| *** chandan_kumar has joined #openstack-climate | 04:52 | |
| *** tomek_adamczewsk has quit IRC | 06:08 | |
| *** tomek_adamczewsk has joined #openstack-climate | 06:48 | |
| *** bauzas has joined #openstack-climate | 07:15 | |
| *** tomek_adamczewsk has quit IRC | 07:39 | |
| *** tomek_adamczewsk has joined #openstack-climate | 07:41 | |
| openstackgerrit | A change was merged to stackforge/python-climateclient: Removing dependencies from test-requirements https://review.openstack.org/86906 | 08:08 |
|---|---|---|
| bauzas | DinaBelova: hi | 08:44 |
| bauzas | DinaBelova: do you know if Nikolay is around today ? | 08:44 |
| bauzas | DinaBelova: https://review.openstack.org/#/c/78309/ | 08:44 |
| bauzas | DinaBelova: seems like he got trouble when testing | 08:44 |
| bauzas | DinaBelova: I want to make sure he properly tuned his devstack | 08:44 |
| bauzas | DinaBelova: oh, maybe I found why | 08:46 |
| bauzas | DinaBelova: probably because I haven't added a devstack conf flag | 08:46 |
| openstackgerrit | Sylvain Bauza proposed a change to stackforge/climate: Add extensions for API v2 controllers https://review.openstack.org/78309 | 09:00 |
| *** chandan_kumar has quit IRC | 10:55 | |
| DinaBelova | bauzas, he has several meetings today | 11:01 |
| DinaBelova | he's in the office | 11:01 |
| DinaBelova | I'll catch him | 11:01 |
| *** chandan_kumar has joined #openstack-climate | 11:09 | |
| *** chandan_kumar has quit IRC | 11:16 | |
| Nikolay_1t | bauzas: hi | 11:26 |
| Nikolay_1t | bauzas: i'm here | 11:26 |
| Nikolay_1t | bauzas: what problem have you found? | 11:26 |
| *** chandan_kumar has joined #openstack-climate | 11:30 | |
| *** cmart has joined #openstack-climate | 12:17 | |
| bauzas | hi Nikolay_1t | 13:04 |
| DinaBelova | :D | 13:14 |
| DinaBelova | he just went from the office :D | 13:14 |
| openstackgerrit | Swann Croiset proposed a change to stackforge/climate: Host creation: use auto-generated uuid as id https://review.openstack.org/87266 | 13:15 |
| scroiset | hi all | 13:16 |
| scroiset | ^^^ a simple patch for postgresql | 13:16 |
| cmart | scroiset, hi! it worked for me, but not for Jenkins.. :( | 13:30 |
| cmart | I think that is failing due to some tests :S | 13:32 |
| *** pafuent has joined #openstack-climate | 13:46 | |
| *** pafuent has quit IRC | 13:54 | |
| bauzas | cmart: we just discussed with scroiset about that | 14:04 |
| bauzas | cmart: a new change will happen | 14:04 |
| cmart | OK! | 14:06 |
| scroiset | humm .. cmart: see my comments on the LP bug .. | 14:08 |
| cmart | got it! I'll talk to Pablo then (I think that he had some progress with the 2nd option). Thanks! | 14:10 |
| *** pafuent has joined #openstack-climate | 14:21 | |
| scroiset | ok thanks | 14:25 |
| pafuent | Hi. DinaBelova, bauzas, scroiset, Could you take a look at https://review.openstack.org/#/c/84158/? | 14:28 |
| bauzas | pafuent: my only concern was about where to store the trust_id | 14:29 |
| bauzas | pafuent: I already reviewed your patch | 14:29 |
| bauzas | pafuent: you're storing it in ComputeHost table | 14:30 |
| bauzas | pafuent: IMHO, ComputeHostAllocation would be the best fit | 14:30 |
| bauzas | because that's a reservation info | 14:30 |
| pafuent | bauzas: Ok, I'll check it | 14:31 |
| bauzas | pafuent: will place a note | 14:31 |
| cmart | scroiset, are you there? | 14:35 |
| scroiset | yes | 14:46 |
| pafuent | Do we need to do something for this patch https://review.openstack.org/#/c/86777/? | 14:47 |
