Monday, 2014-03-10

openstackgerritA change was merged to openstack/keystone: Don't need session.flush in context managed by session
openstackgerritA change was merged to openstack/keystone: Removes a redundant test
openstackgerritBrant Knudson proposed a change to openstack/keystone: Configurable temporary directory for tests
*** leseb has joined #openstack-keystone00:20
*** leseb has quit IRC00:24
openstackgerritBrant Knudson proposed a change to openstack/keystone: Cleanup test setup/config
jamielennoxdolphm: i really doubt your here, but regarding endpoint_type vs interface, yes the correct name is interface but i went with endpoint_type because it is used for this all throughout accessinfo and the service catalog00:53
jamielennoxas far as i can see the only reference to 'interface' is within the endpoint parsing of the service catalog itself00:54
jamielennoxis this something we are attempting to change people on/00:54
openstackgerritA change was merged to openstack/keystone: Fix webob.exc.HTTPForbidden parameter miss
*** zhiyan_ is now known as zhiyan01:24
*** stevemar has joined #openstack-keystone01:46
*** daneyon has quit IRC01:50
*** daneyon has joined #openstack-keystone01:50
*** daneyon has quit IRC01:55
*** daneyon has joined #openstack-keystone01:56
*** daneyon_ has joined #openstack-keystone01:57
*** daneyon has quit IRC01:57
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove unnecessary config_files implementation
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove unnecessary calls to self.config()
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove duplication when picking test config files
*** mberlin1 has joined #openstack-keystone02:08
*** daneyon_ has quit IRC02:08
*** mberlin has quit IRC02:10
*** stevemar has quit IRC02:18
openstackgerritA change was merged to openstack/python-keystoneclient: Provide more data to AuthMethod plugins
openstackgerritDavid Stanek proposed a change to openstack/keystone: Overrides filter builtin using six
*** david-lyle has joined #openstack-keystone03:10
*** bknudson has quit IRC03:14
openstackgerritA change was merged to openstack/python-keystoneclient: increase default revocation_cache_time
openstackgerritDavid Stanek proposed a change to openstack/keystone: Use assertIsNone when comparing against None
openstackgerritDavid Stanek proposed a change to openstack/keystone: Adds style checks to ease reviewer burden
openstackgerritDavid Stanek proposed a change to openstack/keystone: Add a space after the hash for block comments
openstackgerritDavid Stanek proposed a change to openstack/keystone: Removes the use of mutables as default args
*** stevemar has joined #openstack-keystone04:12
*** chandan_kumar has joined #openstack-keystone04:43
*** chandan_kumar has quit IRC04:52
*** saju_m has joined #openstack-keystone04:59
*** chandan_kumar has joined #openstack-keystone05:01
openstackgerritSteve Martinelli proposed a change to openstack/python-keystoneclient: Authenticate via oauth
openstackgerritSteve Martinelli proposed a change to openstack/python-keystoneclient: Add request/access token and consumer support for keystoneclient
*** stevemar has quit IRC05:21
*** gokrokve has quit IRC05:24
*** gokrokve_ has joined #openstack-keystone05:26
*** gokrokve_ has quit IRC05:31
*** gokrokve has joined #openstack-keystone05:38
*** saju_m has quit IRC05:54
openstackgerritJenkins proposed a change to openstack/keystone: Imported Translations from Transifex
*** saju_m has joined #openstack-keystone06:02
*** corsir has quit IRC06:08
*** jaosorior has joined #openstack-keystone06:36
*** gokrokve has quit IRC07:05
*** gokrokve has joined #openstack-keystone07:06
*** gokrokve has quit IRC07:09
*** gokrokve has joined #openstack-keystone07:39
*** gokrokve has quit IRC07:44
openstackgerritMarcos Fermín Lobo proposed a change to openstack/keystone: LDAP global roles and group roles assignments
*** saju_m has quit IRC08:22
*** saju_m has joined #openstack-keystone08:28
*** gokrokve has joined #openstack-keystone08:39
*** leseb has joined #openstack-keystone08:41
*** gokrokve has quit IRC08:44
*** saju_m has quit IRC08:50
*** saju_m has joined #openstack-keystone09:04
*** saju_m has quit IRC09:10
*** saju_m has joined #openstack-keystone09:11
openstackgerritmouad benchchaoui proposed a change to openstack/python-keystoneclient: Fix retry logic
*** henrynash has joined #openstack-keystone09:25
*** henrynash has quit IRC09:36
*** gokrokve has joined #openstack-keystone09:39
*** gokrokve has quit IRC09:44
openstackgerritMarek Denis proposed a change to openstack/keystone: Validate groups presence for federated authn
*** andreaf has joined #openstack-keystone09:56
*** henrynash has joined #openstack-keystone10:14
*** gokrokve has joined #openstack-keystone10:39
*** gokrokve has quit IRC10:44
*** lbragstad has joined #openstack-keystone10:49
*** lbragstad__ has quit IRC10:50
*** henrynash has quit IRC10:51
*** henrynash has joined #openstack-keystone11:00
*** saju_m has quit IRC11:05
*** andreaf has quit IRC11:21
*** gokrokve has joined #openstack-keystone11:39
*** gokrokve_ has joined #openstack-keystone11:41
*** henrynash has quit IRC11:44
*** gokrokve has quit IRC11:44
*** gokrokve_ has quit IRC11:45
*** leseb has quit IRC11:47
*** leseb has joined #openstack-keystone11:47
*** andreaf has joined #openstack-keystone11:48
*** lbragstad has quit IRC11:51
*** leseb has quit IRC11:51
openstackgerritMarek Denis proposed a change to openstack/keystone: Validate groups presence for federated authn
openstackgerritwanghong proposed a change to openstack/keystone: remove hardcoded SQL queries in tests
*** henrynash has joined #openstack-keystone12:05
*** henrynash has quit IRC12:09
*** bvandenh has joined #openstack-keystone12:13
openstackgerritwanghong proposed a change to openstack/keystone: remove the unused variable in test_sql_upgrade
*** henrynash has joined #openstack-keystone12:15
*** henrynash has quit IRC12:18
*** lbragstad has joined #openstack-keystone12:27
*** david-lyle has quit IRC12:28
*** gokrokve has joined #openstack-keystone12:39
*** leseb has joined #openstack-keystone12:39
*** gokrokve has quit IRC12:44
*** dims_ has quit IRC12:44
*** ayoung-ZZZZZZZzz has quit IRC12:45
*** henrynash has joined #openstack-keystone12:53
*** topol has joined #openstack-keystone12:54
*** dims_ has joined #openstack-keystone12:56
*** nkinder_ has quit IRC13:18
*** devlaps has joined #openstack-keystone13:36
*** stevemar has joined #openstack-keystone13:38
*** bknudson has joined #openstack-keystone13:38
*** gokrokve has joined #openstack-keystone13:39
openstackgerritSteve Martinelli proposed a change to openstack/python-keystoneclient: Add request/access token and consumer support for keystoneclient
