Perl 6 - the future is here, just unevenly distributed

IRC log for #openstack-rally, 2015-02-20

| Channels | #openstack-rally index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:38 dmorita joined #openstack-rally
00:44 openstackgerrit Merged stackforge/rally: Supress warning in cli methods that have sensetive output  https://review.openstack.org/157536
01:08 himangi joined #openstack-rally
01:53 abhirc joined #openstack-rally
02:48 ilbot3 joined #openstack-rally
02:48 Topic for #openstack-rally is now ☁ Rally RoadMap: http://goo.gl/JZkmwY ☁ Rally IRC chat logs http://irclog.perlgeek.de/openstack-rally ☁ Key persons to ask:  boris-42, msdubov, rediskin, andreykurilin, amaretskiy  ☁ Documentation: https://rally.readthedocs.org/en/latest/ ☁ To publish changes to Rally:  https://rally.readthedocs.org/en/latest/contribute.html
03:09 openstackgerrit OpenStack Proposal Bot proposed stackforge/rally: Updated from global requirements  https://review.openstack.org/157652
04:30 nkhare joined #openstack-rally
05:32 rdas joined #openstack-rally
05:33 openstackgerrit Yusuke Ide proposed stackforge/rally: Catch a connection refused exception not to show traces  https://review.openstack.org/157659
05:52 yfried|afk joined #openstack-rally
06:03 psd joined #openstack-rally
06:31 neeti joined #openstack-rally
06:32 psd_ joined #openstack-rally
06:53 psd__ joined #openstack-rally
07:40 andreykurilin_ joined #openstack-rally
07:48 psd_ joined #openstack-rally
07:59 openstackgerrit nayna-patel proposed stackforge/rally: Adds Designate server-list scenario  https://review.openstack.org/157691
08:04 openstackgerrit Mikhail Dubov proposed stackforge/rally: [WIP] Unify RPS and constant runners and tune constant runner  https://review.openstack.org/155225
08:12 fhubik joined #openstack-rally
08:12 psd__ joined #openstack-rally
08:12 fhubik_afk joined #openstack-rally
08:15 exploreshaifali joined #openstack-rally
08:21 pboldin joined #openstack-rally
08:37 oanufriev joined #openstack-rally
08:48 openstackgerrit Yusuke Ide proposed stackforge/rally: Catch a connection refused exception not trace  https://review.openstack.org/157659
09:01 arxcruz joined #openstack-rally
09:52 psd_ joined #openstack-rally
09:57 msdubov_ joined #openstack-rally
10:12 aix joined #openstack-rally
10:17 psd__ joined #openstack-rally
10:25 e0ne joined #openstack-rally
10:32 himangi joined #openstack-rally
10:37 pboldin joined #openstack-rally
10:40 e0ne joined #openstack-rally
10:45 himangi joined #openstack-rally
11:00 openstackgerrit Merged stackforge/rally: Fix security group setup when users_per_tenant is > 1  https://review.openstack.org/157433
11:24 openstackgerrit Yusuke Ide proposed stackforge/rally: Catch a connection refused exception not to show traces  https://review.openstack.org/157659
11:31 openstackgerrit Roman Vasilets proposed stackforge/rally: Adding Multiple Request Scenario  https://review.openstack.org/117705
11:42 msdubov_ joined #openstack-rally
11:43 openstackgerrit Mikhail Dubov proposed stackforge/rally: [WIP] Unify RPS and constant runners and tune constant runner  https://review.openstack.org/155225
11:58 psd_ joined #openstack-rally
12:06 yfried|afk joined #openstack-rally
12:07 kairat_kushaev joined #openstack-rally
12:11 exploreshaifali joined #openstack-rally
13:06 fhubik joined #openstack-rally
13:06 pboros joined #openstack-rally
13:25 jaypipes joined #openstack-rally
13:25 openstackgerrit Merged stackforge/rally: Updated from global requirements  https://review.openstack.org/157652
13:42 dpaterson joined #openstack-rally
13:49 pboldin joined #openstack-rally
13:52 psd__ joined #openstack-rally
14:25 cdent joined #openstack-rally
14:29 openstackgerrit Mikhail Dubov proposed stackforge/rally: [WIP] Unify RPS and constant runners and tune constant runner  https://review.openstack.org/155225
14:31 openstackgerrit Mikhail Dubov proposed stackforge/rally: [WIP] Unify RPS and constant runners and tune constant runner  https://review.openstack.org/155225
14:35 cdent joined #openstack-rally
14:36 kairat_kushaev joined #openstack-rally
14:58 psd_ joined #openstack-rally
15:21 cdent_ joined #openstack-rally
15:28 abhirc joined #openstack-rally
15:29 openstackgerrit Marian Krcmarik proposed stackforge/rally: Add scenarios for live migration of an instance with volumes.  https://review.openstack.org/157844
15:57 psd__ joined #openstack-rally
16:02 openstackgerrit Marian Krcmarik proposed stackforge/rally: Add scenarios for live migration of an instance with volumes  https://review.openstack.org/157844
16:16 himangi joined #openstack-rally
16:31 zhiyan joined #openstack-rally
16:41 meteorfox joined #openstack-rally
16:59 psd_ joined #openstack-rally
17:04 ls joined #openstack-rally
17:05 himangi joined #openstack-rally
17:16 abhirc joined #openstack-rally
18:24 meteorfox boris-42: before I waste time, how reliable is the Docker installation for Rally?
