| *** rongze has quit IRC | 00:03 | |
| lifeless | hmmm, triple-cd is awol | 00:14 |
|---|---|---|
| lifeless | oj look, ndercloud fail | 00:14 |
| lifeless | restarted | 00:15 |
| *** cd-undercloud has joined #tripleo | 00:23 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 00:23 |
| *** cd-undercloud has quit IRC | 00:23 | |
| lifeless | and it's unhappy :(.... | 00:23 |
| *** cd-undercloud has joined #tripleo | 00:32 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 00:32 |
| *** cd-undercloud has quit IRC | 00:32 | |
| *** ohadlevy has quit IRC | 00:35 | |
| *** ohadlevy has joined #tripleo | 00:36 | |
| lifeless | really unhappy | 00:36 |
| *** ohadlevy is now known as Guest40138 | 00:36 | |
| *** cd-undercloud has joined #tripleo | 00:42 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 00:42 |
| *** cd-undercloud has quit IRC | 00:42 | |
| *** jcooley_ has joined #tripleo | 00:48 | |
| *** jcooley_ has quit IRC | 00:50 | |
| *** cd-undercloud has joined #tripleo | 00:52 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 00:52 |
| *** cd-undercloud has quit IRC | 00:52 | |
| *** jcooley_ has joined #tripleo | 00:52 | |
| *** sdake has quit IRC | 01:00 | |
| *** cd-undercloud has joined #tripleo | 01:02 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 01:02 |
| *** cd-undercloud has quit IRC | 01:02 | |
| lifeless | nova list | 01:02 |
| lifeless | ERROR: The server has either erred or is incapable of performing the requested operation. (HTTP 500) (Request-ID req-2591518a-3955-4958-bf99-a66f6f800eac) | 01:02 |
| lifeless | thats from the undercloud | 01:02 |
| lifeless | 2013-12-09 01:03:04,232.232 16472 TRACE nova.api.openstack OperationalError: (OperationalError) (1030, 'Got error 28 from storage engine') 'SELECT anon_1.instances_created_at AS anon_1_instances_created_at, anon_1.instances_ | 01:03 |
| lifeless | thats our problem | 01:03 |
| *** lifeless changes topic to "CRITICAL: nova-api broken on undercloud: Using OpenStack to deploy OpenStack; meetings Tuesday 1900 UTC in #openstack-meeting-alt" | 01:03 | |
| *** jcooley_ has quit IRC | 01:08 | |
| *** docaedo has quit IRC | 01:08 | |
| *** jcooley_ has joined #tripleo | 01:11 | |
| *** cd-undercloud has joined #tripleo | 01:12 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 01:12 |
| *** cd-undercloud has quit IRC | 01:12 | |
| openstackgerrit | Jun Jie Nan proposed a change to openstack/os-apply-config: Check file existence before loading it and check errors https://review.openstack.org/60466 | 01:20 |
| *** tzumainn has joined #tripleo | 01:21 | |
| *** cd-undercloud has joined #tripleo | 01:22 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 01:22 |
| *** cd-undercloud has quit IRC | 01:22 | |
| *** sdake has joined #tripleo | 01:28 | |
| *** cd-undercloud has joined #tripleo | 01:32 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 01:32 |
| *** cd-undercloud has quit IRC | 01:32 | |
| lifeless | and ERROR 1030 (HY000): Got error 28 from storage engine | 01:33 |
| lifeless | from mysql itself | 01:33 |
| lifeless | when I manually run the query | 01:33 |
| lifeless | SpamapS: devananda: mordred: ^ ? | 01:33 |
| lifeless | my mysql naivete is showing up now | 01:33 |
| StevenK | lifeless: Apparently, it could be disk space | 01:34 |
| lifeless | /dev/sda1 19G 19G 0 100% / | 01:37 |
| lifeless | the sooner we get everythong on the state partition the better | 01:37 |
| lifeless | 500MB of gzipped log files deleted | 01:38 |
| lifeless | and we can list again. yay | 01:38 |
| lifeless | hopefully we'll get a happier build now | 01:38 |
| lifeless | 6G in /tmp | 01:38 |
| lifeless | also the new lock thing leaks lock files | 01:40 |
| *** CaptTofu has joined #tripleo | 01:40 | |
| lifeless | https://bugs.launchpad.net/tripleo/+bug/1259010 filed | 01:41 |
| uvirtbot | Launchpad bug 1259010 in tripleo "tripleo-cd script leaks lock files" [High,Triaged] | 01:41 |
| lifeless | ok, tripleo-cd started again after /tmp cleaned | 01:41 |
| *** julim has joined #tripleo | 01:45 | |
| *** nosnos has joined #tripleo | 01:48 | |
| *** nosnos has quit IRC | 01:48 | |
| *** nosnos has joined #tripleo | 01:49 | |
| *** julim has quit IRC | 01:55 | |
| *** jcooley_ has quit IRC | 01:55 | |
| *** jcooley_ has joined #tripleo | 02:03 | |
| *** docaedo has joined #tripleo | 02:14 | |
| *** rongze has joined #tripleo | 02:26 | |
| *** marun has joined #tripleo | 02:28 | |
| *** cd-undercloud has joined #tripleo | 02:32 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 02:32 |
| *** cd-undercloud has quit IRC | 02:32 | |
| *** marun has quit IRC | 02:32 | |
| StevenK | lifeless: Does that mean I win? | 02:33 |
| *** Guest40138 has quit IRC | 02:46 | |
| *** ohadlevy has joined #tripleo | 02:46 | |
| *** ohadlevy is now known as Guest78401 | 02:47 | |
| *** jcooley_ has quit IRC | 02:56 | |
| *** sdake has quit IRC | 02:58 | |
| *** jcooley_ has joined #tripleo | 03:04 | |
| lifeless | StevenK: apparently | 03:07 |
| *** csd_ is now known as csd | 03:13 | |
| *** sdake has joined #tripleo | 03:27 | |
| *** cd-undercloud has joined #tripleo | 03:28 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 03:28 |
| *** cd-undercloud has quit IRC | 03:28 | |
| asalkeld | lifeless, what was the library you pointed me to that does parsing? | 03:32 |
| asalkeld | can't remember, brain getting clogged | 03:33 |
| asalkeld | :( | 03:33 |
| asalkeld | no worries, I grep'd my irc log " <lifeless> you guys should totally look into parsley" | 03:39 |
| *** mwagner_lap has joined #tripleo | 04:11 | |
| *** Daviey has quit IRC | 04:12 | |
| SpamapS | hahahahaha | 04:12 |
| SpamapS | parsley makes me giggle | 04:12 |
| *** Daviey has joined #tripleo | 04:19 | |
| *** cd-undercloud has joined #tripleo | 04:21 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 04:21 |
| *** cd-undercloud has quit IRC | 04:21 | |
| *** nosnos_ has joined #tripleo | 04:28 | |
| *** nosnos has quit IRC | 04:31 | |
| *** AaronG_Zzz is now known as AaronGr | 04:38 | |
| *** AaronGr is now known as AaronG_Zzz | 04:42 | |
| *** sdake has quit IRC | 04:53 | |
| *** rongze has quit IRC | 04:54 | |
| *** rushiagr has joined #tripleo | 04:59 | |
| *** boris-42 has joined #tripleo | 05:03 | |
| *** cd-undercloud has joined #tripleo | 05:13 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 05:13 |
| *** cd-undercloud has quit IRC | 05:13 | |
| *** toci-bot has joined #tripleo | 05:16 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_PHxXIXt/ | 05:16 |
| *** toci-bot has quit IRC | 05:16 | |
| *** UtahDave has joined #tripleo | 05:19 | |
| *** Haomeng has joined #tripleo | 05:24 | |
| *** noslzzp has quit IRC | 05:30 | |
| *** rongze has joined #tripleo | 05:37 | |
| *** rpodolyaka1 has joined #tripleo | 05:41 | |
| rpodolyaka1 | morning tripleo | 05:41 |
| *** rongze has quit IRC | 05:42 | |
| *** marun has joined #tripleo | 05:53 | |
| lifeless | rpodolyaka1: morning! | 05:56 |
| *** nosnos_ has quit IRC | 05:57 | |
| *** nosnos has joined #tripleo | 05:58 | |
| *** akuznetsov has quit IRC | 06:01 | |
| *** akuznetsov has joined #tripleo | 06:03 | |
| *** akuznetsov has quit IRC | 06:05 | |