*** gokrokve has quit IRC13:44
*** ChanServ sets mode: +o dolphm13:47
*** jagee has joined #openstack-keystone13:47
*** prad has joined #openstack-keystone13:53
*** prad has quit IRC13:53
*** browne has joined #openstack-keystone13:55
openstackgerritSteve Martinelli proposed a change to openstack/python-keystoneclient: Authenticate via oauth
dstanekanyone else having trouble running the full test suite? it seems that I'm running out of memory again for the last 238 tests13:57
dolphmdstanek: running with tox or testr?13:58
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove unnecessary calls to self.config()
dolphmdstanek: running it now...13:58
dstanekdolphm: how much ram do you have?13:59
dolphmdstanek: 8gb13:59
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove unnecessary config_files implementation
dstanekah, you won't have problems then - i'm in the process of creating a new bigger vm13:59
dstaneki was running a 2g performance server14:00
dolphmdstanek: ah, i'm just running on my laptop14:00
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove duplication when picking test config files
dstaneki think i can fix this with some changes to testtools - tests shouldn't get so big14:01
dstanekdolphm: alright, i just figured out my morning...14:01
bknudsonfix the memory leak14:02
bknudsonjust wait until we're running tests in parallel14:02
dstanekbknudson: that'll be fun - i'm going to plug the hole14:03
bknudsonI got a laptop with more cpus so the lack of parallel testing affects me now14:03
*** jagee has quit IRC14:04
dstanekbknudson: i forget where morganfainberg_Z left off14:05
bknudsonI'm sure there's lots to do.14:05
dolphmdstanek: i also wrote an ansible playbook for most of that raspberry pi project... deployed it temporarily to
jaosoriorI've been trying to run the full test suit in my laptop, but have been getting errors while trying to run keystone.tests.test_keystoneclient.KcMasterTestCase, is this normal?14:05
dstanekjaosorior: what are the errors? can you put the traceback in
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove duplication when picking test config files
openstackgerritBrant Knudson proposed a change to openstack/keystone: Configurable temporary directory for tests
jaosoriorthis is the error message when running ./ -x :
jaosoriorthough, almost every test in keystone.tests.test_keystoneclient.KcMasterTestCase fails for me :/14:08
bknudsonFor assistance, contact <a href="">IT Service Desk.</a>14:09
bknudsonthat's an odd error14:09
bknudsonit looks like it's not connecting to the local debug keystone server but some other system.14:10
jaosorioruhm, maybe it's my local environment14:10
*** nkinder_ has joined #openstack-keystone14:11
jaosorioris there any configuration file for the local debug keystone server?14:11
dstanekwhat is
jaosorioror somewhere were I can tweak this?14:11
jaosoriora proxy server14:11
dstanekit looks to me like BlueCoat-Security-Appliance is intercepting all traffic14:12
dstanekwhich is wierd because you are trying to get to localhost14:12
*** daneyon has joined #openstack-keystone14:12
jaosoriorwhich is odd :/14:13
dstanekyou may need to talk to your IT staff14:13
*** daneyon has quit IRC14:13
dolphmjaosorior: i think your laptop has been compromised by your employer :P14:13
*** daneyon has joined #openstack-keystone14:13
jaosoriorthanks for the help, it seems I'll have to deal with this somehow to get it working14:13
dstaneki've never seem security software that intercepts local traffic, but it doesn't mean it doesn't happen14:14
dstanekjaosorior: virtual box or a cloud vm is how i usually work14:14
dolphmjaosorior: you could also develop on a cloud server or in a local vm to avoid running into that14:14
dolphmdstanek: ++14:14
dstanekalthough you probably can't install VB14:14
dolphmdstanek: VB?14:14
dstanekvirtual box14:15
*** jagee has joined #openstack-keystone14:15
jaosoriorI can use VB, that's no issue, but I would prefer to be able to run this on my machine14:15
dolphmdstanek: you had me thinking visual basic14:15
bknudsonI thought VisualBasic14:15
dstanekdolphm: :-P14:15
jaosorioris there any file were I can configure the local test server?14:15
dstanekjaosorior: to do what?14:16
bknudsonI think the tests read the keystone/tests/test_overrides.conf14:16
dolphmjaosorior: for the most part, tests shouldn't require external configuration14:16
jaosoriorI'm aware of that, but I'm trying to come up with a temporal workaround14:17
dolphmjaosorior: it's literally starting a service on localhost with a high port number and testing against it14:17
dolphmjaosorior: there's no config there, as the ports are chosen randomly14:17
jaosorioroh, alright14:18
dstanekjaosorior: it would be interesting to see what happens if you run a web server on a high port and then access that through the browser14:19
jaosorioris there a parameter I can use for to exclude tests? in the meantime14:19
*** topol has quit IRC14:23
dstanekjaosorior: not that i know of, but you can just delete the file temporarily14:25
*** leseb has quit IRC14:25
dstanekjaosorior: the last time i needed to exclude tests i did it the complicated way and filtered the list of tests generated with subunit using grep - i would probably not do that again14:26
jaosoriorIt's passing now, I had to tweak some environment variable of mine14:26
*** leseb has joined #openstack-keystone14:26
jaosoriorbut if I recall there are some unit testing frameworks that handle these exclusions without the usage of grep14:27
jaosoriorhaven't gotten very acquainted to subunit yet though14:27
jaosoriorwould you consider it useful if I try to add similar functionality to the
dolphmjaosorior: what did you change?14:28
jaosoriorI added some exception to my proxy environment variables. It was just an error in my general setup14:29
*** leseb has quit IRC14:30
*** wchrisj has joined #openstack-keystone14:30
*** leseb has joined #openstack-keystone14:32
dolphmdstanek: test run went fine, but i totally forgot to watch it's memory usage :P14:32
dolphmi did run into this though in py26 and py27 (i've actually seen this before, but ignored it..)
