 | |
| afaranha | lucasagomes, JayF I think it still stuck in the power on, when I try update the node I got conflict again | 18:00 |
| lucasagomes | afaranha, this was before changing the config right? yeah I think that the conductor may be trying to talk to the bmc to sync the power state | 18:01 |
| lucasagomes | and is getting stuck there | 18:01 |
| lucasagomes | probably until some timeout occur | 18:01 |
| afaranha | lucasagomes, okay, I gonna wait then, maybe 5min? | 18:02 |
| *** ifarkas has quit IRC | 18:02 | |
| lucasagomes | afaranha, idk I would just change that option restart conductor | 18:02 |
| lucasagomes | clean that field in the db and try manually to see | 18:02 |
| afaranha | lucasagomes, Or better, as it's trying to power on the server, I'm gonna remove the credentials to use IPMI | 18:03 |
| lucasagomes | afaranha, you can also put the node in maintenance mode so conductors won't touch it | 18:04 |
| afaranha | lucasagomes, I think the node is badly configured | 18:07 |
| afaranha | but When I validate it says power True =/ | 18:08 |
| lucasagomes | the validation mostly check is the params are present | 18:08 |
| jroll | lucasagomes: we haven't talked about it really | 18:09 |
| jroll | lucasagomes: but we're going to work on a graceful conductor shutdown thing real soon (this week?) | 18:09 |
| lucasagomes | jroll, graceful conductor shutdown?! | 18:11 |
| jroll | yeah, finish things and unlock before shutting down | 18:11 |
| lucasagomes | oh | 18:12 |
| jroll | that's how locks tend to get stuck | 18:12 |
| lucasagomes | that would be awesome | 18:12 |
| jroll | :) | 18:12 |
| lucasagomes | but still, I don't whether it can cover all the usecases for the lock getting stuck | 18:12 |
| lucasagomes | we may still need an API/something for that | 18:12 |
| *** Marga_ has quit IRC | 18:13 | |
| jroll | oh, I agree | 18:13 |
| *** Marga_ has joined #openstack-ironic | 18:14 | |
| *** romcheg has joined #openstack-ironic | 18:14 | |
| *** Masahiro has joined #openstack-ironic | 18:16 | |
| *** pensu has quit IRC | 18:16 | |
| lucasagomes | once someone has a time please take a look at https://review.openstack.org/#/c/138729/ | 18:17 |
| NobodyCam | anyone know Tan Lin's irc handle? | 18:18 |
| *** Masahiro has quit IRC | 18:20 | |
| lucasagomes | NobodyCam, lintan | 18:22 |
| NobodyCam | lucasagomes: I'll bet you are correct | 18:22 |
| NobodyCam | I found my answer :) | 18:22 |
| *** penick has quit IRC | 18:23 | |
| *** marcoemorais1 has quit IRC | 18:30 | |
| *** marcoemorais has joined #openstack-ironic | 18:30 | |
| *** ryanpetrello has quit IRC | 18:31 | |
| *** penick has joined #openstack-ironic | 18:31 | |
| *** marcoemorais has quit IRC | 18:31 | |
| *** marcoemorais has joined #openstack-ironic | 18:31 | |
| *** marcoemorais has quit IRC | 18:32 | |
| *** marcoemorais has joined #openstack-ironic | 18:32 | |
| *** igordcard has quit IRC | 18:38 | |
| *** dwalleck has joined #openstack-ironic | 18:40 | |
| *** thiagop has quit IRC | 18:45 | |
| *** Hefeweizen has quit IRC | 18:45 | |
| NobodyCam | JayF: you see Michael's comment on 141188? | 18:47 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 18:48 |
| *** afaranha has quit IRC | 18:49 | |
| *** Marga_ has quit IRC | 18:49 | |
| *** Marga_ has joined #openstack-ironic | 18:49 | |
| *** ryanpetrello has joined #openstack-ironic | 18:50 | |
| *** igordcard has joined #openstack-ironic | 18:51 | |
| *** achanda has quit IRC | 18:54 | |
| *** achanda has joined #openstack-ironic | 18:55 | |
| *** cuihaozhi has quit IRC | 19:00 | |
| *** cuihaozhi has joined #openstack-ironic | 19:00 | |
| *** Marga_ has quit IRC | 19:01 | |
| *** Marga_ has joined #openstack-ironic | 19:02 | |