| *** xuhanp has joined #tripleo | 06:06 | |
| *** cd-undercloud has joined #tripleo | 06:12 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 06:12 |
| *** cd-undercloud has quit IRC | 06:12 | |
| lifeless | nuts | 06:13 |
| * rpodolyaka1 thought tripleo-cd had been working stably lately | 06:15 | |
| lifeless | rpodolyaka1: somewhat | 06:17 |
| lifeless | we ran out of disk space earlier | 06:18 |
| *** lifeless changes topic to "Using OpenStack to deploy OpenStack; meetings Tuesday 1900 UTC in #openstack-meeting-alt" | 06:18 | |
| *** tzumainn has quit IRC | 06:25 | |
| *** rameshg87 has joined #tripleo | 06:29 | |
| rameshg87 | Hi | 06:29 |
| rameshg87 | i had some questions regarding disk-image-builder. is this the right place to ask ? | 06:29 |
| lifeless | as good as any and better than most | 06:29 |
| rameshg87 | thanks :-) | 06:30 |
| rameshg87 | i just had a thought if we could enhance disk-image-builder to build iso images for installing the filesystem image onto the disk | 06:31 |
| rameshg87 | currently it creates whole disk images | 06:31 |
| rameshg87 | where as a new option could be added to create iso images which can directly install the filesystem image onto the disk | 06:32 |
| rameshg87 | how does that sound ? | 06:32 |
| lifeless | in principle, sure. We haven't done any discussion of that. | 06:32 |
| lifeless | though if you're going to install from the iso rather than running from it | 06:33 |
| *** Daviey has quit IRC | 06:33 | |
| lifeless | that seems like a layer you'd put on top of the image | 06:33 |
| lifeless | like a wrapper | 06:33 |
| lifeless | because the contents of the image don't need to be part of the installer | 06:33 |
| rameshg87 | yeah, i was thinking the ubuntu way, where we could pack the image into a squashfs filesystem | 06:34 |
| rameshg87 | and then use casper to either boot live from it | 06:34 |
| *** rpodolyaka1 has quit IRC | 06:34 | |
| rameshg87 | or trigger debian-installer to install from it | 06:34 |
| lifeless | sure | 06:35 |
| rameshg87 | okay.. | 06:35 |
| lifeless | though if you want the ubuntu way, you could also use live-build, which Debian and Ubuntu use to build their live images | 06:35 |
| rameshg87 | oh okay .. | 06:36 |
| rameshg87 | i was following this for my experiments: | 06:36 |
| rameshg87 | https://help.ubuntu.com/community/LiveCDCustomizationFromScratch | 06:36 |
| rameshg87 | will have a look at live-build scripts . | 06:37 |
| rameshg87 | is there something similar on how to create install cds for ubuntu/debian ? | 06:37 |
| rameshg87 | i know they put debian-installer into initramfs and let it trigger in the init scripts .. | 06:38 |
| rameshg87 | but is there something that is already in place that will help me to setup debian-installer into my initramfs with the root disk image ? | 06:38 |
| *** rongze has joined #tripleo | 06:39 | |
| *** arata has joined #tripleo | 06:44 | |
| *** rongze has quit IRC | 06:45 | |
| *** Daviey has joined #tripleo | 06:45 | |
| *** pblaho has joined #tripleo | 06:49 | |
| *** rongze has joined #tripleo | 06:59 | |
| *** nosnos has quit IRC | 07:01 | |
| *** nosnos has joined #tripleo | 07:01 | |
| *** cd-undercloud has joined #tripleo | 07:02 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 07:02 |
| *** cd-undercloud has quit IRC | 07:02 | |
| *** jprovazn has joined #tripleo | 07:16 | |
| *** jcooley_ has quit IRC | 07:18 | |
| openstackgerrit | A change was merged to openstack/os-apply-config: Check file existence before loading it and check errors https://review.openstack.org/60466 | 07:23 |
| lifeless | rameshg87: sorry, family time here | 07:24 |
| lifeless | rameshg87: all the debian tooling is open source and reusbale | 07:24 |
| lifeless | rameshg87: we didn't use it because it's a full OS installer and we wanted something much lighter weight - faster to build new images | 07:25 |
| lifeless | rameshg87: what are aiming to accomplish ? | 07:25 |
| *** jcooley_ has joined #tripleo | 07:29 | |
| *** xuhanp has quit IRC | 07:30 | |
| *** xuhanp has joined #tripleo | 07:30 | |
| rameshg87 | lifeless: thanks for responding back | 07:30 |
| rameshg87 | the requirement that i was thinking is like this: | 07:31 |
| rameshg87 | disk-image-builder prepares a root filesystem with whatever contents the user has chosen | 07:31 |
| rameshg87 | currently it turns that root filesystem into a qemu image | 07:32 |
| rameshg87 | instead of this, i was thinking if we can create an installer iso image which installs this "prepared root filesystem" onto a machine when booted | 07:32 |
| rameshg87 | for that i was investigating in the lines of using debian-installer | 07:33 |
| lifeless | ok | 07:33 |
| lifeless | why do you want to do that? | 07:33 |
| rameshg87 | just prepare the iso image so that it can booted on baremetal machines which support mounting remote media (like hp's ilo) | 07:34 |
| lifeless | sorry, let me try a different way of asking the question | 07:35 |
| lifeless | nova baremetal can deploy the qemu image directly | 07:35 |
| lifeless | with much less hands on effort than ilo with remote media | 07:35 |
| lifeless | why do you want to go qemu -> ISO -> remote media -> image on machine rather than qemu -> nova baremetal -> image on machine | 07:35 |
| rameshg87 | oh okay, now i get the question | 07:36 |
| rameshg87 | i was thinking like this: | 07:38 |
| rameshg87 | disk-image-builder -> prepares root filesystem -> turns into iso image -> put it into glance -> use the iso image for such machines | 07:38 |
| rameshg87 | is this the same thing you had meant ? | 07:39 |
| lifeless | no | 07:39 |
| lifeless | we deploy to HP machines without the ISO at all | 07:39 |
| lifeless | you don't need an ISO - nova baremetal deploys to HP hardware just fine | 07:39 |
| rameshg87 | okay .. | 07:40 |
| lifeless | so I'm just struggling to understand whats better about an ISO | 07:40 |
| lifeless | its manual | 07:40 |
| lifeless | which means its slower and more error prone | 07:40 |
| *** jcooley_ has quit IRC | 07:41 | |
| rameshg87 | yeah, i get it .. | 07:41 |
| *** lsmola has joined #tripleo | 07:41 | |
| *** jcooley_ has joined #tripleo | 07:42 | |
| rameshg87 | but if we could just use the iso to remote mount those images (like hp's ilo) | 07:42 |
| rameshg87 | would that be useful ? | 07:42 |
| lifeless | You tell me | 07:42 |
| rameshg87 | :-) | 07:43 |
| rameshg87 | i was just thinking if it would become an "alternative" | 07:43 |
| rameshg87 | does it sound like it is an overhead ? | 07:43 |
| lifeless | yes | 07:43 |
| lifeless | we'd have to duplicate a bunch of Ironic logic into the ramdisk | 07:44 |
| rameshg87 | okay .. | 07:44 |
| lifeless | so I'd look for a concrete use case to want to do it | 07:44 |
| lifeless | not just can we do it technically | 07:44 |
| lifeless | but what benefit does it give users | 07:44 |
| *** akuznetsov has joined #tripleo | 07:45 | |
| rameshg87 | okay .. i get it .. | 07:45 |
| rameshg87 | let me just rethink if it really makes sense .. | 07:45 |
| lifeless | rameshg87: a good place to start is with a problem statement | 07:46 |
