19:00:43 #startmeeting Ironic 19:00:43 Meeting started Mon Sep 29 19:00:43 2014 UTC and is due to finish in 60 minutes. The chair is devananda. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:44 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:47 The meeting name has been set to 'ironic' 19:00:50 hi all! 19:00:54 o/ 19:00:56 as usual, agenda is posted here: 19:00:58 #link https://wiki.openstack.org/wiki/Meetings/Ironic 19:01:01 morning / afternoon /eveing 19:01:24 o/ 19:01:30 #topic announcements 19:01:37 o/ 19:01:46 \o 19:01:52 Only one from me 19:02:12 I'm coordinating for us to have a presence in the Operator track on Monday in Paris 19:02:28 awesome! 19:02:30 nice! 19:02:40 if anyone is giving a talk that day, please ping me after the meeting so we can make sure the scheduling doesn't conflict 19:02:56 any other announcements from folks? 19:03:17 typing now 19:04:04 I would like to anmounce that will will be having some new folks joining our efforts. Please Welcome BertieFulton from SAP 19:04:17 \o BertieFulton 19:04:19 BertieFulton: hi! 19:04:20 BertieFulton, welcome :) 19:04:22 welcome 19:04:23 BertieFulton, o/ 19:04:35 Thanks everyone 19:04:41 :) 19:05:09 there may also be other folks from sap joining 19:05:28 BertieFulton: feel free to jump in, or just follow along, if taht suits you. there's somewhat of a structure to our meetings, with some open discussion at the end. 19:05:33 I don't see any of the other handles online at this time 19:05:43 #topic Juno RC status 19:05:51 #link https://launchpad.net/ironic/+milestone/juno-rc1 19:06:00 as you can see, we still haven't tagged the RC 19:06:04 which means Kilo is not open yet 19:06:12 that'll all happen when there are no open bugs on that page 19:06:47 several of us have been dog piling on the hash ring bug over the weekend 19:07:12 #link https://review.openstack.org/#/q/status:open+project:openstack/ironic+branch:master+topic:hash-ring,n,z 19:07:41 I'm going to keep working on that until it is done and merged 19:07:55 is there a chance https://bugs.launchpad.net/ironic/+bug/1310135 might get targeted to Juno? 19:07:56 yeah I hope tomorrow I will get something working for the take_over() 19:07:57 Launchpad bug 1310135 in nova "Stopping an instance via the Nova API when using the Nova Ironic driver incorrectly reports powerstate" [High,In progress] 19:07:57 reviews would be appreciated as it's the only thing blocking our release at the moment 19:07:57 devananda: so all four patches, including the one that is WIP? 19:08:03 I have sarted reviewing with the first patch in the set 19:08:20 but ... there are some other bugs that have been found that I wnt to brin gup 19:08:36 adam_g: that's one of them. we can target it, but I haven't coordinated that with Nova yet 19:08:55 mikal: I don't suppose you're around? 19:09:19 jogo: or perhaps you can help -- any chance we can target the fix for https://bugs.launchpad.net/ironic/+bug/1310135 to Juno? 19:09:20 Launchpad bug 1310135 in nova "Stopping an instance via the Nova API when using the Nova Ironic driver incorrectly reports powerstate" [High,In progress] 19:09:31 also not sure this is the correct place in the meeting but I would like to get more eyes on a couple of spec reviews before we open kilo https://review.openstack.org/#/c/103225 and https://review.openstack.org/#/c/121297 19:09:46 this is another bug that's also been requested that we get a fix in Juno - https://bugs.launchpad.net/ironic/+bug/1367182 19:09:48 Launchpad bug 1367182 in ironic "Ilo* drivers should use only ilo credentials to maintain compatibility across releases" [Medium,In progress] 19:10:19 I'm not thrilled with the patch they've proposed, however, so if it takes much longer to rework (longer than the other fixes we have in flight) I dont want to delay the release for it 19:10:43 well, what they have now is not perfect, but it's something to start with 19:10:45 I'm fine with it landing, if a reasonable fix is proposed in time 19:11:11 the fix would be to remove the namespacing from the driver options? 19:11:17 dtantsur: it still exposes ipmi_* properties in the ilo driver, though 19:11:19 at least we won't have to make too many com[atibility layers, if we decide to switch to something shared (like just username or bmc_username) 19:11:21 or using ipmi_ for ilo 19:11:22 there's a workaround for that ilo bug though, right? ie, put in ipmi credentials. 