| pafuent | Fails due to a rebase. The updated lines were deleted in https://review.openstack.org/#/c/86906/ | 14:47 |
| pafuent | Should someone put a -2 that patch? | 14:47 |
| cmart | scroiset, I'm finishing the tempest stuff and I remember that you told me to add execution instructions on the README file.. | 14:51 |
| scroiset | for the patch https://review.openstack.org/#/c/86777/? .. seems we should abandonned it indeed | 14:52 |
| cmart | I ran my tests by adding a new "tox" target on the tempest tox.ini file.. and my idea was to add that instructions on the README.. Is that OK for you? | 14:53 |
| scroiset | tempest : yes I asekd to you :) | 14:53 |
| scroiset | ok for me ... so it will be smth like : tox tempest ? | 14:54 |
| scroiset | btw, there is an error in the README .. | 14:54 |
| cmart | yeah.. tox -eclimate-scenario.. | 14:54 |
| scroiset | not ./run_tests.sh | 14:54 |
| scroiset | but ./run_tempest.sh .. | 14:55 |
| cmart | or maybe I could create one that runs all the climate tests.. | 14:55 |
| cmart | including both cli and scenario tests | 14:55 |
| scroiset | all the tests sounds good | 14:55 |
| scroiset | for me | 14:55 |
| cmart | OK.. | 14:55 |
| cmart | will do that.. | 14:55 |
| cmart | thanks | 14:56 |
| cmart | one quick question.. what is the effect of the "slow" tag? | 15:00 |
| scroiset | for tox? | 15:00 |
| cmart | yeah | 15:01 |
| scroiset | humm .. I think all tests are run sequentially | 15:01 |
| scroiset | so it's slow | 15:01 |
| cmart | because I'm seeing that the heat guys out different targets for those kind of tests.. | 15:01 |
| scroiset | forget what I just said .. it's a testr related '--slowest' | 15:03 |
| cmart | and my idea was to create the new tox target with a regex based on the name of the tests files. For instance: bash tools/pretty_tox.sh '(?!.*\[.*\bslow\b.*\])(^tempest\.(api|scenario|thirdparty|cli)\.resource_reservation) {posargs} | 15:03 |
| cmart | I'm not sure if I should handle them separately.. | 15:03 |
| scroiset | .. depending how long are your scenarios ? | 15:06 |
| scroiset | heat have some tests for autoscaling veryyy long .. (tests are still skipped last time I took a look) | 15:06 |
| cmart | from three to five minutes.. it's only one test.. | 15:07 |
| scroiset | I would say .. keep it simple for the begin. | 15:07 |
| scroiset | one target should be fine | 15:08 |
| cmart | OK.. I'll remove the slow restriction from my target | 15:08 |
| pafuent | bauzas: Regarding the table of trust_id, seems that I can't put that column in other table. The trust_id is needed to delete the compute host, and seems that you could do that without any reservation. | 15:08 |
| pafuent | bauzas: I'm clear? | 15:08 |
| cmart | it will look like: bash tools/pretty_tox.sh '(^tempest\.(api|scenario|thirdparty|cli)\.resource_reservation) {posargs} | 15:08 |
| bauzas | pafuent: indeed, I looked at the code | 15:09 |
| bauzas | pafuent: you're adding trust_id in the dict | 15:09 |
| bauzas | pafuent: but imho, this shouldn't be stored in ComputeHost | 15:09 |
| bauzas | pafuent: trusts are useful for operations that are needed to be deferred | 15:11 |
| bauzas | pafuent: creating hosts is not one of these | 15:11 |
| *** chandan_kumar has quit IRC | 15:11 | |
| bauzas | pafuent: when thinking at this review, we shouldn't even store trusts | 15:11 |
| bauzas | for ComputeHosts I mean | 15:12 |