| *** jcooley_ has quit IRC | 07:46 | |
| lifeless | rameshg87: Something of the form: "User X cannot do Y today." | 07:46 |
| rameshg87 | lifeless: thanks :-), i think i am missing that here .. | 07:49 |
| rameshg87 | thanks for your thoughts | 07:49 |
| lifeless | rameshg87: no probs :) | 07:54 |
| *** cd-undercloud has joined #tripleo | 07:57 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 07:57 |
| *** cd-undercloud has quit IRC | 07:57 | |
| *** rameshg87 has quit IRC | 08:02 | |
| *** xuhanp1 has joined #tripleo | 08:04 | |
| *** xuhanp has quit IRC | 08:05 | |
| *** jtomasek has joined #tripleo | 08:06 | |
| *** jcoufal has joined #tripleo | 08:07 | |
| *** hughsaunders has quit IRC | 08:09 | |
| *** marun has quit IRC | 08:19 | |
| *** hughsaunders has joined #tripleo | 08:24 | |
| *** marun has joined #tripleo | 08:35 | |
| *** rdopieralski has joined #tripleo | 08:41 | |
| *** ifarkas has joined #tripleo | 08:51 | |
| *** cd-undercloud has joined #tripleo | 08:52 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 08:52 |
| *** cd-undercloud has quit IRC | 08:52 | |
| *** jistr has joined #tripleo | 09:03 | |
| Ng | morning | 09:03 |
| lifeless | Ng: morning | 09:04 |
| lifeless | Ng: thanks for putting up the error overnight | 09:04 |
| lifeless | Ng: (in the topic) | 09:04 |
| lifeless | Ng: however, I couldn't see anything wrong with the mlx driver - what I did find was that nova-api was down in the undercloud | 09:04 |
| lifeless | Ng: it was getting error 28 out of mysql, which StevenK noted can be due to running out of spance | 09:05 |
| lifeless | Ng: which indeed the undercloud had no space on / | 09:05 |
| *** UtahDave has quit IRC | 09:05 | |
| *** bauzas has joined #tripleo | 09:07 | |
| *** panda has joined #tripleo | 09:11 | |
| *** vkozhukalov has joined #tripleo | 09:11 | |
| Ng | lifeless: heh, that wasn't overnight, I set that when mellanox went wonky on the 4th | 09:11 |
| *** vkozhukalov has joined #tripleo | 09:12 | |
| lifeless | Ng: lol :P | 09:12 |
| Ng | but it sounds like we could use some alerting on the undercloud :/ | 09:12 |
| *** toci-bot has joined #tripleo | 09:13 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_5RED8EY/ | 09:13 |
| *** toci-bot has quit IRC | 09:13 | |
| *** derekh has joined #tripleo | 09:25 | |
| *** lucasagomes has joined #tripleo | 09:25 | |
| *** hughsaunders has quit IRC | 09:28 | |
| *** hughsaunders has joined #tripleo | 09:32 | |
| *** rushiagr has quit IRC | 09:35 | |
| *** xuhanp1 has quit IRC | 09:40 | |
| *** xuhanp has joined #tripleo | 09:40 | |
| *** cd-undercloud has joined #tripleo | 09:45 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 09:45 |
| *** cd-undercloud has quit IRC | 09:45 | |
| *** hughsaunders has quit IRC | 09:52 | |
| *** rushiagr has joined #tripleo | 10:00 | |
| *** hughsaunders has joined #tripleo | 10:02 | |
| *** max_lobur_afk is now known as max_lobur | 10:02 | |
| *** nosnos has quit IRC | 10:07 | |
| *** nosnos has joined #tripleo | 10:07 | |
| *** akrivoka has joined #tripleo | 10:10 | |
| *** kmsWork_ has quit IRC | 10:10 | |
| *** jeremydei has quit IRC | 10:11 | |
| *** xuhanp has quit IRC | 10:11 | |
| *** jeremydei has joined #tripleo | 10:20 | |
| *** pblaho has quit IRC | 10:23 | |
| *** athomas has joined #tripleo | 10:28 | |
| *** cd-undercloud has joined #tripleo | 10:39 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 10:39 |
| *** cd-undercloud has quit IRC | 10:39 | |
| lsmola | marios_, ping | 10:46 |
| *** akrivoka has quit IRC | 11:10 | |
| * Ng pops out for errands and lunch | 11:12 | |
| *** rushiagr has quit IRC | 11:14 | |
| *** akrivoka has joined #tripleo | 11:17 | |
| *** rongze has quit IRC | 11:17 | |
| *** arata has left #tripleo | 11:19 | |
| *** rongze has joined #tripleo | 11:26 | |
| *** markmc has joined #tripleo | 11:26 | |
| *** cd-undercloud has joined #tripleo | 11:29 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 11:29 |
| *** cd-undercloud has quit IRC | 11:29 | |
| *** toci-bot has joined #tripleo | 11:30 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_MASzXlW/ | 11:30 |
| *** toci-bot has quit IRC | 11:30 | |
| *** Male_21 has joined #tripleo | 11:33 | |
| *** jprovazn has quit IRC | 11:33 | |
| marios_ | lsmola: pong | 11:42 |
| lsmola | marios_, hello | 11:49 |
| lsmola | marios_, I have a few questions about Heat | 11:50 |
| lsmola | marios_, Are you the one who will be working on deployment with Heat templates? | 11:50 |
| lsmola | marios_, will it be part of Tuskar? | 11:50 |
| lsmola | marios_, I want discussing options with shardy , how we can deploy to multiple nodes, using only parameters | 11:52 |
| lsmola | marios_, so without need of adding force_nodes, or definition for each node | 11:52 |
| lsmola | marios_, so e.g. when I want to deploy 20 nova-compute nodes, I would just pass the number 20 | 11:53 |
| marios_ | lsmola: hey, so i am working on getting tuskar to work with tripleo-heat-templates... right now mostly about getting the plumbing in place: /#/c/56947/ (Make merge.py invokable), /#/c/58823/ (Make merge.py installable) and /#/c/52045/ (WIP : sketch of what using merge.py looks like for tuskar) this last one needs updating. also couple other related ones like /#/c/58229/ and /#/c/57210/ which need some more thought | 11:57 |
| lsmola | marios_, though if it will be that simple, and we won't be building any HEat template on the run, the code doesn't have to be in the tuskar-api, but can be right in the tuskar-ui | 11:58 |
| *** rongze has quit IRC | 11:58 | |
| marios_ | lsmola: though, _all_ of this is temporary, afaik. | 11:58 |
| marios_ | lsmola: i.e. eventually merge.py functionality should be pushed into heat | 11:58 |
| lsmola | marios_, yes, I have talked about that with shardy | 11:58 |
| marios_ | lsmola: for your question about n nodes, see /#/c/52045/ from above as an example | 11:58 |
| marios_ | but is still WIP so suggestions welcome | 11:59 |
| lsmola | marios_, ok, so you basically build a template for each node | 12:01 |
| lsmola | marios_, what we were talking about, that we could use one template and say how many of nodes there should be via heat autoscaling | 12:02 |
| marios_ | lsmola: kind of, you need a definition for each node and the nova-compute.yaml gets pulled in for each | 12:02 |
| lsmola | marios_, though thats in progress, I will past you the conversation, as you might be wiser from it :-D | 12:02 |
| marios_ | lsmola: great thanks for the heads up :) | 12:03 |
| lsmola | marios_, http://paste.openstack.org/show/54674/ | 12:04 |
| *** toci-bot has joined #tripleo | 12:07 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_ssrPvJJ/ | 12:07 |
| *** toci-bot has quit IRC | 12:07 | |
| lsmola | marios_, so, what is exactly the plan here, should we work on temporary solution? Or we should focus on how this should be done with heat? | 12:09 |
| lsmola | marios_, I guess it would have to land in i2, so we have a chance to implement that to UI | 12:10 |
| *** jprovazn has joined #tripleo | 12:11 | |
| marios_ | lsmola: on a call gimme few mate | 12:12 |
| lsmola | marios_, np :-) | 12:12 |
