| greghaynes | noama: sorry, I meant to w+1, not -1 | 00:17 |
|---|---|---|
| greghaynes | noama: fixed r | 00:17 |
| *** sudipto has joined #openstack-dib | 04:17 | |
| *** sudipto has quit IRC | 04:38 | |
| *** jaianshu has joined #openstack-dib | 04:39 | |
| *** prometheanfire has left #openstack-dib | 05:10 | |
| *** chhavi has joined #openstack-dib | 05:28 | |
| *** sudipto has joined #openstack-dib | 05:47 | |
| *** sudipto has quit IRC | 05:48 | |
| noama | greghaynes, can you provide more info on the implementation you suggested here https://review.openstack.org/#/c/392170/? | 05:50 |
| *** mgagne has quit IRC | 05:53 | |
| *** mgagne has joined #openstack-dib | 05:55 | |
| *** mgagne is now known as Guest51737 | 05:55 | |
| jaianshu | ianw: can you please have a relook at https://review.openstack.org/#/c/385880/. I have updated few scripts over here following yours and others sugesstions. | 06:04 |
| *** jaianshu_ has joined #openstack-dib | 06:48 | |
| *** remix_auei has joined #openstack-dib | 06:49 | |
| *** jaianshu has quit IRC | 06:52 | |
| *** remix_tj has quit IRC | 06:53 | |
| openstackgerrit | Merged openstack/diskimage-builder: add option to configure cloud-init to allow password authentication https://review.openstack.org/391765 | 06:54 |
| *** sudipto has joined #openstack-dib | 07:21 | |
| *** sudipto_ has joined #openstack-dib | 07:21 | |
| *** openstackgerrit has quit IRC | 08:03 | |
| *** brault has joined #openstack-dib | 08:09 | |
| *** chhavi has quit IRC | 08:16 | |
| *** chhavi has joined #openstack-dib | 08:27 | |
| *** chhavi has quit IRC | 09:00 | |
| *** chhavi has joined #openstack-dib | 09:01 | |
| *** remix_auei is now known as remix_tj | 09:07 | |
| *** jaianshu__ has joined #openstack-dib | 09:33 | |
| *** jaianshu_ has quit IRC | 09:33 | |
| *** jaianshu__ has quit IRC | 09:45 | |
| *** jaianshu__ has joined #openstack-dib | 09:45 | |
| *** chhavi has quit IRC | 09:49 | |
| *** pmannidi has quit IRC | 09:59 | |
| *** chhavi has joined #openstack-dib | 10:45 | |
| *** chhavi has quit IRC | 10:50 | |
| *** tushar_ has joined #openstack-dib | 11:16 | |
| tushar_ | hi | 11:19 |
| *** tushar_ has left #openstack-dib | 11:26 | |
| *** jaianshu_ has joined #openstack-dib | 11:52 | |
| *** fxpester has joined #openstack-dib | 11:53 | |
| fxpester | hi all | 11:53 |
| fxpester | I set up rhel7 image for ironic, but I got interface names new names with each reboot, so I installed `stable interface names` element | 11:54 |
| fxpester | I was expecting new names for interfaces, like ensXX or emXX, but I still got ethX - is this ok or it means that package is failed to do its job ? | 11:55 |
| *** jaianshu__ has quit IRC | 11:56 | |
| fxpester | ok... it`s failed... but package biosdevname is installed | 11:58 |
| *** sudipto has quit IRC | 12:04 | |
| *** sudipto_ has quit IRC | 12:04 | |
| fxpester | got net.ifnames=0 in grub.conf, `surprise motherfucker` ~_~ | 12:10 |
| *** tushar_ has joined #openstack-dib | 12:13 | |
| tushar_ | hii | 12:17 |
| *** owalsh has joined #openstack-dib | 13:09 | |
| tushar_ | hi | 13:57 |
| *** sudipto_ has joined #openstack-dib | 14:25 | |
| *** sudipto has joined #openstack-dib | 14:25 | |
| greghaynes | fxpester: is that something we need to set when using stable interface names? | 14:31 |
| greghaynes | fxpester: if so, can you file a bug? | 14:31 |
| fxpester | not sure actually, cause base dib package failed to build rhel7 image for me, I edit grub.conf by hands after dib error with grubby - so my bug report will not relate | 14:52 |
| greghaynes | oh, so is there another bug which causes dib to fail to make rhel7? | 14:53 |
| fxpester | it actually build it, but it`s not bootable | 14:55 |
| tushar_ | hi | 15:04 |
| tushar_ | I have launched the vm from diskimage....the instance is having hostname as "ubuntu" | 15:05 |
| tushar_ | the configuration file for diskimage is nodepool.yaml | 15:06 |
| tushar_ | How can I get the instance with different hostname than "ubuntu" ? | 15:07 |
| *** dmarlin is now known as dmarlin_ | 15:12 | |
| *** openstackgerrit has joined #openstack-dib | 15:25 | |