| *** marcoemorais has quit IRC | 19:05 | |
| NobodyCam | gah .. you know your tired when switching tabs you come across a half written comment and cann't recall writting the first half | 19:06 |
| *** dwalleck has quit IRC | 19:07 | |
| *** penick has quit IRC | 19:12 | |
| Shrews | NobodyCam: lol. hang in there. a long vacation is coming soon | 19:16 |
| NobodyCam | :) oh I need it bad... | 19:18 |
| NobodyCam | lol | 19:18 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 19:20 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 19:21 |
| Shrews | whoa. zuul queue is waaaaaay backedup | 19:23 |
| NobodyCam | lol .. just a bit | 19:25 |
| *** ndipanov has quit IRC | 19:31 | |
| *** andreykurilin_ has quit IRC | 19:32 | |
| *** 16WAANWAL has joined #openstack-ironic | 19:34 | |
| *** 16WAANWAL has left #openstack-ironic | 19:34 | |
| *** alexpilotti has quit IRC | 19:36 | |
| *** afaranha has joined #openstack-ironic | 19:37 | |
| *** ChuckC has quit IRC | 19:39 | |
| *** penick has joined #openstack-ironic | 19:42 | |
| NobodyCam | are we on for Bug jam tomorrow? | 19:42 |
| openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Add network provider interface and implementations https://review.openstack.org/139687 | 19:42 |
| jroll | would love some eyes on the direction here ^ | 19:42 |
| jroll | spec is yet to be written | 19:42 |
| jroll | but this is basically a saner port of what we're doing downstream | 19:42 |
| * jroll WIP's it | 19:43 | |
| NobodyCam | oh now you've done it.. | 19:45 |
| * NobodyCam now has https://www.youtube.com/watch?v=j_QLzthSkfM stuck in his head | 19:46 | |
| jroll | lol | 19:47 |
| *** jcoufal has joined #openstack-ironic | 19:59 | |
| Shrews | jroll: that code is certainly well organized and clear | 19:59 |
| jroll | hooray | 19:59 |
| jroll | seems sensible? | 20:00 |
| Shrews | i wonder if it's worth trying to share the client within NetworkProvider | 20:00 |
| * jroll will likely write the spec on a plane this weekend or something | 20:00 | |
| jroll | right | 20:00 |
| jroll | so we did that when working on the dhcp stuff | 20:00 |
| jroll | and there were potential problems with a long-running client or something | 20:00 |
| jroll | I don't remember exactly | 20:00 |
| Shrews | yeah, i can understand that | 20:00 |
| NobodyCam | jroll: plane? | 20:00 |
| jroll | NobodyCam: yeah, flying to michigan this weekend | 20:01 |
| jroll | sunday, specifically | 20:01 |
| NobodyCam | awesome!!!! | 20:01 |
| NobodyCam | enjoy! | 20:01 |
| jroll | thanks :) | 20:01 |
| jroll | (family is all there and whatnot, I'll be off next week) | 20:02 |
| NobodyCam | sounds like fun! | 20:02 |
| *** ryanpetrello has quit IRC | 20:03 | |
| Shrews | jroll: yeah, seems sensible, but what would be the use case for using something other than neutron? | 20:03 |
| Shrews | as an example for me | 20:03 |
| jroll | Shrews: 1) not doing it, 2) using neutron with whatever upstream thing eventually talks to switches, rather than rackspace's api extension | 20:04 |
| *** Masahiro has joined #openstack-ironic | 20:04 | |
| jroll | the "plugin" in neutron_plugin is a reference to an out of tree plugin we made: https://github.com/rackerlabs/ironic-neutron-plugin | 20:05 |
| Shrews | ah, beginning to get the picture | 20:09 |
| *** Masahiro has quit IRC | 20:09 | |
| jroll | also, presumably people have written other things to talk to switches, too | 20:09 |
| jroll | you could run ncclient directly from ironic or something (netconf thing) | 20:10 |
| *** Nisha has quit IRC | 20:12 | |
| *** thiagop has joined #openstack-ironic | 20:19 | |
| lucasagomes | I will call it a day | 20:22 |