| bauzas | pafuent: and take use of the one in Leases for operating on PR reservations | 15:12 |
| pafuent | bauzas: So I'll need to modify the delete API in order to have a trust to delete the compute host | 15:12 |
| pafuent | bauzas: Are you OK with that? | 15:14 |
| bauzas | pafuent: well, we're discussing it with scroiset | 15:14 |
| bauzas | pafuent: there is a major concern there | 15:15 |
| bauzas | pafuent: we have 2 API endpoints | 15:15 |
| bauzas | 1/ for creating leases | 15:15 |
| bauzas | 2/ for creating hosts | 15:15 |
| bauzas | for creating hosts, it requires to be admin | 15:15 |
| bauzas | but this operation is immediate, no need to store a trust | 15:16 |
| pafuent | bauzas: I agree | 15:16 |
| bauzas | we can generate a trust and pass it to Nova | 15:16 |
| bauzas | for the deletion, we can also create a trust | 15:16 |
| bauzas | in terms of security, that's even better to generate another trust for deleting | 15:17 |
| bauzas | because the user could have been compromised | 15:17 |
| bauzas | ok with the hosts ? | 15:17 |
| bauzas | trusts should remain in memory for the duration of the request | 15:18 |
| pafuent | bauzas: Yes | 15:18 |
| bauzas | ok | 15:18 |
| bauzas | about leases | 15:18 |
| bauzas | that's another problem | 15:18 |
| bauzas | leases can be performed on a user basis | 15:18 |
| bauzas | not requiring to be admin | 15:18 |
| bauzas | but when manipulating Nova aggregates, it has to be done as admin | 15:19 |
| bauzas | and this action needs to be deferred => trust | 15:20 |
| bauzas | but if we store the user trust and pass it to Nova, it won't work because of not having enough rights for manipulating aggs | 15:21 |
| bauzas | you got it? | 15:21 |
| pafuent | bauzas: Yes | 15:22 |
| bauzas | so we need to store trust rights having admin rights | 15:22 |
| pafuent | bauzas: So we need some kind of validation that the user that uses PR plugin is an admin, right? | 15:23 |
| bauzas | no | 15:23 |
| bauzas | we need to find some way to create an admin trustee with a user trustor | 15:23 |
| bauzas | provided we are a service with admin rightsd | 15:23 |
| bauzas | http://docs.openstack.org/api/openstack-identity-service/3/content/trusts.html | 15:25 |
| pafuent | bauzas: To check if I got the idea... | 15:25 |
| pafuent | bauzas: Regular user => regular trust | 15:25 |
| bauzas | trustor is climate | 15:26 |
| bauzas | trustee is the user | 15:26 |
| pafuent | bauzas: Because we are a service with admin powers: regular trust => admin trust | 15:26 |
| bauzas | unless I'm wrong | 15:26 |
| bauzas | pafuent: but I'm not that good at trusts :) | 15:27 |
| bauzas | pafuent: maybe DinaBelova could help :) | 15:27 |
| * DinaBelova reading the logs | 15:28 | |
| bauzas | seems like I misunderstood trustor != trustee | 15:29 |
| bauzas | let me check code | 15:29 |
| pafuent | OK | 15:29 |
| bauzas | pafuent: indeed I'm wrong | 15:30 |
| pafuent | Trusts are immutable. If the trustee wishes to modify the attributes of the trust, they should create a new trust and delete the old trust. | 15:30 |
| bauzas | pafuent: https://github.com/stackforge/climate/blob/9552fce675eb9c48f86f2338b97ba79dba7951bb/climate/utils/trusts.py#L28 | 15:30 |
| DinaBelova | somehow I can | 15:31 |
| pafuent | DinaBelova: Hi | 15:31 |
| DinaBelova | can't get the point of the argument) | 15:31 |
| DinaBelova | pafuent, o/ | 15:31 |