| openstackgerrit | Miles Gould proposed openstack/diskimage-builder: Allow installing proliant tools from an OS package https://review.openstack.org/344214 | 15:25 |
| *** jaianshu_ has quit IRC | 15:28 | |
| *** tushar_ has left #openstack-dib | 15:36 | |
| *** tushar_ has joined #openstack-dib | 15:40 | |
| greghaynes | tushar_: hrm, usually a tool like cloud-init/glean/dhcp client sets that | 15:53 |
| greghaynes | tushar_: I dont think we have a general way to bake it in to image, so youd have to make a custom element right now. It might be a good idea to support DIB_HOSTNAME as a new dib feature though | 15:54 |
| tushar_ | I have tried by adding the cloud-init under element section in nodepool.yaml... but it didnt work | 16:04 |
| *** fxpester has quit IRC | 16:05 | |
| *** jaianshu has joined #openstack-dib | 16:06 | |
| tushar_ | Earlier when I was working with Kilo release the vms were getting dynamic hostnames ...but now with the newton release i am facing this problem of hostname | 16:06 |
| greghaynes | right, so dib just makes the images, as far as what your cloud provides thats not really something dib can help with. It sounds like your dynamic network config isnt working though | 16:07 |
| greghaynes | you could also try using simple-init | 16:07 |
| tushar_ | ok I will try with simple-init | 16:08 |
| tushar_ | thank you Greghaynes | 16:09 |
| greghaynes | np | 16:10 |
| *** owalsh has quit IRC | 16:16 | |
| *** jaianshu has quit IRC | 16:18 | |
| tushar_ | simple-init didnt work | 16:23 |
| tushar_ | actually in newton neutron networking is there ...while kilo nova network was there | 16:24 |
| greghaynes | yea, this sounds like a cloud issue not a dib issue | 16:24 |
| tushar_ | one more thing i want to mention here ..if I am launching the instance with ubuntu image manually from dashboard ....it takes the hostname dynamically | 16:26 |
| tushar_ | the problem is only through nodepool | 16:26 |
| *** dmarlin has joined #openstack-dib | 16:26 | |
| greghaynes | ah, so if you need to use the neutron metadata service then youll have to use cloud-init | 16:35 |
| tushar_ | yeah | 16:36 |
| tushar_ | tried to configure cloud-init as well | 16:37 |
| tushar_ | but still something went wrong | 16:37 |
| greghaynes | ah, thats probably the thing to debug | 16:52 |
| tushar_ | yup | 16:52 |
| tushar_ | thank you | 16:53 |
| *** tushar_ has quit IRC | 16:55 | |
| *** chhavi has joined #openstack-dib | 17:03 | |
| openstackgerrit | Aparna proposed openstack/diskimage-builder: Update hpssacli to ssacli in proliant-tools element https://review.openstack.org/396504 | 17:22 |
| *** Guest51737 is now known as mgagne | 17:26 | |
| *** mgagne has quit IRC | 17:27 | |
| *** mgagne has joined #openstack-dib | 17:27 | |
| *** sudipto_ has quit IRC | 17:54 | |
| *** sudipto has quit IRC | 17:54 | |
| *** sudipto_ has joined #openstack-dib | 18:00 | |
| *** sudipto has joined #openstack-dib | 18:00 | |
| *** sudipto has quit IRC | 18:03 | |
| *** sudipto_ has quit IRC | 18:03 | |
| *** yolanda has quit IRC | 18:07 | |
| *** chhavi has quit IRC | 18:08 | |
| *** odyssey4me_ has joined #openstack-dib | 19:26 | |
| *** noama has quit IRC | 19:28 | |
| *** odyssey4me has quit IRC | 19:28 | |
| *** yolanda has joined #openstack-dib | 19:32 | |
| *** noama has joined #openstack-dib | 19:35 | |
| yolanda | hi greghaynes , ianw , can you take a look at https://review.openstack.org/#/c/401831/ ? | 19:38 |
| yolanda | it's a tripleo patch to fix CI, but slagle is concerned about how dib v2 is breaking tripleo workflow | 19:39 |
| yolanda | can you take a look at their comments and my code, and see if that makes sense, or can be improved somehow? | 19:39 |
| greghaynes | ianw: ^ Do you remember what all changed here? I agree that we shouldnt be breaking folks like this... | 19:45 |
| greghaynes | oh, we started returning a dict of element objects or somesuch | 19:46 |
| greghaynes | I think there should be an easy fix - make a new method which is the new api, keep the old method there as a mapping from the new format to the old | 19:47 |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Preserve backwards compat for expand_dependencies https://review.openstack.org/407176 | 20:05 |