| *** cd-undercloud has joined #tripleo | 12:29 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 12:29 |
| *** cd-undercloud has quit IRC | 12:29 | |
| *** lucasagomes is now known as lucas-hungry | 12:34 | |
| *** rongze has joined #tripleo | 12:40 | |
| *** athomas has quit IRC | 12:43 | |
| *** julim has joined #tripleo | 12:48 | |
| *** hewbrocc` is now known as hewbrocca | 12:50 | |
| marios_ | lsmola: i think if you have the time to look at the 'doing it properly in heat' side would be good | 12:50 |
| marios_ | lsmola: i suspect there may be blueprints and that shardy may know about them | 12:50 |
| lsmola | marios_, hmm ok | 12:50 |
| lsmola | marios_, so nobody is looking onto that? | 12:51 |
| marios_ | lsmola: the 'temp' solution with merge.py is almost there (as in the plumbing required for merge.py/tht/tuskar) | 12:51 |
| lsmola | marios_, how many people are working on tripleo-heat ? | 12:51 |
| marios_ | lsmola: i was planning to move onto that but have not yet. so if you are interested/in position to, go for it. otherwise i or someone else will at some point | 12:51 |
| lsmola | marios_, well I would say there will be a big learning curve for me :-) | 12:52 |
| marios_ | lsmola: story of my life ;) | 12:52 |
| lsmola | hehe | 12:52 |
| *** max_lobur is now known as max_lobur_afk | 12:52 | |
| lsmola | marios_, so how many people are working on the tripleo-heat? IF only you, we should definitely put there more people. | 12:53 |
| lsmola | marios_, yeah I guess somebody will have to check with shardy about the deadlines, so we confirm when this will be done | 12:53 |
| lsmola | marios_, if it's I3, we should go the temporary way | 12:54 |
| lsmola | marios_, though AFAIK, we should focus on implementing the things in services it belongs to | 12:54 |
| lsmola | marios_, but without at least something working, we will have hard time to work with UI :-) | 12:56 |
| lsmola | marios_, btw. I see your implementation still works with racks, and possibly force nodes? | 12:56 |
| *** rbrady has joined #tripleo | 12:57 | |
| openstackgerrit | Mark McLoughlin proposed a change to openstack/diskimage-builder: Quieten disk-image-get-kernel https://review.openstack.org/60804 | 13:01 |
| *** athomas has joined #tripleo | 13:02 | |
| *** mwagner_lap has quit IRC | 13:11 | |
| *** toci-bot has joined #tripleo | 13:11 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_h7xcZTY/ | 13:11 |
| *** toci-bot has quit IRC | 13:11 | |
| *** nosnos has quit IRC | 13:13 | |
| *** marun has quit IRC | 13:15 | |
| *** vipuls has joined #tripleo | 13:22 | |
| *** marios has joined #tripleo | 13:22 | |
| *** vipul has quit IRC | 13:23 | |
| *** marios_ has quit IRC | 13:23 | |
| *** cd-undercloud has joined #tripleo | 13:24 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 13:24 |
| *** cd-undercloud has quit IRC | 13:24 | |
| *** rongze has quit IRC | 13:25 | |
| openstackgerrit | Jiri Tomasek proposed a change to openstack/tuskar-ui: List views for Nodes in Resources management https://review.openstack.org/60807 | 13:26 |
| mordred | lifeless: fyi - you track that down by: mordred@camelot:~$ perror 28 | 13:28 |
| mordred | OS error code 28: No space left on device | 13:28 |
| *** sballe has joined #tripleo | 13:28 | |
| *** lucas-hungry is now known as lucasagomes | 13:31 | |
| openstackgerrit | Mark McLoughlin proposed a change to openstack/diskimage-builder: Quieten disk-image-get-kernel https://review.openstack.org/60804 | 13:39 |
| *** rongze has joined #tripleo | 13:41 | |
| *** rushiagr has joined #tripleo | 13:46 | |
| *** jdob has joined #tripleo | 13:46 | |
| openstackgerrit | Jiri Tomasek proposed a change to openstack/tuskar-ui: List views for Nodes in Resources management https://review.openstack.org/60807 | 13:48 |
| *** dprince has joined #tripleo | 13:59 | |
| openstackgerrit | Radomir Dopieralski proposed a change to openstack/tuskar-ui: A dialog for registering nodes https://review.openstack.org/60531 | 13:59 |
| *** tzumainn has joined #tripleo | 14:02 | |
| jtomasek | marios: Imho the plan now it to talk to Ironic through Baremetal, although if this changes and Tuskar api will talk to Ironic directly I can adjust node listing to whatever is required | 14:05 |
| marios | jtomasek: thanks, was just wondering if that conversation was had already or what. i guess its for the future for now | 14:05 |
| jtomasek | marios: I would like to have that cleared out too | 14:06 |
| *** max_lobur_afk is now known as max_lobur | 14:06 | |
| *** toci-bot has joined #tripleo | 14:07 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_cMTTOPj/ | 14:07 |
| *** toci-bot has quit IRC | 14:07 | |
| *** jayg|g0n3 is now known as jayg | 14:16 | |
| *** cd-undercloud has joined #tripleo | 14:18 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 14:18 |
| *** cd-undercloud has quit IRC | 14:18 | |
| *** matty_dubs|gone is now known as matty_dubs | 14:21 | |
| openstackgerrit | Sushil Kumar proposed a change to openstack/diskimage-builder: Updates tox.ini to use new features https://review.openstack.org/60832 | 14:24 |
| *** ccrouch has joined #tripleo | 14:37 | |
| *** akuznetsov has quit IRC | 14:56 | |
| *** echohead has quit IRC | 15:01 | |
| *** LinuxJedi has quit IRC | 15:01 | |
| *** shadower has quit IRC | 15:01 | |
| *** rongze has quit IRC | 15:02 | |
| *** rongze has joined #tripleo | 15:03 | |
| *** echohead has joined #tripleo | 15:03 | |
| *** jdob has quit IRC | 15:05 | |
| *** jrist has joined #tripleo | 15:05 | |
| *** LinuxJedi has joined #tripleo | 15:07 | |
| *** shadower has joined #tripleo | 15:07 | |
| *** sdake has joined #tripleo | 15:07 | |
| *** sdake has joined #tripleo | 15:07 | |
| *** noslzzp has joined #tripleo | 15:08 | |
| *** LinuxJedi has quit IRC | 15:08 | |
| *** LinuxJedi has joined #tripleo | 15:08 | |
| *** toci-bot has joined #tripleo | 15:08 | |
| toci-bot | ERROR during toci run, see http://54.228.118.193/toci/toci_logs_PoVwjuW/ | 15:08 |
| *** toci-bot has quit IRC | 15:08 | |
| *** cd-undercloud has joined #tripleo | 15:09 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 15:09 |
| *** cd-undercloud has quit IRC | 15:09 | |
| *** athomas has quit IRC | 15:11 | |
| *** athomas has joined #tripleo | 15:20 | |
| *** jdob has joined #tripleo | 15:22 | |
| *** AaronG_Zzz is now known as AaronGr | 15:25 | |
| *** athomas has quit IRC | 15:28 | |
| *** athomas has joined #tripleo | 15:29 | |
| *** rongze has quit IRC | 15:30 | |
| *** rongze has joined #tripleo | 15:30 | |
| *** akuznetsov has joined #tripleo | 15:34 | |
| *** john-n-seattle2 has joined #tripleo | 15:37 | |
| *** john-n-seattle2 has left #tripleo | 15:37 | |
| derekh | dprince: pleia2 so I managed to loose access to te-1 (I'm pretty sure it was somthign stupid I did, so havn't tried to recover, unless you have something on te-1 that you want I'll zapp it | 15:39 |
| derekh | dprince: pleia2 so I;ve set up te-2, which has everything setup the same, and have started to work on the neutron stuff | 15:39 |
| derekh | dprince: pleia2 eth2 is moved onto a bridge and I'll see if I can figure out how to get seed nodes getting IP's from neutron now | 15:40 |
