Tuesday, 2015-01-20

*** atiwari has quit IRC00:02
*** atiwari has joined #openstack-barbican00:02
*** nkinder has joined #openstack-barbican00:06
*** atiwari has quit IRC00:07
*** ayoung has quit IRC00:13
*** reaperhulk has quit IRC00:18
*** reaperhulk has joined #openstack-barbican00:19
*** ryanpetrello has quit IRC00:21
*** dimtruck is now known as zz_dimtruck00:29
*** sigmavirus24 is now known as sigmavirus24_awa00:38
*** reaperhulk has quit IRC00:43
*** bdpayne has quit IRC01:00
*** ryanpetrello has joined #openstack-barbican01:15
*** reaperhulk has joined #openstack-barbican01:18
*** ryanpetrello has quit IRC01:22
*** crc32 has quit IRC01:24
*** kgriffs is now known as kgriffs|afk01:51
*** crc32 has joined #openstack-barbican02:19
*** SheenaG1 has joined #openstack-barbican02:31
*** SheenaG1 has quit IRC02:55
*** zz_dimtruck is now known as dimtruck02:58
*** kebray has quit IRC03:04
*** crc32 has quit IRC03:20
*** SheenaG1 has joined #openstack-barbican04:04
*** SheenaG1 has quit IRC04:08
*** crc32 has joined #openstack-barbican04:13
*** SheenaG1 has joined #openstack-barbican04:18
*** chlong has quit IRC04:19
*** SheenaG1 has quit IRC04:21
*** chlong has joined #openstack-barbican04:28
*** Nirupama has joined #openstack-barbican05:01
*** david-lyle has joined #openstack-barbican05:15
*** dimtruck is now known as zz_dimtruck05:23
*** jaosorior has joined #openstack-barbican06:06
openstackgerritJamie Lennox proposed openstack/python-barbicanclient: Replace httpretty with requests_mock  https://review.openstack.org/14844606:31
openstackgerritJamie Lennox proposed openstack/python-barbicanclient: Convert CRUD tests to requests-mock  https://review.openstack.org/14844706:31
openstackgerritJamie Lennox proposed openstack/python-barbicanclient: Remove cyclical dependency  https://review.openstack.org/14844806:31
openstackgerritJamie Lennox proposed openstack/python-barbicanclient: Additional requests-mock testing  https://review.openstack.org/14844906:31
openstackgerritJamie Lennox proposed openstack/python-barbicanclient: Use the ksc Adapter instead of custom HTTPClient  https://review.openstack.org/14845006:31
jamielennoxugh06:31
*** nkinder has quit IRC06:32
rm_youjamielennox: lol, just dealt with one of those today myself :P06:33
rm_yousucks06:33
rm_youalways check the commit hashes before you type 'yes' :P06:33
jaosoriorJust saw a bit bunch of babrbicanclient patches in zuul. Was wondering what it was about. Now it makes sense06:35
rm_you-----------------------------------------------------------------------------------------------------------------------------------+06:37
jaosoriorrm_you: wha?06:39
rm_youcat <_<06:41
rm_yousorry06:41
jaosoriorjamielennox: so... are you actually submitting all those?06:43
rm_youi am guessing that was a "git review" lulz06:44
rm_youbut could be wrong06:44
rm_youguessing because of "ugh" :P06:44
jaosoriorHaha well that ugh could mean anything06:44
rm_youtrue06:44
rm_youcould be "ugh, just finished and submitted 5 of my own CRs"06:45
rm_youthat'd be ugh-worthy06:45
jaosoriorHahaha yup06:45
rm_youjaosorior: you coming to the midcycle?06:46
rm_youI hope to be there, so may finally meet you IRL, if you're coming06:46
jaosoriorrm_you: you know of hockeynut made any discoveries regarding the JSONDecodeError (now 404) business?06:46
rm_youno06:46
rm_youbut sounds like something that might be fun to take a look at debugging :P06:47
jaosoriorI'll be there06:47
jaosoriorWill book my flights these days06:47
openstackgerritMerged openstack/barbican: Fix UnicodeDecodeError's in the functional tests  https://review.openstack.org/14684506:48
rm_youwhelp06:48
jaosoriorHaha sounds like fun but it's pure pain :P06:49
jaosoriorBuuuuut will take a swing at it today06:49
rm_youlet me know in the morning (my morning) how it went :P06:49
rm_youi'm still waiting on news of progress for castellan before i do work in that repo, so i don't step on any toes06:50
rm_youmeant to ask during the meeting today but i was in the middle of a group merge-clusterf$#@-fixing session06:50
rm_youwhich is what i was referring to earlier -- mine was 7 CRs long though :P06:51
jaosoriorWhat news on castellan?06:51
rm_yousupposedly JHU was doing the first commit06:51
rm_youBriana was working on it last I heard06:51