18:25 boris-42 meteorfox: ?)
18:26 boris-42 meteorfox: it is tested and it works hm...
18:26 boris-42 meteorfox: as normal code..
18:26 boris-42 meteorfox: no difference=)
18:26 boris-42 meteorfox: here is tutorial https://registry.hub.docker.com/u/rallyforge/rally/
18:30 openstackgerrit Alexander Maretskiy proposed stackforge/rally: Add chunks support and chart `Parallel iterations' to report  https://review.openstack.org/146814
18:54 psd__ joined #openstack-rally
18:56 jjmb joined #openstack-rally
18:59 andreykurilin_ joined #openstack-rally
19:06 meteorfox boris-42: cool thanks
19:12 openstackgerrit Merged stackforge/rally: Updated from global requirements  https://review.openstack.org/157809
19:14 meteorfox boris-42: I'm interested to setup Rally as a gating job, in our private infrastructure, is there any documentation for that?
19:18 boris-42 meteorfox: so what you are looking for exactly?
19:19 meteorfox boris-42: something like upstream's gating job infrastructure for Jenkins
19:20 meteorfox boris-42: how is rally deployed on the jenkins slaves, etc
19:20 boris-42 meteorfox: hm so it depends up to you
19:21 boris-42 meteorfox: jenkins from OpenStack Infra (installs OpenStack with DevStack with Rally plugin)
19:21 boris-42 meteorfox: after that execute bash script that runs rally against it
19:21 boris-42 meteorfox: and publish bunch of files with all kinds of results from Rally
19:21 meteorfox boris-42: ah I see
19:22 boris-42 meteorfox: alternative way is to make a plugin for rally that deploy
19:22 boris-42 meteorfox: with your tool
19:22 boris-42 meteorfox: (devstack plugin is already in upstream)
19:22 boris-42 meteorfox: and then run rally
19:22 boris-42 meteorfox: benchmarks/validation
19:24 meteorfox boris-42: ok, so you use the SLA plugin, to determine if pass or fail?
19:24 boris-42 meteorfox: it's only related to "rally task"
19:28 meteorfox boris-42: ok, I see. thanks, I'll think some more about this, and look at the DevStack plugin and see what they do
19:30 boris-42 meteorfox: so the idea is to keep all results in single db
19:30 boris-42 =)
19:30 boris-42 meteorfox: all deployments and other stuff
19:30 psd_ joined #openstack-rally
19:33 psd joined #openstack-rally
19:34 exploreshaifali joined #openstack-rally
19:34 e0ne joined #openstack-rally
19:38 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano benchmarks  https://review.openstack.org/137661
19:38 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano base for murano benchmarks  https://review.openstack.org/137650
19:39 gugl2 left #openstack-rally
19:43 meteorfox boris-42: right, what's the default database? Sqlite?
19:43 boris-42 meteorfox: so it depends on rally.conf
19:43 boris-42 meteorfox: we support 3 DB backends (mysql, psql, sqlite)
19:43 boris-42 meteorfox: and we are doing functional testing for all of them
19:51 meteorfox boris-42: are they using sqlalchemy?