| DinaBelova | I wanted smth like: admin user created host trusts Climate internal user to commit actions like moving host from freepool to the new user's pool | 15:33 |
| bauzas | DinaBelova: well, within a project, an user can be admin, while in another project, he is not | 15:34 |
| bauzas | DinaBelova: but that's a good option | 15:35 |
| DinaBelova | well, I'm just trying to get the possible solution | 15:36 |
| pafuent | bauzas: In the PR use case, Which kind of user create PR leases? | 15:36 |
| bauzas | pafuent: simple user | 15:36 |
| DinaBelova | I mean if this admin is global keystone admin - it'll work | 15:36 |
| bauzas | then OK | 15:38 |
| pafuent | bauzas: DinaBelova: Ok, so we need to store a trust from the user that creates the host, and use that trust when a regular uses create a lease, right? | 15:38 |
| bauzas | pafuent: only for manipulating aggs IMHO | 15:38 |
| bauzas | pafuent: nevermind | 15:38 |
| bauzas | pafuent: yes | 15:38 |
| pafuent | bauzas: So, we are back at the point of where to store that trust, right? | 15:39 |
| bauzas | pafuent: seems so | 15:39 |
| bauzas | pafuent: but at least we get a clear view :) | 15:39 |
| pafuent | bauzas: Yeah, that's rigth | 15:39 |
| pafuent | pafuent: :-D | 15:40 |
| pafuent | bauzas: The reservations and allocations are created with the lease, right? | 15:40 |
| bauzas | pafuent: yup | 15:40 |
| pafuent | bauzas: So we can store the trust there because don't exists at the moment of the host creation | 15:41 |
| bauzas | :) | 15:41 |
| pafuent | bauzas: Sorry, can't store | 15:41 |
| bauzas | pafuent: yup | 15:41 |
| bauzas | pafuent: let's store it in ComputeHosts | 15:42 |
| bauzas | pafuent: sounds like you were right | 15:42 |
| pafuent | bauzas: OK | 15:43 |
| DinaBelova | okay, cool | 15:44 |
| bauzas | pafuent: maybe a nicer logic would be to get the trust for the host, and use it for moving from agg to agg | 15:46 |
| bauzas | but if so, we need to iterate over the hosts | 15:47 |
| bauzas | if 2 hosts for one reservation | 15:47 |
| pafuent | bauzas: I'll check that | 15:48 |
| *** tomek_adamczewsk has quit IRC | 15:48 | |
| pafuent | bauzas: Sorry, but I need to go | 15:48 |
| bauzas | pafuent: ok me too | 15:51 |
| bauzas | pafuent: bye | 15:51 |
| *** bauzas has quit IRC | 15:53 | |
| *** chandan_kumar has joined #openstack-climate | 16:33 | |
| *** tomek_adamczewsk has joined #openstack-climate | 17:54 | |
| *** tomek_adamczewsk has quit IRC | 18:13 | |
| *** pafuent has left #openstack-climate | 18:18 | |
| *** pafuent has joined #openstack-climate | 18:22 | |
| *** tomek_adamczewsk has joined #openstack-climate | 18:32 | |
| openstackgerrit | Christian Martinez proposed a change to stackforge/climate: Add scenario testing support https://review.openstack.org/87352 | 18:40 |
| cmart | :S:S:S:S | 18:40 |
| *** chandan_kumar has quit IRC | 19:22 | |
| *** tomek_adamczewsk has quit IRC | 20:25 | |
| *** cmart has quit IRC | 20:39 | |
| openstackgerrit | Cristian A Sanchez proposed a change to stackforge/climate: Change references of tenant to project https://review.openstack.org/82192 | 20:47 |
| *** tomek_adamczewsk has joined #openstack-climate | 20:48 | |
| openstackgerrit | Cristian A Sanchez proposed a change to stackforge/climate: Change references of tenant to project https://review.openstack.org/82192 | 21:01 |
| *** pafuent has left #openstack-climate | 21:10 | |
| *** tomek_adamczewsk has quit IRC | 22:21 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!