rm_youI was waiting before I did mine06:51
jaosoriorUok06:52
jaosoriorDidn't know06:52
*** david-lyle has quit IRC06:57
jamielennoxsorry, pressed submit went for coffee07:03
jamielennoxjaosorior: pretty much - i expect to get some problems around patch 307:04
jamielennoxthere is a few changes to how compatability would work and i don't know if anyone will care or not07:04
jamielennoxhopefully the first couple are just ticked off07:05
jamielennoxrm_you: the ugh was just because it was a horrible set of patches that i started because it wouldn't take long07:05
jaosoriorAlright. Will review them today07:06
jaosoriorI'm kind of in a bus chatting on my phone. So hold on :P07:06
jamielennoxthen proceeded to take about a day07:06
rm_youjaosorior: ah, not a bar?07:06
* rm_you makes a checkmark on the "not in a bar" side of his jaosorior whiteboard07:06
jaosoriorHahaha not today07:06
jaosoriorIt's 9am. Not many good ones open07:07
rm_youjamielennox: out of curiosity, what does the change to containers.py to access the secretManager differently have to do with cyclical dependencies?07:07
rm_younot sure I'm seeing it, or else maybe i'm just tired07:07
jamielennoxrm_you: the containers manager calls out to the secrets manager07:08
rm_youtrying to figure out what dependency was cyclical07:08
jamielennoxrm_you: it's the client -> manager -> client07:08
rm_youwell, the client that was passed in07:08
rm_youthat didn't actually look cyclical to me07:08
rm_youthe client calls the manager, passing a copy of itself07:08
jamielennoxreference to itself07:08
rm_youerr, right07:09
rm_youso why can't the manager refer to another manager via that ref?07:09
jamielennoxso it means that the garbage collector struggles to clean up the main client because all the managers are holding a reference to it07:09
rm_youI don't see the problem :P but I guess that's obvious since I wrote the code that does that07:09
rm_youerr, isn't that what you'd want?07:09
jamielennoxso we did the same thing in ksc initially07:09
rm_youi wouldn't imagine you'd want it to be cleaned up if it's still in use (held by a manager)07:10
jamielennoxyou'd create the client, and create the managers07:10
jamielennoxrm_you: right - but because the client is holding a reference to the managers it's not cleaned up either07:10
rm_youhmm07:10
rm_youso once you clean up both... they're both orphaned but held by each other? and so you enter memory leak territory?07:11
jamielennoxi thought python garbage collection was better than that these days, that it would break circular refrences07:11
rm_youthat's ... interesting07:11
jamielennoxhowever we had the problem with horizon creating clients, that created a session which held a connection pool07:11
rm_youso it's a "memory leak" in python?07:11
jamielennoxand the connection pool remained open long after the client had been let go07:11
rm_youthe client doesn't actually hold a connection, IIRC (keystone sessions aren't a keepalive?)07:12
rm_youso it's just the memory for the objects in this case I would assume07:12
rm_youbut I see how that could be a problem for a long-running service07:12
rm_youif you don't hold on to a single barbican-client instance07:12
jamielennoxthey're connection pooled and i think its http 1.1 keep alive07:12
rm_you(the Octavia / Neutron-LBaaS code actually holds one as a singleton)07:12
rm_youhmm07:13
jamielennoxhttp://engineering.hearsaysocial.com/2013/06/16/circular-references-in-python/07:13
rm_youI have to look at how sessions work again07:13
jamielennoxpicked the first one of f the google list07:13
rm_youI didn't THINK they even opened a connection at all until you tried to use them07:13
rm_you(keystone Sessions)07:13
jamielennoxrm_you: they don't, underneath it's all just requests.Sessions07:13
jamielennoxhowever because the client held a reference to the session it wasn't being cleaned up either07:14
rm_youk07:14
rm_youerr07:14
rm_you"Normally Python's garbage collector, which is used to detect these types of cyclic references, would remove it. However, because of the custom destructor (the __del__ method), it marks this item as "uncollectable"."07:14
jamielennoxhmm07:14
jamielennoxi don' have a del07:14
rm_youright07:14
rm_youwhich implies to me that it SHOULD work07:14