| *** jergerber has joined #tripleo | 15:41 | |
| dprince | derekh: feel free to rebuild it, I had no data there | 15:46 |
| *** ifarkas has quit IRC | 15:50 | |
| derekh | dprince: thanks | 15:55 |
| openstackgerrit | Radomir Dopieralski proposed a change to openstack/tuskar-ui: Make the panels nested https://review.openstack.org/60860 | 15:56 |
| openstackgerrit | Radomir Dopieralski proposed a change to openstack/tuskar-ui: Make the panels nested https://review.openstack.org/60860 | 15:59 |
| *** john-n-seattle2 has joined #tripleo | 16:00 | |
| openstackgerrit | Radomir Dopieralski proposed a change to openstack/tuskar-ui: Make the panels nested https://review.openstack.org/60860 | 16:03 |
| *** cd-undercloud has joined #tripleo | 16:03 | |
| cd-undercloud | ************** overcloud complete status=1 ************ | 16:03 |
| *** cd-undercloud has quit IRC | 16:03 | |
| *** jpeeler has quit IRC | 16:11 | |
| *** jpeeler has joined #tripleo | 16:12 | |
| *** spzala has joined #tripleo | 16:13 | |
| SpamapS | noooo not =1 | 16:19 |
| *** vkozhukalov has quit IRC | 16:20 | |
| jdob | i still haven't found out what that message means yet, i'm guessing it's related to the automated testing? | 16:23 |
| openstackgerrit | Clint "SpamapS" Byrum proposed a change to openstack/tripleo-image-elements: Remove bash -x mode in parts of tripleo-cd.sh https://review.openstack.org/60865 | 16:24 |
| SpamapS | jdob: overcloud complete status you mean? | 16:24 |
| jdob | ya | 16:24 |
| SpamapS | jdob: yeah we have a live cloud that we deploy in a loop and that is the result | 16:24 |
| SpamapS | jdob: =0 is a success, =42 is a failure due to a known neutron race condition. =1 could be anything | 16:25 |
| SpamapS | this time it timed out booting a VM in the deployed cloud | 16:25 |
| jdob | so the undercloud stood up correctly but it failed to start one of the overcloud VMs? | 16:27 |
| SpamapS | jdob: right | 16:27 |
| SpamapS | not sure why yet | 16:27 |
| SpamapS | and we're about to lose the artifacts... | 16:27 |
| jdob | and the undercloud deploy is done off of master each time? | 16:27 |
| jdob | guess so :) | 16:27 |
| SpamapS | correction, we lost the artifacts 10 minutes ago :-P | 16:28 |
| * SpamapS wonders if we should start running an rsyslog server and then dump the output to .. wait... | 16:28 | |
| * SpamapS writes on the blackboard 100 times "I will not reimplement OpenStack infra..." | 16:28 | |
| *** tserong has quit IRC | 16:29 | |
| SpamapS | jdob: yes, master of all the services. libraries and stuff are still whatever is in requirements.txt | 16:29 |
| *** bauzas has quit IRC | 16:39 | |
| *** rongze has quit IRC | 16:46 | |
| *** csd has quit IRC | 16:51 | |
| openstackgerrit | A change was merged to openstack/diskimage-builder: Updates tox.ini to use new features https://review.openstack.org/60832 | 16:52 |
| *** matty_dubs is now known as matty_dubs|lunch | 16:54 | |
| *** jcooley_ has joined #tripleo | 16:57 | |
| *** cd-undercloud has joined #tripleo | 16:58 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 16:58 |
| *** cd-undercloud has quit IRC | 16:58 | |
| openstackgerrit | A change was merged to openstack/tripleo-image-elements: Remove bash -x mode in parts of tripleo-cd.sh https://review.openstack.org/60865 | 17:00 |
| *** jcoufal has quit IRC | 17:00 | |
| *** rdopieralski has quit IRC | 17:04 | |
| *** noslzzp has quit IRC | 17:06 | |
| *** ifarkas has joined #tripleo | 17:10 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/tripleo-incubator: Add Ironic to setup-clienttools https://review.openstack.org/60879 | 17:12 |
| derekh | dprince: pleia2 ok, got seed VM running on te-2, ip retrieved from neutron and boots and can be connected to from undercloud, now just need to sort out connectivity from overcloud, | 17:16 |
| pleia2 | derekh: great! | 17:17 |
| *** rongze has joined #tripleo | 17:17 | |
| dprince | derekh: nice | 17:17 |
| derekh | I think once thats done we can start scripting the instructions in the etherpad and bring up a job that simply boots a seed | 17:17 |
| pleia2 | sounds good | 17:17 |
| *** cody-somerville has joined #tripleo | 17:20 | |
| *** rpodolyaka1 has joined #tripleo | 17:20 | |
| pleia2 | I got the noop for running create-vm done on friday, that was actually very easy, now on "don't create image" | 17:21 |
| pleia2 | err, configure-vm | 17:21 |
| derekh | pleia2: sounds good | 17:22 |
| *** noslzzp has joined #tripleo | 17:24 | |
| *** jcoufal has joined #tripleo | 17:28 | |
| *** rongze has quit IRC | 17:28 | |
| *** jtomasek has quit IRC | 17:30 | |
| *** jcoufal has quit IRC | 17:32 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/tripleo-incubator: Add Ironic to setup-clienttools https://review.openstack.org/60879 | 17:34 |
| *** john-n-seattle2 has left #tripleo | 17:44 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/tripleo-incubator: Add Ironic to setup-clienttools https://review.openstack.org/60879 | 17:46 |
| *** cody-somerville has quit IRC | 17:47 | |
| *** bauzas has joined #tripleo | 17:48 | |
| *** cd-undercloud has joined #tripleo | 17:49 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 17:49 |
| *** cd-undercloud has quit IRC | 17:49 | |
| *** sdake_ has joined #tripleo | 17:52 | |
| *** sdake_ has joined #tripleo | 17:52 | |
| *** rongze has joined #tripleo | 17:54 | |
| *** jcooley_ has quit IRC | 17:54 | |
| *** cody-somerville has joined #tripleo | 17:57 | |
| *** athomas has quit IRC | 18:01 | |
| *** jistr has quit IRC | 18:02 | |
| *** derekh has quit IRC | 18:05 | |
| *** cody-somerville has quit IRC | 18:06 | |
| *** sdake has quit IRC | 18:08 | |
| *** sdake has joined #tripleo | 18:09 | |
| *** sdake has joined #tripleo | 18:09 | |
| *** dprince has quit IRC | 18:10 | |
| SpamapS | hm, do we have a way of making iptables rules stick on the cd-undercloud? | 18:12 |
| *** vkozhukalov has joined #tripleo | 18:20 | |
| *** ifarkas has quit IRC | 18:21 | |
| *** lucasagomes is now known as lucas-afk | 18:22 | |
| *** jcooley_ has joined #tripleo | 18:22 | |
| *** max_lobur is now known as max_lobur_afk | 18:23 | |
| *** krotscheck has joined #tripleo | 18:24 | |
| *** rongze has quit IRC | 18:37 | |
| openstackgerrit | A change was merged to openstack/tripleo-incubator: Add Ironic to setup-clienttools https://review.openstack.org/60879 | 18:37 |
| *** matty_dubs|lunch is now known as matty_dubs | 18:38 | |
| *** bauzas has quit IRC | 18:40 | |
| *** cd-undercloud has joined #tripleo | 18:40 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 18:40 |
| *** cd-undercloud has quit IRC | 18:40 | |
| *** noslzzp has quit IRC | 18:42 | |
| *** rushiagr has quit IRC | 18:49 | |
| *** noslzzp has joined #tripleo | 18:53 | |
| *** lucas-afk is now known as lucasagomes | 18:55 | |
| lifeless | SpamapS: not yet, AFAIK | 18:55 |
| lifeless | mordred: thanks | 18:56 |
| SpamapS | lifeless: I've been logging unexpected IP traffic on the public interface for the last week or so. | 19:00 |
| SpamapS | lifeless: would like to start dropping a lot of it.. but .. no point if we lose it all on reboot. | 19:01 |