*** david-lyle has joined #openstack-keystone14:33
dstanekdolphm: did the test for that fail?14:34
dolphmdstanek: yeah14:35
*** rwsu has joined #openstack-keystone14:36
dolphmi'm still on tox 1.6.1 (did 1.7 ever get fixed?)14:36
*** mberlin1 has quit IRC14:36
bknudsoni've never seen a failure in test_keepalive_and_keepidle_set14:37
*** gokrokve has joined #openstack-keystone14:39
dstanekdolphm: i'm on 1.6.1 and i haven't seen that error14:40
*** ayoung has joined #openstack-keystone14:41
*** gokrokve_ has joined #openstack-keystone14:41
*** gokrokve has quit IRC14:43
*** dolphm has quit IRC14:48
*** packet has joined #openstack-keystone14:49
*** tellesnobrega has quit IRC14:49
*** raildo has quit IRC14:49
*** dolphm_ has joined #openstack-keystone14:49
openstackgerritJenkins proposed a change to openstack/keystone: Updated from global requirements
bknudsonjenkins will never get that change merged if it keeps changing14:50
*** ayoung has quit IRC14:50
*** ayoung has joined #openstack-keystone14:50
*** dolphm_ is now known as dolphm14:50
*** ChanServ sets mode: +o dolphm14:50
*** gokrokve_ has quit IRC14:51
openstackgerritJenkins proposed a change to openstack/python-keystoneclient: Updated from global requirements
*** devlaps has quit IRC14:54
dolphmdstanek: the total duration for running tox -e py27 has basically doubled for me in the past week :-/14:55
dolphmrunning again...14:56
dolphm(last run: )14:56
bknudsonI think new migrations wind up being O(n*n)14:57
bknudsonif so, squashing migrations would speed it up again14:58
*** thedodd has joined #openstack-keystone14:58
dstanekbknudson: is that a work in progress or a wishlist item?15:03
bknudsondstanek: morganfainberg_Z has a change lined up for J15:04
*** leseb has quit IRC15:05
*** leseb has joined #openstack-keystone15:05
*** bvandenh has quit IRC15:06
*** gokrokve has joined #openstack-keystone15:08
*** leseb has quit IRC15:09
*** leseb has joined #openstack-keystone15:11
*** jraim has quit IRC15:15
*** jraim has joined #openstack-keystone15:17
*** stevemar has quit IRC15:21
openstackgerritMarek Denis proposed a change to openstack/keystone: Validate groups presence for federated authn
*** stevemar has joined #openstack-keystone15:25
mhuHi, is there a documented way to update the sql backend for a contrib ? Namely, run the scripts there:
*** gokrokve has quit IRC15:40
mhuI am using a devstack and the identity_providers table is not created when installing keystone15:43
lbragstadmhu: db_sync?15:50
mhulbragstad, doesn't seem to work15:51
mhuI did 'keystone-manage db_sync', it exited silently, and the table hasn't been created15:52
dstanekmhu: did you enable the extension?15:54
dstanekmhu: ^15:55
mhudstanek, thanks a lot !!15:55
mhuI bet it's the part with db_sync15:55
mhudstanek, lbragstad : yep, that did it alright - thanks for your help15:56
*** devlaps has joined #openstack-keystone16:16
*** gyee has joined #openstack-keystone16:19
*** mberlin has joined #openstack-keystone16:21
*** raildo has joined #openstack-keystone16:26
*** gokrokve has joined #openstack-keystone16:26
dstanekmhu: np16:27
*** harlowja has joined #openstack-keystone16:49
*** browne has quit IRC16:53
*** marcoemorais has joined #openstack-keystone16:55
*** browne has joined #openstack-keystone17:00
*** amcrn has quit IRC17:04
ayoungnkinder_, can you bring back to life.17:06
*** leseb has quit IRC17:06
ayoungmorganfainberg_Z, let me know when you are up, and we can talk about the revoke/clear patch17:08
*** leseb has joined #openstack-keystone17:11
*** morganfainberg_Z is now known as morganfainberg17:11
*** richm has joined #openstack-keystone17:12
morganfainbergbknudson, huh what do i have lined up for J?17:12
morganfainbergayoung, here now17:13
ayoungmorganfainberg, so, do you need help in coming up with the test ?17:14
openstackgerritayoung proposed a change to openstack/python-keystoneclient: Split sample PKI token generation
ayoungmorganfainberg, I'mreferring to
ayoungmorganfainberg, the test would be slightly artificial17:15
morganfainbergi saw your comment17:15
ayoungthe isdea was that in the real world, we would use the token expiry as the point after we could drop a revocation event17:16
ayoungbut if a token was set with an artificially long expiry17:16
ayoungwe could drop the event, and then that token would start passing17:16
ayoungin geenral, revocation events don't revoke into the future17:16
ayoung,but we could do that, too17:16
morganfainbergthat can't be done short of mucking with the data outside of the control of keystone17:16
morganfainbergthe artificial expiry17:16
morganfainbergor custom provider17:16
ayoungtest_revoke should be able to do that17:17
ayoungyou can call cms to sign a token yourself if you really want17:17
morganfainbergi think that fall outside of the scope of the test i wrote17:17
morganfainbergthat sounds like a completely different test case17:17
ayounginfact, since token_values is all that is handed off, you can just hand jam them17:18
ayoungmorganfainberg, ok, let me give it a stab17:18
ayoungwe can add out +1s together to get to a singele +2 if desired17:18
morganfainbergayoung, sure.17:19
morganfainbergin either case i want to get that review in today17:19
morganfainberghowever we need to do that17:19
morganfainbergs/through/in gate17:19
*** jaosorior has quit IRC17:20
*** lbragstad has left #openstack-keystone17:23
*** lbragstad has joined #openstack-keystone17:24
*** amcrn has joined #openstack-keystone17:27
ayoungmorganfainberg, first off, you were passing the token iteslef into the event17:28
ayounginto the tree17:28
ayoungit needs to be the flattened version from17:29
ayoungIt can be a simple dictionary like17:29
morganfainbergayoung, i am using the v2 token values method17:29
morganfainbergayoung, look at line 15017:30
ayoungAh...I see it17:30
ayoungmissed that17:30
morganfainbergayoung, yep17:30
morganfainbergayoung, no worries17:30
ayoungmorganfainberg, no need, I think17:30
ayounglets just test the dictionary here17:30
morganfainbergi was aiming for the "expected" value based on something that a token kind of looks like17:31
morganfainbergbut if you'd rather it be a hand-crafted flattend value set, sure17:31
morganfainbergthough, i think that is how we got into the mess where our sample tokens don't look like "real" tokens17:32
*** andreaf has quit IRC17:34
ayoungmorganfainberg, wel...I don't disagree....the code was originally split out or other reasons.17:34
ayoungI like the idea of beating on the v2 and v3 translation functions17:34
lbragstadmorganfainberg: quick question on the region migration review whenever you've got a  minute17:35
morganfainberglbragstad, sure17:35
lbragstadso, I took your and jay's suggestions and implemented them17:35
morganfainbergwow i forgot i even looked at that review :P17:35
lbragstadI sync'd down to 36, since that was the migration before jay's region table addition17:36
ayoungmorganfainberg, blank_token_data is in the post form, though17:36
ayoungnot pre17:36
morganfainbergayoung, ah, that isn't clear17:36