| greghaynes | yolanda: ianw ^ | 20:05 |
| yolanda | hi greghaynes, thanks. What about the print of get_image_element_array function? | 20:13 |
| greghaynes | yolanda: why do you need that? | 20:13 |
| yolanda | i'm actually not sure if that's needed, just tried to replicate the code in dib | 20:13 |
| yolanda | but i can try to remove | 20:14 |
| greghaynes | Yes, I dont think you should need it | 20:14 |
| greghaynes | if you do then something very wierd is going on | 20:14 |
| yolanda | let me amend the patch, and depend on yours from dib | 20:14 |
| greghaynes | I think with mine the patch should be able to go away entirely | 20:17 |
| greghaynes | if you can test that | 20:17 |
| yolanda | well, that bits. The bit to point to lib directory is still needed | 20:17 |
| greghaynes | I actually dont understand why there needs to be an explicit point to lib | 20:18 |
| yolanda | because it went one level down | 20:18 |
| yolanda | and we removed that directory creation in the package | 20:18 |
| yolanda | ianw suggested we better pick with that python method | 20:18 |
| greghaynes | what fails if you just dont set _LIB? | 20:19 |
| yolanda | it was giving an error, cannot remember now but was failing | 20:19 |
| greghaynes | hrm, the reason we moved this stuff is so we could use python entrypoints for our scripts, the idea being once you pip install the scripts are accessible rather than having to much search paths | 20:20 |
| greghaynes | we might have missed something which is probably what failed | 20:20 |
| greghaynes | but it might be nice to fix that in dib too | 20:20 |
| yolanda | ok let's see first output of that patch now, and after that, i can try removing the os.environ['_LIB'] var, and see | 20:25 |
| yolanda | i need to update it slightly anyway, because right now instack gives an exception if the var is not set | 20:25 |
| yolanda | greghaynes, actually i think the depends-on needs to go in the other direction. your change in dib needs to depend on mine that sets up lib path for v2, or it won't pass the CI | 20:28 |
| yolanda | it's quite a chain | 20:28 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder: Preserve backwards compat for expand_dependencies https://review.openstack.org/407176 | 20:33 |
| ianw | hmm, i mean, _LIB is supposed to be private, it's right there in the "_" :) | 20:33 |
| yolanda | i don't get where is that used | 20:34 |
| ianw | yeah, IIRC that was really only used inside disk-image-create | 20:34 |
| greghaynes | yolanda: why do you want the dib patch to depend on the instack change? shouldnt it be the other way around? | 20:35 |
| yolanda | greghaynes, no because right now the code in instack fails if it cannot find _LIB path | 20:35 |
| yolanda | and in v2 , /usr/share/diskimage-builder/lib does not exist | 20:35 |
| yolanda | ianw, you mean, _LIB is only used for dib v1? | 20:35 |
| greghaynes | right, but why does the dib change require the instack change? | 20:36 |
| greghaynes | I think that means instack requires the dib change? | 20:36 |
| yolanda | greghaynes, no... if we test dib without that fix, it will fail on DIB CI | 20:36 |
| ianw | yolanda: no, we still use _LIB, but I don't think any elements depend on it? | 20:36 |
| ianw | actually, a quick grep shows that's not true | 20:37 |
| yolanda | i cannot find anywhere | 20:37 |
| yolanda | maybe i can drop it from the instack code completely? or only if using v2? | 20:37 |
| ianw | yolanda: no, i don't think you can. but i think https://review.openstack.org/#/c/401831/12/instack/main.py essentially looks correct? | 20:38 |
| yolanda | ok let's see how CI behaves now | 20:39 |
| greghaynes | I just dont understand why that var needs to be set for instack to work, wonedering what instack is doing there | 20:39 |
| greghaynes | the value is correct there, more just wondering why | 20:39 |
| yolanda | yep, i don't understand as well. I think my failure was just falling into exception, so that's why i added the workaround for paths in v2 | 20:40 |