jamielennoxhttps://bugs.launchpad.net/python-keystoneclient/+bug/1282089 is the bug against ksc07:15
rm_youhave you tried any of the verification methods it listed to validate the bug?07:15
jamielennoxmaybe i need to read through it again - my memory had just come down to don't have circular references07:16
rm_youobjgraph or manually inspecting the gc array?07:16
rm_youyeah, I mean, I think I'm with you "in a general sense"07:16
jamielennoxrm_you: no i haven't - that was a while ago07:16
rm_youalright, might be worth looking into07:17
jamielennoxso i can change that, the purpose of it all was to try to move towards using the Adapter07:17
rm_youI think I understand the issue now so I can look at it07:17
rm_youmaybe that's what i'll do tomorrow morning07:17
rm_youI have been meaning to learn more about the python garbage-collector07:17
jamielennoxthe idea with adapter was you just pass all the Client(**kwargs) options to it and then make requests via it07:17
rm_youyeah either way I guess I wouldn't -1 your approach if it works07:17
jamielennoxgave consistent options across all clients and let me add options like endpoint_override to all clients just be updating ksc07:18
rm_youI just am not sure it was worth doing if this *wasn't* a problem07:18
rm_youbut that ship has sailed07:18
rm_youi'll review that whole chain tomorrow07:18
jamielennoxthe ksc hack was trivial, but i'd love to know what was being left open if it's only __del__07:18
rm_youI just glanced at the most interesting CR first :P07:18
rm_youwell, you can always checkout an offending commit-point and take a look :P07:19
rm_youbut anyway,yeah i suppose there are other benefits of the new approach07:22
jamielennoxrm_you: have a look through the session and adapter stuff, i don't hae anything pressing for barbican at the moment - i only came over to look as i saw you just released a major version07:24
rm_youheh yeah07:24
rm_youwell, thanks for taking a swing at fixing up the code a bit :P07:24
jamielennoxi'm also fairly biased into just making all the clients consistent on this front07:25
rm_youyes07:25
jamielennoxi'm more or less at critical mass amongst the clients now - even if they all do it slightly differently07:25
rm_youthough eventually the unified openstack client should take care of that... right? lol07:25
jamielennoxthe SDK, probably07:25
jamielennoxi'm not so involved in that one any more - it might still be a while07:25
*** david-lyle has joined #openstack-barbican07:31
*** david-lyle has quit IRC07:35
*** jkf has quit IRC07:37
jaosoriorjust got to the office. Damn, this got pretty interesting07:38
*** chlong has quit IRC07:41
*** david-lyle has joined #openstack-barbican07:42
*** david-lyle has quit IRC07:47
*** miqui_ has quit IRC07:54
*** david-lyle has joined #openstack-barbican08:01
*** david-lyle has quit IRC08:05
jaosoriorjamielennox: you around?08:20
jamielennoxjaosorior: a few more moments08:20
jaosoriorregarding this: https://review.openstack.org/#/c/148446/1/barbicanclient/test/test_barbican.py08:21
jaosoriormiiiight be just me08:21
jaosoriorbut when reading the code, I forget every once in a while that it's a fixture coming from requests_mock08:21
jaosorioryou think it's too much of a nit pick if I ask you to import it with a more verbose name?08:22
jamielennoxjaosorior: i don't mind, leave a -1 and i'll do it tomorrow08:24
jaosoriorthat was all. Just wanted to know if it's just me being kinda picky, or if it makes sense08:24
jaosoriorother than that, looks pretty good08:24
jaosoriorwhat time is it over there?08:25
jamielennox6.2508:25
jamielennoxso time i went home08:25
jamielennox:)08:25
jaosorioryeah man, seems legit08:25
jaosoriorhave a good one08:25
jamielennoxjaosorior: see you later08:26
jamielennoxjaosorior: my normal two letters of a nick and tab means i have almost messaged myself a few times here08:26
jaosoriorhahaha that would be pretty funny08:27
jamielennoxalright cya08:27
*** jamielennox is now known as jamielennox|away08:28
*** woodster_ has quit IRC09:00
*** david-lyle has joined #openstack-barbican09:19
*** chlong has joined #openstack-barbican09:28
*** tkelsey has joined #openstack-barbican10:36
openstackgerritJuan Antonio Osorio Robles proposed openstack/barbican: ** DO NOT MERGE **  https://review.openstack.org/14660811:03