ayoungmorganfainberg, I don't think I wrote that17:37
morganfainbergalso... why is _blank_token_data in it's only used for testing17:37
lbragstadfor some reason, it doesn't look like it's picking up the right constraint,17:37
morganfainberglbragstad, i see that looking at that now.17:37
ayoungUm...guess I did...17:37
ayoungdoesn't really seem to belong in the model...looks like test code17:38
ayoungonly your code uses it17:38
*** jaypipes has joined #openstack-keystone17:39
morganfainberg_sample_blank_token uses it, and that is used in a few places17:39
* lbragstad waves at jaypipes17:39
morganfainbergjaypipes, such comment17:39
ayoungI think that it should be used to default the token_data17:40
ayoungbut it probably should be _17:40
morganfainbergoh ick. lbragstad17:40
lbragstadlol the sql error?17:40
morganfainbergthe "unique" keyword might be the issue17:41
morganfainbergvs. how we "mostly" just apply the constraint explicitly as a constraint afterwards17:41
morganfainbergi'm guessing "unique" does something very dumb17:41
lbragstad.. kinda what I was thinking...17:41
lbragstadso that would leave us with having to create a temp table, move the data from region table over, drop the old region table, and recreate without the 'unique=True' part... right?17:42
morganfainbergi'm looking at the SQLA code now17:42
jaypipeslbragstad: if you do a print of region_table.indexes right before the uc.drop() call, what does it print?17:42
lbragstadjaypipes: checking17:42
jaypipesmorganfainberg: I'm thinking the same as you.. might be the default "name" of the index...17:43
morganfainbergunique constraints with unique kw looks like a non-named constraint17:43
morganfainbergwhich is bad17:43
jaypipesmorganfainberg: right. oslo.db has all that fixed up, though.17:43
lbragstadset([Index(u'description', Column(u'description', VARCHAR(length=255), table=<region>, nullable=False), unique=True)])17:43
jaypipesmorganfainberg: I don't like the UC naming convention in oslo.db, but at least it is consistent.17:44
morganfainberglbragstad, can you do a 'show create table' ?17:44
jaypipeslbragstad: OK, one sec... checking to see if UniqueConstraint() has a "name" kwarg17:44
morganfainbergjaypipes, fair, i like consistent naming.17:44
morganfainbergjaypipes, it does.17:44
morganfainbergjaypipes. look at line 20417:45
morganfainbergfor example17:45
morganfainbergit's a sql-a migrate ism it looks like17:45
jaypipeslbragstad: ok, change line 16 in that paste to desc_uc = constraint.UniqueConstraint('description', table=region_table, name="description")17:45
morganfainbergjaypipes, sec17:46
morganfainbergok, so i need to see what the key ends up being17:47
jaypipesmorganfainberg: "description" see lbragstad paste above.17:47
morganfainbergnot sure if thats what SQLA is doing17:47
* lbragstad fist pump! 17:47
lbragstadthat works17:47
* lbragstad pasting17:47
morganfainbergoh ok17:47
morganfainbergi am concerned that it wont "work" in all SQL backends though17:48
morganfainbergsince it seems like SQLA lets the backend do some magic name assignment sometimes17:48
morganfainbergbut i haven't tracked that down 100% to be sure17:48
jaypipesmorganfainberg: fair point. lbragstad, instead of hard-coding "description" as the name, probably worth taking the index name from the return of region_table.indexes[0]17:48
morganfainbergthat sounds more correct17:48
morganfainbergjaypipes, ++17:48
*** gyee has quit IRC17:49
lbragstadgood diea17:49
jaypipeslbragstad: might want to double check the structure of the Index class with a print dir(region_table.indexes[0]) to see if there is a name attribute.17:49
jaypipeslbragstad: did you want me to paste something? ;)17:49
morganfainbergjaypipes, if there isn't a consistent name, we might want a temp/pivot and recreate.17:50
jaypipesmorganfainberg: ?17:50
morganfainbergjaypipes, it sucks to need to do, but it ensures we have consistency17:50
lbragstadjaypipes: hah nope... my ^ - C are a little fast today17:50
morganfainbergcreate new table, migrate data, rename table17:50
jaypipesmorganfainberg: not sure that is needed, as the __table__.indexes collection should give us the introspection we need (all we need is the index/constraint name in this case, no?)17:51
morganfainbergif the tables.indexes[0] doesn't have a name17:51
lbragstadmorganfainberg: jaypipes yeah, we have done that before, migration 16 iirc17:51
morganfainbergjaypipes, well, the issue is non-named constaints can't be migrated17:51
morganfainbergjaypipes, sql-a has this issue17:51
morganfainbergjaypipes at least last i heard17:51
morganfainbergsql-a-migrate that is17:51
jaypipesmorganfainberg: ah, yes, that is indeed true... poop.17:51
morganfainbergjaypipes, so if we dno't have a name, lets just migrate a new "fixed" table (sucky but solves the issue cleanly)17:52
lbragstadso we will have to create tmp, migrate data, drop old, and recreate regions without unique=True17:52
jaypipesyes, I suppose so. :(17:52
morganfainberglbragstad, i _think_ you can do create tmp, migrate data to tmp (with correct structure), drop old, rename tmp17:52
jaypipeslbragstad: but don't we need to do that anyway with the change to nullable=True on description?17:52
morganfainbergone less table create17:52
lbragstadmorganfainberg: ok, sounds like a plan17:53
morganfainbergjaypipes, nullalble should be a alterable attribute17:53
jaypipeslbragstad: thx for your patience, man!17:53
morganfainbergjaypipes, since it's not held as an index/key17:53
lbragstadjaypipes: I think dolphm wanted to keep that nullable=False17:53
morganfainbergjaypipes, indexes and keys are not held in the column on all engines17:53
lbragstadand we can carry a fix in the manager to set it to '' if it isn't provided17:53
jaypipesmorganfainberg: hmm, yes, that is true for the upgrade. downgrade needs to be guarded, though.17:53
morganfainbergjaypipes, sure, downgrade would need de-conflicting17:54
jaypipesmorganfainberg needs deconflicting :P17:54
morganfainbergjaypipes, though tbh, i'd probably just make the downgrade a no-op and fix the migrate up to include it.17:54
morganfainbergif is nullable, do nothing, if isn't drop nullable. fix previous migrate to not include nullable, and downgrade is "do nothing"17:54
*** browne has quit IRC17:55
*** browne has joined #openstack-keystone17:55
jaypipesfair enough.17:55
jaypipesno harm no foul on that one.17:55
morganfainbergi actually like that methodology for non-shipped code.17:56
morganfainberglets people track master and lets us "fix" issues without worrying about conflict resolution etc.17:56
lbragstadI'll work on another patch and hope to get something pushed up today17:57
morganfainberglbragstad, sounds good17:57
lbragstadmorganfainberg: jaypipes thanks again for the help, it's really appreciated17:57
morganfainberglbragstad, thanks for working on this, it is a bit painful so patience is also appreciated.17:58
jaypipeslbragstad: any time!17:58
jaypipestrust me, I know how painful the migration crap is, so again, ty for helping with it.17:58