| yolanda | but don't know the core reason for that _LIB var | 20:40 |
| ianw | greghaynes: it looks like a couple of elements use it to find files to source | 20:40 |
| greghaynes | oh wait, so is this running the elements but not using the dib runner? | 20:41 |
| ianw | greghaynes: yeah, that's always been the weird thing about instack | 20:41 |
| greghaynes | wow | 20:41 |
| greghaynes | whyyyy | 20:41 |
| greghaynes | yolanda: ok, so then yes youre going to need that export | 20:42 |
| ianw | ??? it's bitten me before too. elements are *mostly* separate, but there are some hook points like env vars like this | 20:42 |
| greghaynes | yolanda: but really, this isnt a public interface. the python module api is a public interface but the interface between dib and what context is passed to elements is not | 20:43 |
| greghaynes | we wont remove things but we totally can and do add things | 20:43 |
| greghaynes | so... this is just going to break randomly when we do | 20:43 |
| greghaynes | anyhow, I still dont get about the depends-on. If youre going to break without the dib fix, then you want to depend on the dib fix I would think? | 20:45 |
| yolanda | greghaynes, so it's a chain problem now.. | 20:51 |
| yolanda | dib will break without instack fix | 20:51 |
| yolanda | and instack will break without dib | 20:51 |
| greghaynes | wait, dib shouldnt break at all | 20:51 |
| yolanda | well, tripleo-ci for dib | 20:51 |
| greghaynes | that change works fine without any instack change - dib doenst depend on instack | 20:51 |
| yolanda | the tripleo-ci for dib won't pass | 20:51 |
| greghaynes | oh! | 20:51 |
| greghaynes | ah, ty | 20:52 |
| greghaynes | I forgot that we co-gate with instack | 20:52 |
| greghaynes | sorry about that :) | 20:52 |
| yolanda | but i also think that instack will fail without the other change, so it's a circular dep | 20:52 |
| greghaynes | yea, so IMO we should just remove the depends-on from dib, merge it anyways (tripleo is nonvoting) | 20:52 |
| greghaynes | then the circular dep is broken | 20:52 |
| yolanda | ah ok | 20:53 |
| yolanda | let me amend it then | 20:53 |
| greghaynes | ianw: curious what you think on the backwards compat element_deps patch | 20:53 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder: Preserve backwards compat for expand_dependencies https://review.openstack.org/407176 | 20:53 |
| yolanda | greghaynes, i'm also a bit worried about tripleo-ci being non-voting | 20:54 |
| yolanda | that led to ignoring the errors... | 20:54 |
| yolanda | sometimes they are just fake errors, but some other times were legit ones | 20:54 |
| greghaynes | right, its not actually possible to make it voting since its 3rd party CI style, but more so the issue is weve had long periods where it failed continuously where it sort of trained us to not treat it as voting | 20:55 |
| yolanda | yep, i agree on that | 20:57 |
| yolanda | anyway let's see if things are better now | 20:57 |
| yolanda | i need to step out for dinner, will check tomorrow morning | 20:57 |
| yolanda | greghaynes, also, what's progress on the partitioning? how do you see it? | 20:57 |
| greghaynes | It seemed like Andre updated the patches a bunch and started adding some tests, I think theres the perms issue we have to figure out still | 20:58 |
| greghaynes | I need to look at it again though, I havent since he updated over the weekedn | 20:58 |
| yolanda | i've been focused a lot on the ci fixes and lost track on progress about it. But i'm more confident now on the integration with tripleo itself | 21:00 |
| greghaynes | thats good to hear - now that theres some tests its a ton easier for me to review and fix up | 21:01 |
| greghaynes | so I think itll finish up more quickly, once I get some time to review it again | 21:02 |
| openstackgerrit | Michael Johnson proposed openstack/diskimage-builder: pip-and-virtualenv element needs python installed https://review.openstack.org/406420 | 21:39 |
| * hwoarang review time | 23:27 | |
| *** pmannidi has joined #openstack-dib | 23:46 | |
| *** dmarlin has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!