| *** sdake has quit IRC | 19:04 | |
| lifeless | SpamapS: seems to me that in the short term we need a host firewall and in the longer term we should interrogate neutron for security group definitions for us and apply them locally. | 19:08 |
| *** jcooley_ has quit IRC | 19:11 | |
| SpamapS | lifeless: I like that idea. :) | 19:11 |
| *** sdake has joined #tripleo | 19:17 | |
| *** jcooley_ has joined #tripleo | 19:17 | |
| *** sdake has quit IRC | 19:32 | |
| *** sdake has joined #tripleo | 19:34 | |
| *** cd-undercloud has joined #tripleo | 19:34 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 19:34 |
| *** cd-undercloud has quit IRC | 19:34 | |
| *** AaronGr is now known as AaronGr_Lunch | 19:35 | |
| *** jcooley_ has quit IRC | 19:40 | |
| *** bauzas has joined #tripleo | 19:41 | |
| *** sdake has quit IRC | 19:46 | |
| *** Guest78401 is now known as ohadlevy | 19:51 | |
| *** ohadlevy has quit IRC | 19:51 | |
| *** ohadlevy has joined #tripleo | 19:51 | |
| *** epim has joined #tripleo | 20:01 | |
| *** rongze has joined #tripleo | 20:04 | |
| jdob | there's no way to edit a comment on a review request after you post it, is there? | 20:06 |
| *** AaronGr_Lunch is now known as AaronGr | 20:09 | |
| *** lucasagomes has quit IRC | 20:09 | |
| *** akrivoka has quit IRC | 20:09 | |
| *** sdake_ has quit IRC | 20:12 | |
| *** sdake_ has joined #tripleo | 20:13 | |
| lifeless | jdob: don't think so | 20:19 |
| openstackgerrit | Ralf Haferkamp proposed a change to openstack/diskimage-builder: local-config: Configure proxy for zypper repos https://review.openstack.org/60927 | 20:20 |
| *** markmc has quit IRC | 20:21 | |
| *** vkozhukalov has quit IRC | 20:28 | |
| *** cd-undercloud has joined #tripleo | 20:30 | |
| cd-undercloud | ************** overcloud complete status=42 ************ | 20:30 |
| *** cd-undercloud has quit IRC | 20:30 | |
| *** sdake has joined #tripleo | 20:32 | |
| *** sdake has joined #tripleo | 20:32 | |
| jdob | doh. ok, thanks lifeless | 20:33 |
| *** jcooley_ has joined #tripleo | 20:34 | |
| lifeless | SpamapS: hey, got a sec? I want to talk service enablement more | 20:36 |
| lifeless | and where is dan | 20:36 |
| *** gothicmindfood has joined #tripleo | 20:41 | |
| *** dprince has joined #tripleo | 20:45 | |
| *** jcooley_ has quit IRC | 20:46 | |
| *** cody-somerville has joined #tripleo | 20:47 | |
| *** cody-somerville has joined #tripleo | 20:47 | |
| *** jcooley_ has joined #tripleo | 20:47 | |
| dprince | lifeless: did you ping me on Google hangouts? | 20:48 |
| lifeless | I did | 20:48 |
| dprince | lifeless: whats up? | 20:49 |
| lifeless | I want to talk through the semantics of service enablement | 20:49 |
| lifeless | specifically in the context of cold reboots of datacentres | 20:49 |
| lifeless | which slipped my mind when we talked last week | 20:49 |
| lifeless | but is a fairly key use case | 20:49 |
| dprince | lifeless: okay. is now still a good time then. I've got my crappy headset today again. :( But we can deal with that I think. | 20:50 |
| lifeless | if we can find SpamapS | 20:50 |
| lifeless | I'd like to rope him in too | 20:50 |
| dprince | lifeless: tomorrow I'll be back in the office... so that works better for me if the two of you are up for it. | 20:50 |
| lifeless | dprince: either way is fine; I felt some urgency because I don't want you or him plowing effort into a faulty design | 20:51 |
| lifeless | the thing that reminded me was writing this mail - http://lists.openstack.org/pipermail/openstack-dev/2013-December/021668.html | 20:51 |
| lifeless | basically the cold power on story requires some degree of 'handle nothing being available when we restart' | 20:52 |
| lifeless | which is in tension with 'don't start services until we've actively gotten current configuration' | 20:52 |
| lifeless | so I'd like to have an approach outlined (doesn't need to be implemented) with the new stuff | 20:53 |
| lifeless | because | 20:53 |
| lifeless | I think the cold power on story has to take precedence - it's a fundamental | 20:53 |
| dprince | lifeless: My approach was always based around the fact that services would autostart once they have been configured once. It was the first boot scenario I had issues with. | 20:53 |
| lifeless | dprince: ok; so the thing that I learnt while discussing that with you was that we hadn't analyzed the hot reboot story properly | 20:54 |
| *** jcooley_ has quit IRC | 20:54 | |
| lifeless | because that has 'new software starting up with old configuration' which is a problem | 20:54 |
| *** marun has joined #tripleo | 20:55 | |
| *** jcooley_ has joined #tripleo | 20:55 | |
| lifeless | its that bit that has me convinced we need to do something different (the cold power on aspect still doesn't really bother me any which way - we could just zap the logs on first config :)) | 20:55 |
| dprince | lifeless: well. I would say it *could* have new software. Not necissarily that it *does*. | 20:55 |
| dprince | lifeless: and zapping logs is a 'sweep this problem under the rug' solution I think. | 20:56 |
| lifeless | but it's also that bit that is in tension, I think - like, how do we ensure there is valid config if the config sources have gone awol? | 20:56 |
| lifeless | so maybe it's an occ caching thing | 20:56 |
| lifeless | dprince: anyhow, just a thing to have in mind | 20:56 |
| *** jcooley_ has quit IRC | 20:57 | |
| dprince | lifeless: it would help as we discuss things if we had documented stories for these things. I think I follow you here but for example documenting somewhere the cold reboot, hot reboot, etc might speed our discussion along. | 20:58 |
| dprince | lifeless: similar to my previous misunderstandings about the "read only story" | 20:58 |
| dprince | lifeless: not saying it is all on you to write these up... just that we might could benifit from consolidated docs on these important use cases | 20:59 |
| lifeless | I agree | 20:59 |
| lifeless | want to put your understanding in some wiki pages / etherpads and we can go from there? | 20:59 |
| *** rpodolyaka1 has quit IRC | 21:00 | |
| dprince | lifeless: perhaps that is the way to start this, yes. And then we can have a google Hangout to argue about the details | 21:00 |
| lifeless | cool! | 21:01 |
| *** gothicmindfood has quit IRC | 21:04 | |
| *** jprovazn has quit IRC | 21:11 | |
| *** gothicmindfood has joined #tripleo | 21:15 | |
| SpamapS | lifeless: was at lunch. Back now. | 21:15 |
| lifeless | SpamapS: see scrollback | 21:16 |
| SpamapS | lifeless: reading | 21:16 |
| SpamapS | lifeless: hm | 21:17 |
| SpamapS | lifeless: perhaps cachedir should be /var/cache/os-collect-config instead of /var/run | 21:17 |
| SpamapS | lifeless: /var/cache is at least allowed to survive reboot | 21:17 |
| lifeless | SpamapS: perhaps, but that then runs new software with old config | 21:17 |
| lifeless | SpamapS: I'm leaning towards caching in occ | 21:17 |
| SpamapS | not sure what you mean | 21:18 |
| lifeless | I made two statements; which should I clarify? Or both? | 21:18 |
| SpamapS | lifeless: I get old/new software/config tension | 21:18 |
| SpamapS | lifeless: but caching in occ ?? | 21:18 |