| lucasagomes | have a good night everyone | 20:22 |
| *** lucasagomes is now known as lucas-dinner | 20:22 | |
| *** kevinbenton has quit IRC | 20:22 | |
| NobodyCam | have a good night lucas-dinner | 20:22 |
| *** penick has quit IRC | 20:24 | |
| *** penick has joined #openstack-ironic | 20:31 | |
| *** jdob has joined #openstack-ironic | 20:42 | |
| jdob | NobodyCam: btw, my ironic issues from yesterday seem to be much better now that I resolved the fact that SSH logins were taking so long. thanks again :D | 20:42 |
| NobodyCam | jdob: awesome!!!! | 20:43 |
| slagle_ | hi, in common/disk_partitioner.py is the point of this method to wait for the disk to become available? https://github.com/openstack/ironic/blob/master/ironic/common/disk_partitioner.py#L112 | 20:46 |
| slagle_ | (ok, i feel slightly silly asking that given the name of the method is _wait_for_disk_to_become_available) | 20:47 |
| slagle_ | but that call to fuser there allows for a 1 exit code, and fuser /nonexistent/file returns 1 | 20:48 |
| *** penick has quit IRC | 20:48 | |
| *** achanda has quit IRC | 20:49 | |
| *** r-daneel has joined #openstack-ironic | 20:49 | |
| *** achanda has joined #openstack-ironic | 20:49 | |
| *** jcoufal has quit IRC | 20:50 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 20:51 |
| *** kevinbenton has joined #openstack-ironic | 20:51 | |
| slagle_ | hmm, i guess the checking of stderr would catch that case | 20:53 |
| *** jcoufal has joined #openstack-ironic | 20:53 | |
| *** igordcard has quit IRC | 20:53 | |
| PaulCzar | devananda: NobodyCam I just successfully deployed on actual physical hardware via the pxe_ipmi driver! | 20:53 |
| *** achanda has quit IRC | 20:54 | |
| NobodyCam | PaulCzar: awesome! :) | 20:54 |
| NobodyCam | PaulCzar: now comes my /shameless_plug: Having just gone thru what you did are there any Ironic Doc improvments you think would help. | 20:58 |
| *** penick has joined #openstack-ironic | 21:00 | |
| PaulCzar | NobodyCam: definitely would love to see better docs around the drivers and their different uses and requirements | 21:02 |
| PaulCzar | and I think better documentation for using ironic outside of devstack in general | 21:02 |
| *** dprince has quit IRC | 21:04 | |
| NobodyCam | want to toss up a patch for us to review? (<- Shameless bit) | 21:08 |
| * NobodyCam steps afk for a bit... | 21:10 | |
| jroll | PaulCzar: \o/ awesome | 21:13 |
| *** Marga_ has quit IRC | 21:21 | |
| Shrews | FYI, py33 jobs seem to be failing. Infra is aware, it would seem. | 21:26 |
| Shrews | and as a reminder, today is my last day until 2015 :) | 21:30 |
| Shrews | see you all next year | 21:30 |
| *** yjiang5_away is now known as yjiang5 | 21:30 | |
| PaulCzar | crazy question ... does ironic handle LVM partitioned images? | 21:31 |
| jroll | have a good one Shrews :D | 21:32 |
| *** andreykurilin_ has joined #openstack-ironic | 21:33 | |
| *** Shrews has quit IRC | 21:37 | |
| *** Hefeweizen has joined #openstack-ironic | 21:38 | |
| NobodyCam | PaulCzar: currently ironic does not. thou there is a spec up that could provide that support when completed | 21:40 |
| *** chuckC_ has joined #openstack-ironic | 21:45 | |
| PaulCzar | how are people currently handling RAID? preconfigure it and put the device name in disk_devices= ? | 21:46 |
| NobodyCam | j* have a second to speak to ^^^^^ | 21:46 |
| jroll | we don't do raid? | 21:47 |
| NobodyCam | PaulCzar: that would work for test/poc env's but not production | 21:47 |
| devananda | PaulCzar: out of band / preconfigure. yes. | 21:47 |
| jroll | s/?// | 21:47 |
| NobodyCam | jroll: nm then... | 21:47 |
| jroll | :) | 21:47 |
| jroll | I didn't answer for a reason :P | 21:48 |