19:11:48 rloo: yes. the bug is just that the ilo driver -also- rquiers the user to put in ipmi user/pass 19:11:53 which are effectively the same values 19:11:58 devananda, but it no longer _requires_ them at least, right? 19:12:09 so it's an API bug, essentially. not a functionality bug. 19:12:23 devananda: right, so still can use. just not great. the hash-ring stuff though seems more urgent to fix. 19:12:30 rloo: right 19:12:31 deva, 19:12:38 rloo, +1 19:12:40 dtantsur: it requires them but automatically adds them for you 19:12:55 devananda, IIRC it does not call node.save() 19:13:03 so they're not persisted 19:13:07 (I may be wrong) 19:13:19 deva, we have teh fix for 1367182 already. we will submit that for review if upsteam is willing to consider it for juno. 19:13:50 wanyen, isn't it https://review.openstack.org/#/c/124704/ ? 19:14:23 our last proposal was to dynamically convert ilo credentail to ipmi credential 19:14:45 dtantsur: ah, if it's not persisted, then I thnk it's fine 19:15:10 adam_g: on the Nova bug, what do you think of at least landing the Ironic side of the fix? 19:15:37 devananda, I checked - it's not. so unless save() is called anywhere else, we're fine 19:15:40 devananda, i think that woudl narrow the race for sure 19:15:45 are there any other bugs that folks feel should be targeted to Juno? 19:16:04 hmm I think I got one 19:16:05 lemme see 19:16:25 https://launchpad.net/bugs/1373671 19:16:26 Launchpad bug 1373671 in ironic "SSH driver does not work with non-english locale" [High,In progress] 19:16:34 lucasagomes: ++ 19:16:42 there's a fix in the review already, and tripleo ci have been rebuilt 19:16:44 right, quite bad one 19:16:47 ok tagging 19:17:16 I guess we'll land it long before hashring series :) 19:17:35 the hash ring changes, up to but not including the periodic task, are probably ready to land now 19:17:46 at least IMO :) 19:18:09 devananda, is the "online" field better documented? I will take a look after the meeting 19:18:13 cause that was a bit confusing 19:18:26 ++ 19:18:30 lucasagomes: I posted that just before the meeting. but i'm happy to add more inline docs to it or change the name 19:18:47 dtantsur: yes https://review.openstack.org/#/c/124704/ 19:18:48 devananda, ack, grand them :) 19:18:50 cheers 19:18:54 then* 19:19:02 devananda: if there are minor issues with the hash ring (not that I've found any) better to land and address with anothwer patch or fix before landng 19:19:32 NobodyCam: if we find them, fix before landing, only because this is a release blocker 19:19:42 ack :) 19:19:48 meaning that the release may be tagged at any time after all the targeted bugs are closed 19:19:59 so if we need to do a follow on patch, we actually need to target it too 19:20:11 *we would actually ... 19:20:23 to communicate to the openstack release team that we aren't ready for them to release 19:20:35 ahh yes very good point 19:20:55 any other questions on the release process, targeted bugs, etc? 19:20:57 there's still rc2 or 3 before actual release, although it gets more onorous (sp) to fix bugs then, right? 19:21:12 rloo: we should treat RC1 as though it's final 19:21:26 rloo: and not tag it unless we actually dont know of any more blocking bugs 19:21:27 devananda: of course! 19:21:40 what happens when we tag RC1 is that we re-open master for Kilo development 19:21:54 and after that, any bug fixes taht need to be applied to Juno need to be handled separately as a backport 19:22:24 there's a deadline to tag rc1 right? 19:22:27 * lucasagomes checks wiki 19:22:30 rloo: so, hopefully, there actually won't be any rc2 or rc3 :) 19:22:50 lucasagomes: the goal was last thursday. the deadline is when Juno final is released 19:22:50 #link https://wiki.openstack.org/wiki/Juno_Release_Schedule 19:23:00 gotcha 19:23:07 fair enuff 19:23:33 ok, moving on 19:23:36 #topic Kilo planning 19:23:38 deva, so can we target https://bugs.launchpad.net/ironic/+bug/1367182 for RC1? 