| tzumainn | lifeless, ping | 21:19 |
| *** tserong has joined #tripleo | 21:21 | |
| SpamapS | lifeless: occ right now caches everything in /var/run/os-collect-config ... what are you suggesting should be cached? | 21:22 |
| lifeless | tzumainn: pong | 21:22 |
| lifeless | tzumainn: replying to your mail atm | 21:22 |
| *** cd-undercloud has joined #tripleo | 21:22 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 21:22 |
| *** cd-undercloud has quit IRC | 21:22 | |
| lifeless | SpamapS: I'm suggesting that if you reboot an occ using instance with no metadata sources available that occ should configure everything with the last successful metadata | 21:23 |
| tzumainn | lifeless, np, I'll wait for that then, I just thought it might be easier to clear up one of the issues more directly | 21:23 |
| lifeless | tzumainn: cool; lets do that - be just a sec | 21:23 |
| tzumainn | lifeless, sure, just let me know | 21:24 |
| SpamapS | lifeless: thats what I was suggesting. :) | 21:24 |
| SpamapS | lifeless: moving the cache to /var/cache/os-collect-config would achieve that. | 21:24 |
| SpamapS | I think | 21:24 |
| lifeless | tzumainn: sent | 21:25 |
| SpamapS | we actually have a stale config problem right now.. hm | 21:25 |
| SpamapS | we don't delete things from the cache if the collector for it fails or is not queried | 21:25 |
| lifeless | SpamapS: and separately, when you reboot an occ using instance and there is metadata available, nothing should start until occ has pulled down the current metadata | 21:25 |
| lifeless | 'nothing' meaning occ configured services | 21:26 |
| *** sdake has quit IRC | 21:26 | |
| *** sdake has joined #tripleo | 21:27 | |
| *** sdake has quit IRC | 21:27 | |
| *** sdake has joined #tripleo | 21:27 | |
| SpamapS | lifeless: the nothing should start is easy enough to achieve, we can make some meta attributes and use them, like if os-apply-config --key _os-collect-config.all_collected --type boolean ; then | 21:27 |
| SpamapS | to be clear, occ doesn't configure services :) | 21:27 |
| lifeless | SpamapS: I know it doesn't | 21:27 |
| lifeless | SpamapS: but think holistically | 21:28 |
| SpamapS | holistically occ should signal what the state of its cache is. | 21:28 |
| lifeless | SpamapS: do we want a new release of nova starting when we've done a rebuild until the new metadata is in place? | 21:28 |
| lifeless | SpamapS: this is about upstart jobs | 21:28 |
| lifeless | SpamapS: and the interaction between our layers | 21:28 |
| lifeless | and systemd | 21:28 |
| SpamapS | lifeless: right I get what we're aiming at. | 21:28 |
| SpamapS | lifeless: so to me occ should record the set of collectors asked for, and the ones that ended up in-use... | 21:29 |
| lifeless | dprince's concerns about first boot errors are a special case of 'software on system does not match configuration on system' | 21:29 |
| SpamapS | lifeless: if those are equal then we are fully configured. | 21:29 |
| lifeless | SpamapS: in fact, we had a case of the cold power on thing when I rebooted the undercloud | 21:30 |
| *** sdake_ has quit IRC | 21:30 | |
| SpamapS | ohhmmm... software and config versioning.. could we version expected config in elements and actual config in Heat templates? | 21:30 |
| SpamapS | lifeless: yes, it went horribly didn't it? :-P | 21:30 |
| lifeless | SpamapS: yes :) | 21:30 |
| lifeless | SpamapS: perhaps; one thing I thought about was the proposed uuid in the image | 21:30 |
| lifeless | and per-image services | 21:30 |
| lifeless | or per-image enablement or something | 21:31 |
| SpamapS | Yeah if we were recording the link between template <-> image somewhere that would work. | 21:31 |
| lifeless | there is a third layer which is metadata definition vs whats in the image | 21:31 |
| lifeless | for that though, I think we can ignore for now | 21:31 |
| lifeless | I'm assuming that at any point in time the user has a heat template consistent with the images they have asked to deploy | 21:32 |
| tzumainn | lifeless, thanks for replying to the email - do you have a moment to answer a question I have about manual assignment? | 21:32 |
| lifeless | absolutely | 21:32 |
| lifeless | I have another mail I have to send on that,w hen I find the right thread ;) | 21:32 |
| SpamapS | There's only one Metadata in Heat for any given resource. So we've talked about putting the image identifier in that metadata as the way to signal takeovernode. | 21:32 |
| lifeless | your question may well help me answer it more effectively! | 21:33 |
| tzumainn | lifeless, thanks! let me just see if I can phrase it right, I have trouble with phrasing sometimes :P | 21:33 |
| SpamapS | Its possible that same ID could drive the config vs. software service control as well. | 21:33 |
| SpamapS | Effectively if that ID isn't the same, you're half configured. | 21:33 |
| tzumainn | lifeless, anyway, so in the debate vs automated vs manual assignment, my understanding of the wireframes was that "manual assignment" was just using resource classes and node profiles to supply extra filter parameters to nova scheduler | 21:33 |
| lifeless | SpamapS: I have a couple of hesitations here though; what about vanilla image use - no uuid? | 21:33 |
| lifeless | SpamapS: and, maybe this is complecting things | 21:34 |
| tzumainn | if that's the case, do you have the same objections? | 21:34 |
| SpamapS | I think it is | 21:34 |
| SpamapS | lifeless: I'm sitting here trying to get out of a "respond to the problem" loop in my head :) | 21:34 |
| SpamapS | instead need to identify the actual problems | 21:34 |
| lifeless | yah | 21:36 |
| lifeless | tzumainn: so, the implementation of manual assignment is hugely concerning; the very idea of it also. | 21:36 |
| lifeless | tzumainn: it's like RAID - when was the last time a sysadmin manually said which disk in a RAID away was the hot spare? | 21:36 |
| *** openstackgerrit has quit IRC | 21:37 | |
| *** openstackgerrit has joined #tripleo | 21:37 | |
| lifeless | There is a meme being expressed that sysadmins want manual control, but none of the sysadmins I know want it. | 21:37 |
| tzumainn | lifeless, that's true if nodes are homogeneous, I had thought that node profiles would primarily be used for heterogeneous nodes | 21:38 |
| lifeless | [*] ok -very- mild exaggeration for effect. | 21:38 |
| lifeless | tzumainn: consider a machine with 8 4TB disks. | 21:38 |
| lifeless | tzumainn: a reasonable amount of ram, and a moderate amount of CPU. | 21:38 |
| dprince | lifeless: first boot: a very important case though. | 21:39 |
| lifeless | dprince: yes, I'm not suggesting we don't service it; I'm saying its a special case of a general problem we also haven't solved. | 21:39 |
| lifeless | dprince: so I'd like to try and solve the general case and then see whats left of the special case | 21:39 |
| lifeless | dprince: it might not be particularly special after that. Or it might be. | 21:39 |
| lifeless | tzumainn: we can use that machine as a cinder store if we RAID6 the disks, or as a swift store if we JBOD the disks.s | 21:40 |