| devananda | PaulCzar: there's a spec up describing where raid will fit in the provisioning workflow, and several proposals to add it | 21:48 |
| NobodyCam | morning / afternoondevananda :) | 21:48 |
| devananda | eg, for drac, etc | 21:48 |
| devananda | which I'd really like us to land this cycle ... | 21:48 |
| PaulCzar | I would expect it wouldn't be all that hard to embed the raid cli tools in the deployer image and run the cli to build the raid before doing the callback to ironic-api to kick off the iscsi mount etc | 21:48 |
| devananda | PaulCzar: nope. that is actuyally possible today | 21:48 |
| devananda | PaulCzar: you can customize the depl;oy ramdisk to statically do that on your hardware today | 21:49 |
| *** lucas-dinner has quit IRC | 21:49 | |
| PaulCzar | devananda: yeah, that's what I suspect we'll want to do | 21:50 |
| devananda | PaulCzar: the ongoing work is for an API for ironic to control that // change RAID config in response to user requests , nova flavors, etx | 21:50 |
| PaulCzar | devananda: thx, that will help me when the equipment with raid in it shows up | 21:51 |
| PaulCzar | right now my boxes just have a single disk | 21:51 |
| *** Masahiro has joined #openstack-ironic | 21:53 | |
| *** Masahiro has quit IRC | 21:58 | |
| *** jdob has quit IRC | 21:59 | |
| *** pcrews has quit IRC | 22:06 | |
| *** Marga_ has joined #openstack-ironic | 22:08 | |
| NobodyCam | devananda: still about? | 22:08 |
| devananda | yep | 22:08 |
| NobodyCam | gots a second to take a look at https://review.openstack.org/#/c/100951 (its got two +2's) | 22:09 |
| *** jcoufal has quit IRC | 22:10 | |
| NobodyCam | brb | 22:13 |
| *** romcheg has quit IRC | 22:20 | |
| *** romcheg has joined #openstack-ironic | 22:22 | |
| *** romcheg has quit IRC | 22:24 | |
| *** marcoemorais has joined #openstack-ironic | 22:40 | |
| NobodyCam | jroll: got a second to take another quick look at https://review.openstack.org/#/c/141766 you had +2 two revs ago :) | 22:49 |
| *** ryanpetrello has joined #openstack-ironic | 22:54 | |
| jroll | why'd you all override my +2? >:( | 22:54 |
| * jroll looks :P | 22:54 | |
| NobodyCam | lol | 22:54 |
| *** rlr0329 has joined #openstack-ironic | 22:55 | |
| rlr0329 | Trying to setup iPXE in juno release of ironic. Followed steps 1-7 for iPXE setup in the install-guide. Getting 'File not found' in tftp server log for filename http://<ip address>/ironic/boot.ipxe Have verified the URL is valid using wget. Any idea what I am missing? | 22:55 |
| NobodyCam | did you create the tftp map file? | 22:56 |
| NobodyCam | oh wait | 22:56 |
| JayF | It looks like your dhcp server is not sending an ipxe rom first | 22:58 |
| JayF | because it's trying to pull down the filename with tftp which indicates it's your NIC boot rom still running; not ipxe | 22:59 |
| rlr0329 | JayF: I am using neutron dhcp server. Is there a neutron config change? | 22:59 |
| devananda | NobodyCam: on the inspect spec, was there a discussion on creating a new interface vs. adding this to the management interface? | 23:00 |
| JayF | I am unfamiliar with neutron; I use Ironic with the "None" DHCP provider. | 23:00 |
| JayF | But I'd suspect so, yes | 23:00 |
| devananda | also, i'm still not thrilled about adding a new API endpoint for changing xtate, rather than using /state/procision | 23:01 |
| rlr0329 | JayF: What do you use for dhcp? | 23:01 |
| jroll | rlr0329: we run isc-dhcp-server outside of ironic | 23:01 |
| jroll | rlr0329: with static configuration | 23:02 |
| jroll | devananda: there was discussion at some point, I remember some sort of consensus on the new interface (sadface) | 23:02 |
| rlr0329 | jroll: Do you have an example of your config and how you setup the ipxe rom? | 23:02 |