19:52 boris-42 meteorfox: who they?0
19:52 meteorfox =) that came out wrong
19:52 boris-42 meteorfox: yep rally is using sqlalchemy
19:52 boris-42 =)
19:52 meteorfox boris-42: are you using the core or the orm libraries?
19:53 boris-42 meteorfox: for now just orm
19:55 meteorfox boris-42: ok, what's being used in upstream for the backend?
19:59 psd boris-42, can you please elaborate upon your cleanup resources comment here: https://review.openstack.org/#/c/157458/1/rally/benchmark/scenarios/neutron/security_group.py
20:06 boris-42 meteorfox:  I don't get a question
20:06 boris-42 meteorfox: we are testing all backends
20:07 meteorfox boris-42: I was only curious what RDBMS is being used? No, technical reason
20:09 boris-42 meteorfox: maybe I don't get a question, you can use one of the next option: MySql, PSQL, SQLite
20:10 boris-42 meteorfox: depending on your goals
20:10 boris-42 meteorfox: if you are building CI it's better to chose MySql or PSQL
20:10 boris-42 meteorfox: all 3 backends are supported and tested
20:12 meteorfox boris-42: yeah, I know all of them are tests, I was just curious. :) will definitely go with either MySQL or PSQL, have you tried running with MariaDB?
20:12 meteorfox boris-42: it should work with the mysql bindings, i believe
20:13 boris-42 meteorfox: I am sure that it should work well with MariaDB
20:13 boris-42 meteorfox: but we don't have functional testing for it, so I can't guarantee 100% just 99.9% =)
20:14 meteorfox boris-42: that's fine
20:14 meteorfox :)
20:32 meteorfox boris-42: one more thing, can I configure Rally to use the internal API endpoints instead of the publicURLs?
20:34 boris-42 meteorfox: soo
20:34 boris-42 meteorfox: you can try
20:35 boris-42 meteorfox: https://github.com/stackforge/rally/blob/master/samples/deployments/existing.json put them here
20:36 meteorfox boris-42: do I just change public -> internal? Or do I add the URLs for every service?
20:36 psd_ joined #openstack-rally
20:37 boris-42 meteorfox: so actually seem like we will have issues https://github.com/stackforge/rally/blob/master/rally/deploy/engines/existing.py#L118-L119
20:37 boris-42 meteorfox: it's hardcoded here to use public api
20:37 boris-42 meteorfox: seems like code changes required
20:41 meteorfox boris-42: will that constant apply to all services or just keystone? Like when it grabs the URLs from the catalog, will it pick the internalURL of the other services too?
20:43 boris-42 meteorfox: here is the code how we are working with it
20:44 boris-42 meteorfox: https://github.com/stackforge/rally/blob/master/rally/osclients.py#L81-L319
20:45 meteorfox boris-42: and all of that is driven by these lines right? https://github.com/stackforge/rally/blob/master/rally/deploy/engines/existing.py#L129-L130
20:46 boris-42 meteorfox: in case of existing users
20:47 meteorfox ok
20:48 boris-42 meteorfox: acutally this line is more important https://github.com/stackforge/rally/blob/master/rally/deploy/engines/existing.py#L119
20:48 meteorfox boris-42: right, but isn't that that default?
20:49 boris-42 meteorfox: ?
20:49 meteorfox boris-42: isnt that a get() method from dict object?
20:49 boris-42 meteorfox: yep
20:49 boris-42 meteorfox: it is
20:49 boris-42 meteorfox: so actually maybe it works..
20:49 boris-42 meteorfox: could you try?)
20:50 meteorfox boris-42: sure
20:51 meteorfox boris-42: so in theory, setting the endpoint_type = 'internal' should use the internal endpoints?