| dprince | lifeless: well, I'm happy to talk about whatever case. But for me this just seems simple and straight forward. Don't start things until they have been configured (at least once) | 21:40 |
| lifeless | tzumainn: or cinder with JBOD'd disks if it has no hardware raid controller and we throw LVM / mdadm at them | 21:40 |
| dprince | lifeless: It is unclear if autostarting after that is a good or bad thing. | 21:41 |
| lifeless | dprince: sure; i'm rephrasing that as 'don't start things until they have configuration that matches their software' | 21:41 |
| lifeless | dprince: which I think solves it universally | 21:41 |
| lifeless | tzumainn: the question then, is who, or what, should decide that that machine becomes a cinder machine or a swift machine? | 21:41 |
| dprince | lifeless: Well. Once could argue that implementationally that leads to a house of cards. | 21:41 |
| lifeless | Implementational house of pancakes! | 21:42 |
| lifeless | </sorry> | 21:42 |
| *** boris-42 has quit IRC | 21:43 | |
| * SpamapS perks up | 21:43 | |
| SpamapS | pancakes? | 21:43 |
| SpamapS | where? | 21:43 |
| dprince | SpamapS: neutron ate the all | 21:44 |
| tzumainn | lifeless, okay, so - is the question here, if there's only one machine, who should decide whether it becomes a cinder or swift machine? | 21:44 |
| pleia2 | heh | 21:44 |
| * dprince neutron misteriously eats everything sometimes | 21:44 | |
| * dprince with a y even | 21:44 | |
| SpamapS | and then regurgitates them | 21:44 |
| lifeless | tzumainn: it's an exemplar of the problem | 21:44 |
| lifeless | tzumainn: more generally, we'll have M hardware configurations and N roles, and M{1, >>N}, but N {3-10} | 21:46 |
| tzumainn | lifeless, okay, sure - well, I'm not sure it's addressing the problem that others are raising, namely - if there are two machines with different attributes, one more suited for task X and one more suited for task Y - how can we make sure that the right machine gets the right task? | 21:46 |
| tzumainn | or are you saying that shouldn't be an issue | 21:46 |
| lifeless | thats not my understandind of the the problem others are raising | 21:46 |
| lifeless | so the general thing I'm pointing at is that the M hardware configurations are directly exposed as undercloud flavors. | 21:47 |
| lifeless | And that we need to tell heat which set of flavors, per role, are acceptable, and it's up to heat <-> nova to map it out. | 21:47 |
| lifeless | specifically we can't can't say that role maps to one flavor, because hardware shifts over time and we should expect enough variation that we can't mask it with deliberately inaccurate flavors | 21:48 |
| tzumainn | lifeless, I'm in agreement with that, I guess I had thought that others were raising the problem I had mentioned | 21:48 |
| lifeless | When Jarda says 'manual allocation' it's my understanding he is talking about the ability to say 'I want that machine there to be a control plane machine' | 21:48 |
| lifeless | irrespective of the presence or lack of distinguishing hardware characteristics | 21:49 |
| tzumainn | lifeless, yep, and I'm trying to go backwards to understand the true requirement, hoping that by defining that, we can come to a consensus | 21:49 |
| lifeless | yup | 21:49 |
| lifeless | me too :) | 21:49 |
| tzumainn | lifeless, so if I were to say that a) the process should never care about homogeneous nodes, and b) we will let resource classes and node profiles deal with heterogeneous nodes by passing in parameters to scheduler | 21:50 |
| tzumainn | would that fit your model? | 21:50 |
| lifeless | I worry about the meaning of 'passing in parameters to scheduler' | 21:51 |
| lifeless | because | 21:51 |
| *** rongze has quit IRC | 21:52 | |
| lifeless | that could mean 'tuskar implements a scheduler and then uses scheduler hints' or it could mean 'users specify the flavors to use and we pass that to heat' | 21:52 |
| lifeless | the former is a huge layering violation; the latter I think will work fine | 21:52 |
| tzumainn | lifeless, that's fair, and you're right that we need to be precise in the requirements - if I meant the latter would you be | 21:52 |
| tzumainn | ah, okay | 21:52 |
| lifeless | There may be many other approaches that work fine - I don't want to rule them out. But there's been at least one that doesn't, and I don't think it makes sense to do that one again. | 21:53 |
| lifeless | My focus is to ensure we're building something that enhances Nova rather than works around it | 21:53 |
| tzumainn | in that case, I might propose that I specify the above as a requirement, and challenge others to find a use case that it doesn't fulfill | 21:54 |
| lifeless | Cool :) | 21:54 |
| tzumainn | lifeless, thanks for talking through this, I like to think that everyone's coming closer to finding common ground | 21:55 |
| tzumainn | I'm going to update the requirements on the wiki to be far more specific, send that out, and specifically ask for objections regarding how assignment works | 21:56 |
| *** openstackgerrit has quit IRC | 21:57 | |
| *** openstackgerrit has joined #tripleo | 21:57 | |
| *** jdob has quit IRC | 22:00 | |
| lifeless | tzumainn: Jay's latest mail and my reply might help too | 22:07 |
| SpamapS | dprince: so first boot and "out of sync config and software boot".. | 22:09 |
| SpamapS | lifeless: ^^ | 22:09 |
| *** cody-somerville has quit IRC | 22:13 | |
| *** cd-undercloud has joined #tripleo | 22:17 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 22:17 |
| *** cd-undercloud has quit IRC | 22:17 | |
| *** rongze has joined #tripleo | 22:18 | |
| *** cody-somerville has joined #tripleo | 22:28 | |
| lifeless | yays | 22:28 |
| *** jergerber has quit IRC | 22:34 | |
| *** vipuls is now known as vipuls-away | 22:35 | |
| *** matty_dubs is now known as matty_dubs|gone | 22:45 | |
| *** dprince has quit IRC | 22:46 | |
| *** vipuls-away is now known as vipuls | 22:47 | |
| *** vipuls is now known as vipuls-away | 22:50 | |
| *** rongze has quit IRC | 22:51 | |
| *** epim has quit IRC | 22:54 | |
| *** epim has joined #tripleo | 22:54 | |
| *** panda_ has joined #tripleo | 22:57 | |
| *** panda has quit IRC | 22:58 | |
| *** vipuls-away is now known as vipuls | 22:58 | |
| *** jayg is now known as jayg|g0n3 | 23:01 | |
| *** gothicmindfood has quit IRC | 23:08 | |
| *** cd-undercloud has joined #tripleo | 23:11 | |
| cd-undercloud | ************** overcloud complete status=0 ************ | 23:11 |
| *** cd-undercloud has quit IRC | 23:11 | |
| *** epim has quit IRC | 23:15 | |
| *** bauzas has quit IRC | 23:28 | |
| *** gothicmindfood has joined #tripleo | 23:28 | |
| *** gothicmindfood has quit IRC | 23:30 | |
| *** gothicmindfood has joined #tripleo | 23:32 | |
| *** gothicmindfood has left #tripleo | 23:43 | |
| *** gothicmindfood has joined #tripleo | 23:45 | |
| *** gothicmindfood has quit IRC | 23:45 | |
| *** rongze has joined #tripleo | 23:51 | |
| openstackgerrit | Elizabeth Krumbach Joseph proposed a change to openstack/tripleo-incubator: Add image already exists option to boot-seed-vm https://review.openstack.org/60968 | 23:55 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!