morganfainbergjaypipes, i have some work to do for that to land in J, but yes, migrations can be painful17:59
jaypipesheh, yeah.18:00
*** marcoemorais has quit IRC18:04
*** marcoemorais has joined #openstack-keystone18:05
*** marcoemorais has quit IRC18:05
*** marcoemorais has joined #openstack-keystone18:05
openstackgerritayoung proposed a change to openstack/keystone: Call an existing method in sync cache for revoke events
*** zhiyan is now known as zhiyan_18:08
morganfainbergayoung, that looks reasonable, running a test locally before tossing a +1 on it18:09
ayoungmorganfainberg, thanks18:10
morganfainbergayoung, works as expected! nice :)18:11
morganfainbergalso i'm loving mock.patch.object18:11
ayoungmorganfainberg, took me a second to debug, cuz I had token[user_id] = user_id18:11
ayoungyeah, kinda panicked there for a second18:12
morganfainbergi've _never_ done that before /s18:12
morganfainberg"Why isn't this dict working?! i set the key!!!"18:12
morganfainbergok +1 added specifically calling out the expanded test.18:13
morganfainbergminor nit should be fixed.18:14
morganfainbergline 169, remove it instead of commenting it out18:14
morganfainbergayoung, ^18:14
*** zhiyan_ is now known as zhiyan18:19
ayoungrunning unit tests in tox provides a real lack of feedback18:21
morganfainbergayoung, testr18:22
morganfainbergayoung, it's not tox's fault18:22
morganfainbergit's the runner being annoying18:22
ayoungmorganfainberg, provides feedback18:22
*** leseb has quit IRC18:22
ayoungand that uses testr, too18:22
*** leseb has joined #openstack-keystone18:22
morganfainbergayoung, hm.18:23
morganfainbergwe could probably do the same thing in the standard tox runs18:23
ayoungmorganfainberg, Or provide a switch...I suspect that the CI doesn't need the log spew18:24
morganfainbergayoung, looking to see if we can do something posargs specific18:25
ayoungmorganfainberg, it turns out that I had rebooted my machine and not run testmount, so it was not doing the ramdisk trick18:25
ayoungmorganfainberg, I usually realize that when I see how slow each test is running18:26
*** leseb has quit IRC18:27
ayoungmorganfainberg, so, I just ran the master version of python-keystoneclient, but with the popen args changes to : universal_newlines=False  and it runs fine in 2718:29
ayounggoing to try with -epy3318:29 fails like immediately18:29
ayoungdb type could not be determined18:30
*** zhiyan is now known as zhiyan_18:34
ayoungmorganfainberg, OK...I learned something18:34
*** gyee has joined #openstack-keystone18:37
openstackgerritayoung proposed a change to openstack/keystone: Call an existing method in sync cache for revoke events
morganfainbergayoung, LGTM18:42
*** leseb has joined #openstack-keystone18:45
*** leseb has quit IRC18:45
*** leseb has joined #openstack-keystone18:45
dstaneki wonder is there is an option for testr to not store the test history18:47
*** leseb has quit IRC18:50
morganfainbergdstanek, does not look like it18:52
*** daneyon has quit IRC18:54
morganfainbergayoung, tox -epy27 -- --subunit | subunit-2to1 | tools/colorizer.py18:54
morganfainbergthat should, afaict do what run_tests does18:55
morganfainbergchecking now18:55
ayoungmorganfainberg, we can submit a change to to do that18:55
dstanekmorganfainberg: really? i wouldn't expect the pipes to work there18:55
morganfainbergayoung, yeah if it works18:55
dstanekayoung: that is what run_tests is doing right?18:55
ayoungI would be willing to bet that using to run tox would be the right long term direction18:55
ayoungit means we don't duplicate how we test, we just provide a shortcut and "teach people tox" approach18:56
morganfainbergdstanek, might need a little extra guarding for bash-isms18:56
morganfainbergdstanek, sure.18:57
dstanektestr make all this stuff so overly complicated18:57
ayoungdstanek, um...if you recommend switching out our testing infrastructure again, I will beat you with morganfainberg 's +2 pin in a wool sock18:58
dstanekayoung: i didn't recommend it in the first place :-)  that was the official open stack direction :-(18:58
ayoungdstanek, I meant that should just do  "tox -epy27"18:59
dstanekit's part of the reinvent everything game plan18:59
ayoungand -pep8  do to -epep818:59
ayoungand so forth.  I'm collecting notes for a cheat sheet18:59
dstanekayoung: yeah, i think that would be a good plan18:59
ayounglike,  whenc toc -epy33 fails remove the .test--repost times.db file18:59
ayoungalthough I suspect we should probably just recommend wiping that whole subdir19:00
dstanekayoung: although for that to work we'd have to figure out the pipelining19:01
ayoungwhat about the pipelining?19:02
morganfainbergayoung, dstanek tox -epy27 -- --subunit | subunit-2to1 | tools/ works just fine19:02
morganfainbergjust need to make sure python-subunit and six are installed on the system19:02
dstanekmorganfainberg: oh, nice19:02
ayoung from testtools import StreamToExtendedDecorator19:03
ayoungImportError: cannot import name StreamToExtendedDecorator19:03
dstaneki wonder if you can use tox to run those19:03
ayoungmust be able to19:03
morganfainbergi can absolutely get a change for run_tests to do that, the only question is making run tests use the "right" env for python version19:03
ayoungmorganfainberg, for Keystone can default to 27 until we can get a clean 33 run19:03
ayoungonce we do, lets switch it to 3319:03
morganfainbergor we could just add to tox.ini -e "19:03
ayoungand make a gentlemen's pact to code primarily in 33, with 27 being the afterthought19:04
morganfainberga subunit-2to1 env19:04
dstanekmorganfainberg: the way i have thought about it in the past was to create a bash helper script and have to run that19:04
morganfainbergso it would use whatever the "base" version is19:04
ayoungoh, and run pep before the unit tests19:04
dstanekbut i never did it because the script would call tox and tox would run yet another shell script - it just made me too sad19:05
morganfainbergwe could just make run_tests "smart"19:05
morganfainbergit already is the helper script19:05
morganfainbergand tox already has env's for everything19:05
morganfainbergso, two options, new tox env "subunit_2to1"19:06
morganfainbergwhich just uses the base/default python version19:06
dstanekmorganfainberg: pipelining doesn't work directly in tox does it?19:06
morganfainbergeg. what pep8 does19:06
morganfainbergsure it does.19:06
morganfainbergit just executes whatever commands19:06
dstanekwhy not put it there then and call it a day?19:06
dstaneki swear i tried it and ran into issues19:07
* morganfainberg checks19:07
dstanekso instead of the 'python testr' command you would copy the pipeline from run_tests.sh19:07
morganfainbergok let me see what i can convince tox to do.19:08
ayoungdstanek, how is your python 33 kungfu?  I am trying to make a swtich in the popen call to deal with binary data and, well it is crapping out19:08
morganfainbergmight need to explicitly invoke bash19:08
ayoungto use atechnical term19:08
ayoungTypeError: 'str' does not support the buffer interface19:08