20:52 boris-42 meteorfox: yep
20:52 boris-42 meteorfox: in theory it should work
20:52 meteorfox boris-42: I wonder if in the future with should allow to override services explicitly, to ignore the ones in the catalog, something like this https://github.com/paypal/aurora/blob/master/Config.json#L8-L17
20:52 meteorfox s/with/we
20:53 boris-42 meteorfox: I think it's valid case
20:53 boris-42 meteorfox: we can work on it
20:54 meteorfox boris-42: Once, I had to do something like that were I wanted to use a particular cell, instead of the global service. It'd be useful
20:55 meteorfox boris-42: I'm trying to get rally installed in this weird environment, but  I constantly get this error: http://rpc.cloudnull.io/python_packages/9.0.3/ uses an insecure transport scheme (http). Consider using https if rpc.cloudnull.io has it availabl
20:55 meteorfox boris-42: i'm using ./install_rally.sh -v
20:56 meteorfox boris-42: even when I tried pip install virtualenv I still get that same error
20:56 meteorfox it could be a pip issue
20:57 meteorfox boris-42: wait, I just realized that could be our internal pypi server
20:58 boris-42 meteorfox: hm it's strange
20:58 boris-42 meteorfox: intall-rally.sh installs rally from master
20:58 meteorfox boris-42: yeah, nevermind, it's an environmental issue on my side
21:07 meteorfox boris-42: our private pypi repository is outdated, and even after working around that issues, it can't still find other packages :( I'll see if I can just set it up to use the original pypi repo
21:14 boris-42 meteorfox:  we will have soon packages
21:14 boris-42 meteorfox: that may help
21:15 meteorfox boris-42: I'm just going to force pip to ignore our repo, and use the public one
21:20 rook joined #openstack-rally
21:32 meteorfox boris-42: finally got it working. I configured the endpoint_type to internal and now I'm running a task, rally show * and rally deployment check commands worked
21:33 meteorfox boris-42: nope, it didn't work got this error, '('Connection aborted.', error(110, 'Connection timed out'))' from the task
21:34 * meteorfox debugging
21:34 boris-42 meteorfox: what command?
21:34 meteorfox rally task
21:34 boris-42 meteorfox: start?
21:34 meteorfox yes
21:35 boris-42 meteorfox: run first command "rally task validate"
21:35 boris-42 meteorfox: with input task
21:35 meteorfox boris-42: ok, I'll try that
21:36 meteorfox boris-42: quick question, isn't 'rally task start' supposed to go to the background? It blocks, and I can't use the shell
21:38 meteorfox boris-42: I see the problem, it's using the publicURL, which I can't access from my env
21:38 meteorfox boris-42: and it times out
21:40 boris-42 meteorfox: it blocks shell
21:40 boris-42 meteorfox: btw could you vote  https://www.openstack.org/vote-vancouver/presentation/using-rally-for-openstack-certification-at-scale ?)
21:41 meteorfox boris-42: sure :)
21:41 boris-42 meteorfox: thanks
21:43 meteorfox boris-42: done, np
21:43 meteorfox boris-42: so, it seems endpoint_type didn't work, it's still trying to use the publicURL
21:47 meteorfox boris-42: the user context worked, it failed when it tried to validate the flavor, because it use the publicURL, instead of the internal
21:59 psd__ joined #openstack-rally
21:59 meteorfox boris-42: where do you create the Endpoint objects for new users?
21:59 boris-42 meteorfox: in user context
22:00 boris-42 meteorfox: https://github.com/stackforge/rally/blob/master/rally/benchmark/context/users.py#L185-L188
22:00 boris-42 meteorfox: yep and it is hardcoded
22:01 meteorfox boris-42: heh, ok that's where we need to propagate the endpoint_type
22:01 boris-42 meteorfox: so maybe by default we will take from admin endpoint
22:01 boris-42 meteorfox: and add new argument to context
22:01 boris-42 meteorfox: that allows to override that value ?
22:02 meteorfox boris-42: yeah, I think so, that could work.
22:02 boris-42 meteorfox:  so if you wish you can report a bug and publish fix
22:02 boris-42 =)
22:02 meteorfox boris-42: awesome, I will do
22:03 boris-42 meteorfox: it shouldn't be hard to do
22:06 meteorfox boris-42: yeah, it shouldn't be. I think in the long run we might want to set an endpoint_type for all new users, that could be overridden for each service
22:09 boris-42 meteorfox: it should be actually in deployment information
22:09 boris-42 meteorfox: so we will need to make ExistingCloud plugin on steriods
22:09 boris-42 )
22:09 meteorfox boris-42: yes, I agree.