19:23:39 Launchpad bug 1367182 in ironic "Ilo* drivers should use only ilo credentials to maintain compatibility across releases" [Medium,In progress] 19:24:10 wanyen: I already did :) 19:24:22 two things for this topic 19:24:25 deav, ty 19:24:35 s/deav/deva 19:24:56 - kilo summit 19:24:59 - kilo specs 19:25:25 NobodyCam asked if we could open kilo specs today. I think that's fair 19:25:32 though I still wont look at them until Juno is tagged :) 19:25:45 I'm not against some good read :) 19:25:56 yeah, and lets remember that we've changed the way kilo specs are proposed 19:26:03 we better update the template first etc 19:26:08 lucasagomes: right 19:26:14 we should land before we open kilo https://review.openstack.org/#/c/103225 and https://review.openstack.org/#/c/121297 19:26:21 I also think we should follow the lessons we learned (same ones Nova learned, too) 19:26:36 specs are kinda heavy handed, and some patches just don't require a spec 19:26:43 eg, cleanup and refactoring 19:27:15 but don't be afrai to say "this change should be discussed, please propose a spec" if you feel that it would benefit from that 19:27:30 how about the stuff we talked about with the initial spec being just a paragraph or two 19:27:48 NobodyCam: yep. for things that are probably larger, that's definitely good 19:27:55 deva, uefi spec already approved in Juno but we would like to do more in Kilo such as adding uefi support for ilo virtual meida ipa driver, do we need to submit a new spec for kilo? 19:28:05 wanyen: yes 19:28:13 devananda: we should send an email to the list about these things, when we open specs for kilo 19:28:16 deva, ok. 19:28:20 about the short initial spec etc 19:28:21 jroll, +1 19:28:28 wanyen: though it might be fairly simple, eg. just refer to the already approved spec and describe any differences 19:28:32 so everyone is up to date on process 19:28:37 jroll: indeed we should. sounds like you are volunteering :) 19:28:46 bah 19:28:50 * jroll stops talking 19:28:52 lol 19:28:52 there was an email after the mid-cycle but it would be good to reinforce it 19:28:54 jroll: lol 19:28:57 jroll, hah 19:29:02 devananda: I can do that, I'll run a draft past you before I do 19:29:25 jroll: thanks. also that seems to be the sort of thing JayF likes to do 19:29:32 oh 19:29:37 * jroll steers his bus toward JayF 19:29:37 deva, okay. We will do that . 19:29:38 # action for jroll ? 19:29:45 NobodyCam: yeah, that's fine 19:29:46 Hey, I thought I had to say things to be volunteered to do things 19:29:57 JayF: you've said things in the past. that's good enough :) 19:30:04 You've removed a serious tool in getting me to stay quiet ;) 19:30:12 \o/ 19:30:31 * devananda likes it when everyoen has things to say 19:30:43 * JayF nominates jroll to do the email thing 19:31:00 #agreed kilo specs to re-open this week, preceeded by an email describing the changes over Juno process 19:31:04 heh seems nobody really likes to send emails to the list 19:31:23 lucasagomes: words are hard :) 19:31:33 jroll, yup 19:31:36 #agreed Kilo specs may be initially done as a light-weight proposal only, to guage direction and interest before investing lots of time in a full spec 19:31:58 may be or should be? 19:32:07 * dtantsur likes nit-picking 19:32:10 may be 19:32:11 may be 19:32:14 ack 19:32:19 I dislike process for process' sake 19:32:24 ++ 19:32:29 +! 19:32:32 ++ 19:32:36 so if I know this thing is an approved direction, doing a light weight spec is still just a waste of time 19:32:38 dtantsur: it's more of "if you are not sure that people will care at all about this feature" 19:32:40 anyway :) 19:33:05 heh it would help me a lot with my Juno discovery efforts :) 19:33:11 any questions on Kilo at this point, without going into a discussion of actual features (lets save that for the specs or ML) 19:34:27 I'm happy to see the gDoc getting longer :) 19:34:44 oh, for reference, here's the link again for the summit planning 19:34:45 #link https://docs.google.com/spreadsheets/d/1XBKdeDeGfaRYaThjIIoYRwe_zPensECnxsKUuqdoVmQ/edit#gid=0 19:35:22 ok, moving on then 19:35:28 #topic subteam reports 19:35:37 heya 19:35:47 so the parallel tempest jobs are now part of our check pipeline, non-voting 19:35:48 adam_g, dtantsur, GheRivero - hi! how's everything? 19:35:51 devananda: anything more about us having a presence in the Operator track on Monday in Paris 19:35:54 oh nm 19:36:05 and the sideways grenade job should be showing up there soon too https://review.openstack.org/#/c/124562/ 19:36:07 NobodyCam: not yet. I'll pass on details when I have them 19:36:20 nothing new for Juno 19:36:26 adam_g: awesome work! thanks! 