dstanekmorganfainberg: like a "bash -c 'blah'""?19:09
morganfainbergdstanek, ah you're right, we're in pure python19:09
morganfainbergdstanek, can't use bash-isms19:09
dstanekayoung: decent19:09
morganfainbergdstanek, lets not do that19:09
ayoungdstanek, when calling popen, I was passing universal_newlines=True19:09
morganfainbergbut it works19:10
ayoungbut, I suspect it is using that to detect "is this a text going both directions"19:10
dstanekmorganfainberg: when i thought about it i didn't like any solution i came up with so i just stopped thinking about it :-)19:10
morganfainbergwell.. it looks like it works19:10
ayoungIf I set universal_newlines = False, it seems like code that is expecting binary is getting a string and barfing on it19:10
dstanekayoung: which review?19:10
dstaneki've been using nosetests because everytime i have an import error testr won't give me a traceback19:11
ayoungdstanek,  has it ,but I think I am going to post a simpler one for just this change first19:11
ayoungI'll post a draft19:11
morganfainbergdstanek, eyah /bin/bash -c "<stuff>" does exactly that19:13
morganfainbergnot sure i want to spew all the timing stuff into the CI logs though19:13
morganfainbergayoung, dstanek, i'm more inclined to make a new tox target.19:13
ayoungdstanek, you can see that just encoding the outgoing text does not seem to be enough19:14
ayoungin fact, it seems to be irrelevant19:14
morganfainbergdstanek, ayoung, so new tox target run_tests uses? or ... just spew the content into the CI logs?19:15
morganfainbergnew target can just use the base python version19:16
morganfainbergso whatever you're using locally would be the target.19:16
ayoungmorganfainberg, if we add a new target, it will not be run yet by CI, right?  It explicitly runs pep8 and py27 in different targets anyway19:16
morganfainbergbut do we want that info in the CI logs.19:17
morganfainbergactually... i think this is an infra question19:17 could do tox -epep8  and then tox -epy2719:17
morganfainbergif they have a strong feeling one side or the other.19:17
morganfainbergayoung, yeah i can have it do it just like that19:17
ayoungwith all of the coloring19:17
morganfainbergw/ the extra args to make it colorful19:17
morganfainberglike i said, provided you have subunit and six installed19:17
morganfainbergif we make it a tox target, six and subunit are installed in the venvs19:18
morganfainbergrun_tests would just use the new tox target vs. py2719:18
ayoungmorganfainberg, sounds right to me19:19
ayoungmorganfainberg, 2 reviews19:19
ayoungone for the new tox target, a second to use it in run_tests.sh19:19
morganfainbergeh, nah, 1 review19:19
ayoungthe second that is going to require some finagling19:19
ayoungbut the first review is  valuable all by itself19:20
dstanekwhat's the benefit using having a new venv?19:20
ayoungnah, use the same env as the py27 call19:20
ayoungjust different args19:20
morganfainbergayoung, subunit_27 and subunit_3319:20
*** raildo has quit IRC19:21
ayoungmorganfainberg, so...could we do the same kind of thing to make it easy to run  postgres, mysql, and LDAP live tests?19:21
dstanekmorganfainberg: and then have it reuse the existing py27 and py33 envs right?19:21
morganfainbergdstanek, ++19:21
morganfainbergayoung, uhm... that would be harder since that is based on config files19:21
ayoungyeah, I know....but we need something like that.  How does the autodiscovery work?19:22
ayoungCould we somehow make those tests triggered by putting them in a different directory or require some env var?19:22
dstanekayoung: my guess is that if you have in py33 you have to convert the value of text to bytes19:22
ayoungdstanek, wouldn't that be "encode"19:23
*** leseb has joined #openstack-keystone19:23
dstanekalthought i would have expected encode to work19:26
ayoungdstanek, so strutils.to_bytes  ?19:26
morganfainbergso we can't make tox behave automatically19:26
morganfainbergbecause tox uses py27, py26, and py33 as magic venvs19:27
morganfainbergi'm going to make run_tests smarter19:27
dstanekayoung: i don't see you doing the encode on the text19:28
ayoung  line 10819:28
ayoungand 22219:28
dstanekmorganfainberg: you can't have tox use one of those as the venv for a new target?19:28
morganfainbergdstanek, you can, but if you run subunit outside of py33, it'll install the "base" version19:29
morganfainbergso you might be calling it subunit_py33 and end up with a py27 install19:29
morganfainberg"pyXX" is a magic venv19:29
morganfainbergi'd rather keep that19:29
morganfainbergand just tell people "use run_tests, no really"19:29
morganfainbergand i'll add an option to run py33 or py26 to run_tests19:30
dstanekayoung: i don't think strutils is actually encoding the data and giving you bytes19:30
morganfainbergbut it'll do the same thing as tox -epy27 but just with subunit output19:30
ayoungdstanek, um... I think you are right19:31
dstanekayoung: it seems that function is just stoopid!19:31
ayoungdstanek, but I tried a plain six call with similar results19:31
openstackgerritA change was merged to openstack/python-keystoneclient: Updated from global requirements
openstackgerritMorgan Fainberg proposed a change to openstack/keystone: Add slowest output to tox runs (testr)
*** raildo has joined #openstack-keystone19:33
dstanekayoung: what happens when you just do a plain old str.encode('utf-8')19:35
*** daneyon has joined #openstack-keystone19:37
*** gokrokve_ has joined #openstack-keystone19:40
ayoungdstanek, just tried that...I'll paste19:41
ayoung  File "./keystoneclient/common/", line 90, in _process_communicate_handle_oserror19:42
ayoung    err +=
ayoungTypeError: Can't convert 'bytes' object to str implicitly19:42
ayoungthat one looks dealable19:42
ayoungstill instances of TypeError: Type str doesn't support the buffer API19:42
raildodolphm: pig19:44
*** gokrokve has quit IRC19:44
*** daneyon has left #openstack-keystone19:48
ayounggyee,  you are going to want that for your X509  work.19:52
bknudsonI'm getting all sorts of warnings from doc build with the requirements change.19:53
*** serapontiy has joined #openstack-keystone19:54
raildoayoung: I was interested in contributing to the federation and would like to know how I can do this, is there any functionality that can help, or test, or extend to the client19:56
raildoCould you help me or you know who should I contact?19:57
openstackgerritRichard Megginson proposed a change to openstack/keystone: better handling for empty/None ldap values
morganfainbergayoung, so, --fail-fast is the onlything i'm running up against issues duplicating19:58
morganfainbergerm --failfast19:58
morganfainberghow much does that option get used in testing?19:59
morganfainbergdstanek, bknudson, ^20:00
bknudsonmorganfainberg: I've never used it.20:01
bknudsonI just end the job when it fails20:02
morganfainbergif we can get rid of that, i can make run_tests use tox exclusively20:02
bknudsonI don't use run_tests.20:02