22:14 andreykurilin_ joined #openstack-rally
22:15 meteorfox boris-42: I opened the bug here, https://bugs.launchpad.net/rally/+bug/1424082
22:16 meteorfox boris-42: I'll work on the patch
22:16 rook joined #openstack-rally
22:21 boris-42 meteorfox: I confirmed it
22:23 meteorfox boris-42: cool thanks. Here's what I'm thinking to do, am I missing something else? https://gist.github.com/meteorfox/291a2caa2a2b6d860195
22:23 boris-42 meteorfox: unit tests=)
22:24 meteorfox boris-42: endpoint is not a dict, that might not work. Yeah unit tests
22:25 boris-42 meteorfox: ?
22:26 boris-42 meteorfox: it will work actually because there is trick=)
22:26 meteorfox boris-42: I think this self.context["admin"]["endpoint"] returns an objects.Endpoint object, so I can't do ["endpoint_type"] on it
22:26 boris-42 meteorfox: ah no
22:26 boris-42 meteorfox: it won
22:26 boris-42 meteorfox: my fault
22:27 boris-42 meteorfox: use  endpoint_type = self.context["admin"]["endpoint"].endpoint_type
22:28 meteorfox boris-42: yeah, that's exactly what I did. I'm running the unit tests, I hope it covers that function
22:29 meteorfox boris-42: tests passed! This changes the behavior for the new users, do you know if other users were relying on this behavior?
22:30 boris-42 meteorfox: other users?
22:30 himangi joined #openstack-rally
22:30 meteorfox boris-42: I mean like other people
22:30 boris-42 meteorfox: you mean rally end users?)
22:30 meteorfox exactly
22:30 boris-42 meteorfox: ah I think they will be happy
22:31 boris-42 meteorfox: if there will be -1 bug
22:31 boris-42 =)
22:32 meteorfox boris-42: =), yeah, I was just thinking of some end-user that assumed that the internal endpoint_type, only applied for the existing users, and admin, but he/she was relying that new users used public endpoint_type. Probably not, that would be a strange scenario
22:33 boris-42 meteorfox: if you are worried about that you can add to users context one new argument
22:33 boris-42 meteorfox: fix jsonschema
22:33 boris-42 meteorfox: to accept "endpointType"
22:34 boris-42 meteorfox: so if user will need some specific behavior for new users then he will be able to do it
22:36 meteorfox boris-42: something like this https://gist.github.com/meteorfox/5e8131d5e56c7d67aa35
22:38 boris-42 meteorfox: yep it will be input
22:38 boris-42 meteorfox: but you should change user context validation schema
22:38 meteorfox boris-42: right, I know. =)
22:39 meteorfox boris-42: but it feels not right to define the endpoint_type, in context
22:39 meteorfox boris-42: it seems that should be part of the deployment configuration
22:40 boris-42 meteorfox: then don't add it
22:40 boris-42 meteorfox: =)
22:40 boris-42 meteorfox: I think as well it's better to avoid new arguments
22:40 boris-42 meteorfox: let's avoid argument
22:40 boris-42 meteorfox: if somebody ask about it we will return to this question
22:41 meteorfox boris-42: alright, sounds good.
22:51 openstackgerrit Carlos L. Torres proposed stackforge/rally: Sets new users' endpoint_type to default to the admin endpoint_type  https://review.openstack.org/157988
22:53 openstackgerrit Carlos L. Torres proposed stackforge/rally: Sets new users' endpoint_type to default to admin's  https://review.openstack.org/157988
22:53 meteorfox boris-42: sorry, I forgot that first line for commit has to be <65
22:55 boris-42 meteorfox: please add unit tests
22:55 boris-42 meteorfox: tests/unit/benchmark/context/test_users.py
22:55 boris-42 meteorfox: that will validate new behaviour
22:55 meteorfox boris-42: ok, right. I'll do that
22:55 meteorfox boris-42: I need to get to the habit
23:02 psd_ joined #openstack-rally
23:03 meteorfox boris-42: sorry I have almost no experience writing tests in OpenStack. How would I go about to specify, that deployment configuration uses the endpoint_type -> "internal"
23:04 meteorfox boris-42: I'm guessing using some kind of mock
23:10 boris-42 meteorfox so it's not different from other python projects
23:10 boris-42 meteorfox: just unittest + mocking
23:13 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano benchmarks  https://review.openstack.org/137661
23:13 openstackgerrit Roman Vasilets proposed stackforge/rally: Add murano base for murano benchmarks  https://review.openstack.org/137650
23:14 rook joined #openstack-rally

| Channels | #openstack-rally index | Today | | Search | Google Search | Plain-Text | summary