19:36:42 enabling parallel tempest tests tickles the race condition addressed by https://launchpad.net/bugs/1310135 19:36:43 Launchpad bug 1310135 in nova "Stopping an instance via the Nova API when using the Nova Ironic driver incorrectly reports powerstate" [High,In progress] 19:36:52 once that is addressed the job should be more stable and we can make it voting 19:37:18 adam_g: I'd like to keep it non-voting for a few weeks to get some more data in logstash 19:37:28 devananda, +1 19:37:28 adam_g: so we can spot other races it may be exposing 19:37:54 +1 19:38:04 jroll, did you ever make any progress on getting the agent ramdisk booting with lower memory? 19:38:04 adam_g: do you know if its logs are going to be available there already? I haven't actualy checked ... 19:38:39 adam_g: not really, I think DIB images will be our only option, I got pulled into some other things 19:38:39 devananda, for those tests? yeah, they should all be available in logstash AFAIK 19:38:50 jroll, ah, ok 19:38:51 #info tempest parallel test has exposed https://launchpad.net/bugs/1310135 - fixes are proposed to nova and ironic 19:38:53 Launchpad bug 1310135 in nova "Stopping an instance via the Nova API when using the Nova Ironic driver incorrectly reports powerstate" [High,In progress] 19:39:01 adam_g: still on my radar to get done asap 19:39:19 #info tempest parallel test shoudl stay non-voting for a few weeks of stability while we gather logstash data and ensure it is stable before enabling voting 19:39:35 adam_g: cool 19:39:54 GheRivero: thanks! no news is good news in this case :) 19:40:02 ++ 19:40:18 jroll: anything else to report on IPA? 19:40:32 devananda: I don't think so; JayF anything from you? 19:40:45 linggao: hi! anything to report on ipminative / pyghmi ? 19:40:59 I don't believe I have anything to report either. 19:41:01 wanyen: hi! anything (besides the bug we already covered) to report on the iLo driver? 19:41:02 I would mention that 19:41:02 hi devananda, nop 19:41:13 w/r/t IPA tempest testing, getting it running in less ram requires getting DIB elements 19:41:17 linggao: just big thank you fir check-ironic-xcat-third-party 19:41:20 and I believe that NobodyCam is working on that atm 19:41:30 we submitted ios script to dib. we did tsetingfor bios-only, legacy and uefi mode. 19:41:33 JayF: ah, cool. NobodyCam, how's that going? 19:41:37 s/ios/iso 19:41:52 wanyen: nice! can you paste a link to that review here for us? 19:42:00 NobodyCam, glad to contribute something. 19:42:01 JayF: devananda: yuriy was also working on that :) 19:42:19 wanyen: you probably know this but for the sake of the meeting, the `make iso` target landed for the IPA CoreOS builder 19:42:20 (neither work yet) 19:42:46 devananda: I will prob be picking up from yuriz work here: https://review.openstack.org/#/c/103105 19:43:20 NobodyCam: ++ 19:43:46 anyone wants bug numbers? :) 19:43:55 * NobodyCam does 19:43:56 yes? 19:44:04 Open: 94 (-6). 3 new (0), 26 in progress (-6), 0 critical, 12 high (+2) and 4 incomplete (0); juno-rc1: 6 (-9) 19:44:18 so good job closing tasks in progress 19:44:18 oh luv the - 19:44:21 dtantsur: of course :) thought I pinged you at the start of this topic, sorry 19:44:31 :) nice 19:44:37 eav, link https://review.openstack.org/#/c/123642/ 19:44:46 also, we reduced number of open bugs by ~40 during this month 19:44:57 o/ 19:44:58 (also thanks to nova folks taking part of our headache)) 19:45:10 #info ISO target for iLO driver proposed to disk-image-builder at https://review.openstack.org/#/c/123642/ 19:45:11 great work everyone!!!! 19:45:23 I see open bugs <100 as a great milestone 19:45:31 JayF: yes. Tx 19:45:32 and that's all from me :) 19:45:41 dtantsur: that reminds me, for bugs that affect Ironic but where the fix is only necessary within the Nova driver 19:45:48 dtantsur, cheers, +1 for the < 100 ob 19:45:49 dtantsur: we should probably mark the Ironic side as wont-fix 19:45:52 I did that for several last week 19:46:14 devananda, me too some time ago, but I didn't finish back then 19:46:14 otherwise they are likely to be left dangling and open on our side, and require manually closing them later 19:46:26 dtantsur: cool, thanks. I think we're good now :) 19:46:34 won't fix or invalid? 