| devananda | I -1'd rev 52. I dont see any reply to my question | 23:03 |
| *** ryanpetrello has quit IRC | 23:03 | |
| devananda | should we be creating a new resource, /v1/nodes/<uuid>/inspect, which directly affects a *different* resource, namely /v1/nodes/<uuid>/states/provision? | 23:03 |
| jroll | rlr0329: something like this https://gist.github.com/jimrollenhagen/0d35f7de9307bcad4a61 | 23:03 |
| NobodyCam | devananda: I have to look back over the spec. I recall a discussion like that gennerally | 23:03 |
| devananda | NobodyCam: where? | 23:03 |
| NobodyCam | trying to recall | 23:04 |
| rlr0329 | jroll: will do. Thanks! | 23:04 |
| jroll | rlr0329: not saying I recommend this configuration, it's just how we run ironic :) | 23:04 |
| rlr0329 | jroll: understand | 23:05 |
| *** marcoemorais has quit IRC | 23:13 | |
| *** marcoemorais has joined #openstack-ironic | 23:13 | |
| *** andreykurilin_ has quit IRC | 23:14 | |
| NobodyCam | devananda: I can not find that discussion :( | 23:36 |
| devananda | NobodyCam: jroll seems to recall it as well. perhaps one of you remembers the outcome? | 23:37 |
| jroll | I remember "++ IntrospectionINterface" or something | 23:37 |
| jroll | let me dig logs | 23:37 |
| jroll | huh, wheretf | 23:41 |
| devananda | right | 23:41 |
| devananda | i just grepped irc logs, dont see it | 23:41 |
| *** Masahiro has joined #openstack-ironic | 23:42 | |
| jroll | yeah, I feel like that happened | 23:43 |
| jroll | sigh | 23:44 |
| NobodyCam | jroll: maybe we need a holiday | 23:44 |
| jroll | I just need a better memory | 23:45 |
| devananda | if you recall the reasoning for it...? | 23:45 |
| * jroll looks at the review | 23:45 | |
| jroll | devananda: because screw the exploding driver matrix | 23:45 |
| jroll | idk honestly | 23:45 |
| devananda | right. lets explode it more :( | 23:46 |
| devananda | :) even | 23:46 |
| devananda | in what way is that not a management function? | 23:46 |
| *** Masahiro has quit IRC | 23:46 | |
| devananda | when it's done out-of-band , it's going to use the same library and very similar code to the mgmt interface | 23:47 |
| *** naohirot has joined #openstack-ironic | 23:47 | |
| naohirot | good morning ironic | 23:48 |
| devananda | oh sorry. that's not what I asked | 23:48 |
| devananda | jroll: so I dont think it needs a new driver interface. but i cxare less about that than the API | 23:49 |
| devananda | jroll: why is there a new REST API endpoint? | 23:49 |
| jroll | devananda: dtantsur|afk has a comment with: @Nisha, just for the record: we agreed on having a separate IntrospectionInterface. Also people seem to hate term discovery, should we call it 'introspect' everywhere (I'm ok with both)? | 23:56 |
| jroll | on patch set 46 | 23:56 |
| jroll | if that helps | 23:56 |
| devananda | jroll: thanks. but that wasn't my initial question | 23:56 |
| jroll | devananda: that's a great question... | 23:56 |
| devananda | jroll: why a new rest api endpoint? | 23:57 |
| jroll | devananda: that's a great question... | 23:57 |
| devananda | and why does PUT to /node/inspect result in a change to /node/state ??? | 23:57 |
| jroll | idk if I had enough coffee before that +2 | 23:57 |
| devananda | that seems ... obviously wrong | 23:57 |
| * jroll taks it back | 23:57 | |
| devananda | but maybe it's not obvious why it's right? | 23:57 |
| * jroll -1's | 23:58 | |
| jroll | I'm not sure | 23:58 |
| jroll | I think just nobody objected to it, ever | 23:59 |
| jroll | and maybe we all thought we had agreed on it :) | 23:59 |
| jroll | nish<tab> boo | 23:59 |
| * jroll steps outside for a few | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!