morganfainbergbknudson, right, but if did the same thing as tox -epy27 (for example) just w/ subunit output, would you be more inclined to use it?20:02
bknudsonsince the tests are run essentially in random order what's the difference the first one to fail?20:03
morganfainbergbknudson, use the same venv etc.20:03
morganfainbergbknudson, eh, some people like to fix each test along the way20:03
dstanekmorganfainberg: i think i added that for ayoung20:03
morganfainbergor have it explode on the first error vs. just keep going20:03
morganfainbergi personally never use it20:04
bknudsonI've done that with -- --failing20:04
dstanekmorganfainberg: i almost exclusively use nosetests20:04
morganfainbergdstanek, sure.20:05
bknudsonI wouldn't trust that somebody would change run_tests on me. So I'll continue to use tox20:05
bknudsonSince that's what the build does20:06
morganfainbergbknudson, aye20:06
morganfainbergthough i'm going to start using tox -e<env> -- --subunit | subunit-2to1 | tools/colorizer.py20:07
bknudsoncould make an alias for it20:07
morganfainbergbknudson, ++20:08
bknudsonmorganfainberg: can't put it in tox.ini?20:08
bknudsondue to the pipes?20:08
morganfainbergbknudson, well, we could make that the default.20:08
morganfainbergbknudson, but the command needs to be /bin/bash -c "python testr --<things> --testr-args='--subunit {posargs}' | subunit-2to1 | tools/"20:09
morganfainbergbknudson, it's kinda uglty20:09
morganfainbergbut it does work.20:09
morganfainbergalso, not sure if infra wants the "speed of each test printed in the CI logs from jenkins"20:10
bknudsonlet's not break infra20:10
morganfainbergi think i need to submit a feature add for tox, to allow some form of {env:KEY} to just be optional20:12
morganfainbergvs. raising an error of the envvar isn't set20:12
bknudsonLooks like broke on the doc build... started generating warnings.20:12
bknudsonI was able to recreate locally, but don't have time to look at it right now.20:12
*** jamielennox is now known as jamielennox|away20:15
*** serapontiy has quit IRC20:26
gyeeayoung, about
gyeewhy are we still using v2 auth?20:27
gyeeline 450-46020:27
bknudsonmorganfainberg: sounds like tox is re-implementing bash20:29
dolphmanyone know Peter Feiner on IRC? (alias=pete5 according to launchpad)20:30
morganfainbergdolphm, that name sounds familiar20:30
lbragstaddolphm: I've reviewed his code before20:30
*** ayoung has quit IRC20:32
dstanekdolphm: i'm pretty sure i've seen his code, but i've never talked to him on irc20:34
*** leseb has quit IRC20:38
*** raildo has quit IRC20:55
*** leseb has joined #openstack-keystone21:01
*** pete5 has joined #openstack-keystone21:01
*** leseb has quit IRC21:06
*** amcrn has quit IRC21:06
*** packet has quit IRC21:06
*** pete5 has quit IRC21:07
*** pete5 has joined #openstack-keystone21:08
*** packet has joined #openstack-keystone21:11
*** amcrn has joined #openstack-keystone21:12
*** packet has quit IRC21:12
*** jamielennox|away is now known as jamielennox21:17
openstackgerritA change was merged to openstack/keystone: Use config fixture
jamielennoxdolphm: will won't make icehouse if i fix it up? (Guessing no)21:29
bknudsonjamielennox: it's keystoneclient... isn't linked to a release.21:30
morganfainbergjamielennox, keystoneclient isn't linked to a release21:30
morganfainbergbknudson, beat me to it21:30
bknudsonmorganfainberg: jinx!21:30
morganfainbergbknudson OH NOES!21:30
jamielennoxdamnit, wrong link21:30
openstackgerritJenkins proposed a change to openstack/keystone: Updated from global requirements
bknudsonmorganfainberg: I saw that the config fixture went in... that's great.21:31
morganfainbergjamielennox, i would aim that for J121:31
jamielennoxit can be considered a bug fix or a feature addition - if it's a feature then i think i'll drop it because it's ugly and will be much easier with pecan21:31
morganfainbergjamielennox, it's not "broken" in I21:31
morganfainbergjamielennox, land that first thing J121:32
jamielennoxmorganfainberg: debatable21:32
morganfainbergjamielennox, how "broken" is it?21:32
morganfainbergis it "this is really bad an unusable why are we even trying"21:32
jamielennoxno, you can somewhat fix it21:32
jamielennoxok, leave it21:33
morganfainbergjamielennox, i'm not opposed to it going in21:33
jamielennoxyou can put the right url in mostly because no-one really uses the internalURL anyway21:33
morganfainbergif there is a real concern, lets get it in.21:33
jamielennoxthere's about 2 or 3 reviews i put in right after freeze21:34
morganfainbergso... it doesn't have a red x21:34
morganfainberglet me look it over real quick21:34
jamielennoxright, i've got a hold on a couple of others so i wasn't sure21:34
morganfainbergjamielennox, ok so it's bigger than I'd want in this late21:36
morganfainbergbut i could buy that this is a valid item to get in21:36
morganfainbergi think it's a risk evaluation, how likely will this be to break anyone if it lands21:36
jamielennoxalright, i'll fix it today and let us debate it at the meeting21:36
morganfainbergif there is more than a nominal risk, i'd want to hold it.21:36
morganfainberggood idea21:36
jamielennoxbut it's an ugly patch21:36
morganfainbergeh, we've landed worse close to the end21:37
*** jimbaker has quit IRC21:39
openstackgerritJamie Lennox proposed a change to openstack/python-keystoneclient: Add 'methods' to all v3 test tokens
openstackgerritJamie Lennox proposed a change to openstack/python-keystoneclient: Use AccessInfo in auth_token middleware
*** ayoung has joined #openstack-keystone21:42
openstackgerritBrant Knudson proposed a change to openstack/keystone: Remove duplication when picking test config files
openstackgerritBrant Knudson proposed a change to openstack/keystone: Configurable temporary directory for tests
morganfainbergbknudson, +++++++++21:48
lbragstadbknudson: nice!21:48
bknudsonI hope it makes things easier.21:49
*** marekd|email-me is now known as marekd21:49
morganfainbergbknudson, next step in the long road to parallel testing is to rip out the large volumn of "config" files21:53
bknudsonmorganfainberg: this gets rid of one of them.21:53
morganfainbergbknudson, :)21:53
morganfainbergbknudson, i can't fix the LDAP ones and the "livetests" require some special magic to keep working21:54
bknudsonmorganfainberg: it's going to be hard to make the live tests run in parallel... especially ldap21:54
morganfainbergbut i think i can remove 100% of the rest of the configs21:54
marekdstevemar dolphm bknudson ayoung dstanek: Gentlemen, anyone willing to take a look at ? :-)21:54
bknudsonyou'd need a separate tree for each thread21:54
morganfainbergnah, live tests you should run with explicit non-parallel options21:54
bknudsonfor sql you'd need a separate schema (in db2, or database in mysql)21:55
morganfainbergthere are some tools to do that on demand iirc21:55
morganfainbergbut those should (for the time being) only be run in single-worker mode (env var)21:55
bknudsonmorganfainberg: I finally got a computer with more than 2 cpus, so the lack of parallel testing actually affects me now21:56