19:46:34 devananda, I'll revisit bug list this week just in case 19:47:19 lucasagomes: I'm going to go with wontfix ... which I think is in line with what nova wants 19:47:19 it sounds more like invalid for me because the code is not in our tree so can't be fixed in ironic 19:47:26 right 19:47:32 but the bug is not itself invalid 19:47:35 just to confirm in case I see some and have to close 19:47:38 it's just not fixable within this project 19:47:42 devananda, invalid for ironic I meant 19:47:45 oh 19:47:58 right, yeah... I happy with any, just to know which one to use 19:48:09 yea. i'm not sure it matters, but we should be consistent 19:48:14 so let's go with "wontfix" :) 19:48:19 ok :) 19:48:34 also, following sdague's email about nova bug statuses, and the edits to the openstack Bug wiki page 19:48:50 please start using status:confirmed for bugs taht we have confimred 19:49:11 and status:triaged *only* if the bug report contains a suggested solution/fix, but not any code 19:49:47 #topic open discussion 19:49:54 devananda, right, I also noticed that. Launchpad has different definition :( 19:49:54 hi , have anyone launched instance in ironic using ubuntu/centos image? 19:49:55 look! 10 minutes of open discussion :) 19:50:11 BertieFulton: how was your first meeting. Can I ask are there any areas you would to focus your efforts on began with? 19:50:20 harshada_kakad: hi! I don't understand your question? 19:50:53 harshada_kakad, you mean ubuntu being the host os for the ironic services and centos the tenant image? 19:50:58 i want to launch baremetal instance using image ubuntu/centos /// 19:51:15 harshada_kakad: it has been done, I don't see what the issue would be 19:51:19 NobodyCam: Very informative would be the best description. 19:51:22 harshada_kakad: ironic just writes an image to disk 19:51:30 harshada_kakad, ok, but did u try and it fail? what happened? 19:51:55 harshada_kakad: ubuntu and centos are different distributions of linux. Ironic can write out -any- disk image, not just those. 19:52:11 i m trying to get kernel and initeramfs using disk-image builder for centos7.. 19:52:39 harshada_kakad, right, I think you should talk to the guys at #tripleo about it 19:52:53 but it fails as there is no busybox for centos7 ... 19:53:09 harshada_kakad: the developer meeting is not the best time to ask usage questions. Please ask usage questions on the openstack@ mailing list 19:53:15 and due to which cnt launch instance using centos 19:53:17 BertieFulton: : I know the first meeting can be overwhelming 19:53:32 ok 19:53:45 BertieFulton: We're here (mostly in #openstack-ironic) to help 19:54:17 NobodyCam: I think we have a number of areas we would like to focus our efforts on. Boot from Volume is a big area of interest for us 19:54:49 so, I dunno about everyone else, but I'm really looking forward to Paris :) 19:54:56 o/ 19:55:09 BertieFulton: there have been several discussions about boot from volume, along two avenues 19:55:14 BertieFulton, boot from SAN volume? 19:55:30 wanyen: and NFS 19:55:36 BertieFulton: a) using iPXE to mount an iSCSI volume from the ramdisk, then boot from that 19:55:42 BertieFulton: that would a great area to jump in with ... but as you can see ^^^^ also a very active area 19:56:09 ^ ! /me has big plans for ipxe for K 19:56:10 BertieFulton: using an FC card and HBA to attach at the hardware level, then config the machine's bios to boot from that 19:56:13 NobodyCam: I'm thinking I maybe should backtrack and fix some bugs for a while :) 19:56:41 lucasagomes: BertieFulton: I would love boot from block from block with ipxe support 19:56:51 * JayF needs to start hacking on bonding support for pxe 19:56:53 lol 19:56:57 BertieFulton: starting with bug fixes is great! look for the low-hanging-fruit tag 19:57:30 BertieFulton: https://bugs.launchpad.net/ironic/+bugs?field.tag=low-hanging-fruit 19:57:31 JayF, yeah once we separate the boot interface from deploy interface that will be easier 19:57:36 oh we have a LHF tag... kewl... /me learns something new 19:58:16 devananda: thanks, will do :) 19:58:30 * devananda reminds the bug team to use the LHF tag :) 19:58:54 two minutes 19:59:12 might as well end a minute early 19:59:17 cheers - thanks everyone! 19:59:23 tx 19:59:23 thanks everyone 19:59:25 Thanks very much 19:59:30 thanks y'all :) 19:59:35 #endmeeting