*** dims_ has quit IRC21:58
morganfainbergbknudson, hehe21:58
stevemarmarekd, i will be looking at it in about an hour or two21:59
stevemarfood time firts21:59
marekdstevemar: thanks a lot. It's enough to have some feedback in next 12 hours :-)22:02
*** marekd is now known as marekd|away22:03
*** stevemar has quit IRC22:04
jamielennoxhey guys, so client review22:05
jamielennoxand if you want more click on my name :)22:06
*** petertoft has joined #openstack-keystone22:09
*** andreaf has joined #openstack-keystone22:10
*** jimbaker has joined #openstack-keystone22:10
*** jimbaker has quit IRC22:10
*** jimbaker has joined #openstack-keystone22:10
openstackgerritldbragst proposed a change to openstack/keystone: Use class attribute to represent 'user' and 'group'
openstackgerritJamie Lennox proposed a change to openstack/python-keystoneclient: Handle Token/Endpoint authentication
*** lbragstad has quit IRC22:15
*** dims_ has joined #openstack-keystone22:16
*** leseb has joined #openstack-keystone22:18
dstanekmarekd|away: i just posted a review - i -1ed it because i think there is a bug22:19
dstanekmarekd|away: but i also think that there maybe a way to make it more reusable22:19
*** nkinder_ has quit IRC22:20
dstanekbknudson: take a look at the review i just did on
dstanekbknudson: i'm interested in hearing your opinion on my idea22:20
bknudsondstanek: I like that idea. I'm not a fan of changing the identity API every time there's a change somewhere.22:20
morganfainbergtotally idle thought...22:21
morganfainberghow opposed to making the identity SQL backend able to be in a different DB would everyone be (not required, optionally)22:21
morganfainbergnot sure how... that would work with oslo.db22:21
*** nkinder has joined #openstack-keystone22:22
bknudsoncan't imagine anyone opposed to that.22:22
morganfainbergbut might be a useful deployment strategy22:22
bknudsonmorganfainberg: the identity SQL backend would have its own engine.22:22
morganfainbergbknudson, yeah but people are opposed to things for odd reasons22:22
morganfainbergbknudson, ah ++ ok22:22
bknudsonwe could have all the sql backends have their own engine22:23
morganfainbergbknudson, i'm kindof thinking that might be a solid idea22:23
morganfainbergbknudson, if undefined use the standard one.22:23
morganfainbergerm "default"22:23
morganfainberge.g. connect string22:23
bknudsonmorganfainberg: I think there's a foreign key from assignment to identity that would have to be broken22:23
morganfainbergbknudson, then it should be broken :)22:23
bknudsonor identity to assignment.22:24
bknudsonotherwise LDAP wouldn't work22:24
morganfainbergi'll toss a BP up for J22:24
openstackgerritJamie Lennox proposed a change to openstack/python-keystoneclient: Convert auth_token to use session
*** petertoft has quit IRC22:32
*** thedodd has quit IRC22:35
morganfainbergbknudson, do you want to stick in the per-testr worker logic in _calc_tmpdir?22:35
morganfainbergbknudson, rather than making that change later?22:35
morganfainbergbknudson, we'll need it in either case, but i'm fine with adding it when we get down the line22:36
bknudsonmorganfainberg: I'd rather make that change later since it's a separate enhancement22:36
morganfainbergchecking before i +2 this change22:36
bknudsonmorganfainberg: that one will probably have to create the directory, too, so it will be more complicated22:37
bknudsonmorganfainberg: maybe it will be as easy as doing a mkdtemp... wasn't thinking about it before.22:38
morganfainbergit will be, but lets just do that when we get to that bridge22:38
morganfainberg+2 across your test changes (4 i count)22:38
bknudsonyou'd get multiple pristine databases but that's no big deal22:38
bknudsonI tried switching some of those tests to in-memory but that didn't work.22:39
bknudsondidn't look into it too much22:39
bknudsonit should work.22:39
morganfainbergi _think_ we could do the pristine db check w/ lockutils magic.22:40
morganfainbergmake 1 pristine db, and copy from that22:40
bknudsonmorganfainberg: right... I'm not sure it's worth it though. They don't take up that much space.22:41
*** bknudson has left #openstack-keystone22:51
openstackgerritMatthieu Huin proposed a change to openstack/keystone: Add missing documentation for enabling federation auth plugin
*** marekd|away is now known as marekd22:55
*** david-lyle has quit IRC22:56
marekddstanek: what bug?22:57
*** gokrokve_ has quit IRC22:58
*** henrynash has quit IRC23:00
marekddstanek: anyway, thanks.I am not sure if I understood you correctly. You suggested to hold off with my changes until we all get some consensus, or with switching to driver's hints?23:00
marekddstanek: ah, just noticed your comment.23:03
marekddstanek: you are probably right....but I would go even further -> compare sets so we are sure all the values are equal, not only lengths.23:04
dstanekmarekd: yeah, the bug is in the comparing of the lists23:05
*** andreaf has quit IRC23:06
mhumarekd, got a few questions about federation, do you have time to talk about it now ?23:06
*** amcrn has quit IRC23:07
dstanekmarekd: i meant hold off on make changes related to my hints comment - i23:07
dstaneki'd like to hear what others have to say on the matter23:07
openstackgerritMatthieu Huin proposed a change to openstack/keystone: Add missing documentation for enabling federation auth plugin
*** leseb has quit IRC23:10
*** amcrn has joined #openstack-keystone23:10
marekddstanek: ok, we can discuss it during tomorrows keystone meeting.23:12
marekdmhu: hey, are you USA/Canada timezone?23:12
mhumarekd, no, french tz23:13
mhubut not sleepy yet :)23:13
marekdI am Switzerland/France tz.23:14
marekdmhu: but, whatever, go ahead :-)23:14
mhumarekd, it's not an emergency, it can wait until tomorrow morning23:15
marekdmhu: I should be at work before 9am our tz, is it okay to talk then ?23:17
mhuI'll send you an e-mail in a minute explaining my problem, I should be at work between 9:30 and 1023:18
mhuwe can talk about it then23:18
mhuthanks and good night !23:18
marekdat least one person who is online during 'normal' hours :D23:18
*** leseb has joined #openstack-keystone23:27
openstackgerritMatthieu Huin proposed a change to openstack/identity-api: Fix for federation token request examples
*** gokrokve has joined #openstack-keystone23:40
morganfainbergayoung, this will (if accepted) make it so we can support things like --failfast in run_tests if using tox23:46
morganfainbergayoung, for now, i can't keep feature parity if we want --failfast support23:46
*** bknudson has joined #openstack-keystone23:52
*** jagee has quit IRC23:52
morganfainbergdstanek, for J are we interested in converting our internal notification/callback stuff to blinker? I know you brought that up.23:57
*** leseb has quit IRC23:57
morganfainbergdstanek, looks like it might be a much cleaner interface23:58
*** leseb has joined #openstack-keystone23:58
morganfainbergdstanek, but I am not sure we could do the de-dupe work we currently do.23:59

Generated by 2.14.0 by Marius Gedminas - find it at!