*** atiwari has joined #openstack-barbican11:46
*** atiwari has quit IRC11:51
*** atiwari has joined #openstack-barbican11:53
*** zz_dimtruck is now known as dimtruck12:06
*** atiwari has quit IRC12:07
*** SheenaG1 has joined #openstack-barbican12:55
*** ryanpetrello has joined #openstack-barbican13:13
*** Nirupama has quit IRC13:24
*** SheenaG1 has quit IRC13:27
*** atiwari has joined #openstack-barbican13:29
*** darrenmoffat has joined #openstack-barbican13:35
*** atiwari has quit IRC13:37
*** samueldmq-away is now known as samueldmq13:54
*** lisaclark1 has joined #openstack-barbican14:10
*** rellerreller has joined #openstack-barbican14:31
jaosoriorhockeynut: ping14:42
*** lisaclark1 has quit IRC14:47
*** lisaclark1 has joined #openstack-barbican14:49
openstackgerritJohn Wood proposed openstack/barbican-specs: Support concurrent updates on the orders API resource  https://review.openstack.org/14858714:56
*** sigmavirus24_awa is now known as sigmavirus2414:59
*** dimtruck is now known as zz_dimtruck15:01
*** lisaclark1 has quit IRC15:01
*** dave-mccowan has joined #openstack-barbican15:03
*** atiwari has joined #openstack-barbican15:03
*** lisaclark1 has joined #openstack-barbican15:05
*** atiwari has quit IRC15:10
*** atiwari has joined #openstack-barbican15:10
aleedave-mccowan, ping15:17
dave-mccowanalee, +ack15:17
*** kgriffs|afk is now known as kgriffs15:18
aleedave-mccowan, hey -- any progress on any of the cert patches?15:18
*** paul_glass has joined #openstack-barbican15:20
dave-mccowanalee, not yet.  i was out last week.  i'll ping you with a better answer at the end of today.  it's top of my list now.15:20
aleedave-mccowan, great - thanks15:20
aleerm_work, ping15:24
openstackgerritJohn Wood proposed openstack/barbican-specs: Save order information on created secrets/containers  https://review.openstack.org/14726715:26
*** woodster_ has joined #openstack-barbican15:28
*** atiwari has quit IRC15:29
*** lisaclark1 has quit IRC15:31
*** atiwari has joined #openstack-barbican15:33
*** zz_dimtruck is now known as dimtruck15:34
*** lisaclark1 has joined #openstack-barbican15:35
openstackgerritJohn Wood proposed openstack/barbican-specs: Save order information on created secrets/containers  https://review.openstack.org/14726715:35
*** ayoung has joined #openstack-barbican15:36
*** gyee has joined #openstack-barbican15:38
*** atiwari has quit IRC15:38
openstackgerritJohn Wood proposed openstack/barbican-specs: Support concurrent updates on the orders API resource  https://review.openstack.org/14858715:39
*** SheenaG1 has joined #openstack-barbican15:43
*** jorge_munoz has joined #openstack-barbican15:47
*** lisaclark1 has quit IRC15:47
*** atiwari has joined #openstack-barbican15:56
*** lisaclark1 has joined #openstack-barbican15:58
*** SheenaG11 has joined #openstack-barbican15:58
*** SheenaG1 has quit IRC16:01
hockeynutjaosorior greetings sir16:08
hockeynutsaw your update16:09
jaosoriorhockeynut: hey man16:10
jaosoriorYeah, quite strange16:10
hockeynutthe barbican server was up before the tests were run - we do a curl before running the tests16:10
jaosoriorIf you look at the barbican log from when thr tests succeed. You do see the relevant debugn logs and everything16:11
hockeynutI've been banging my head against a wall trying to see where the logging goes.  running testr locally shows the same behaviour (ie we don;t see the logs from the test)16:11
jaosoriorBut when it fails, there's only the keystonemiddleware...and no barbican stuff there... what the hell16:12
hockeynutyep. it almost like someone took barbican away...hence the 404s16:12
jaosoriorYeah16:13
hockeynutI wonder if there is some timing issue where the server is stopped  before the tests are run16:13
hockeynutor in between when we curl to verify that the  server is there (which works) and when we run the functional tests16:13
hockeynutintermittent problems suck16:13
jaosoriorAnd the testr stuff is the way it works. Subunit will supress stdout because that's where it streams the workers' output16:13
jaosoriorI was reading a bunch on it today... but then have up and decided to use the server's log instead. And that's where I stumbled upon what we saw now16:16
hockeynuti thought the os_capture_stdout env vars would control that so subunit would save the stdout (and stderr)16:19
hockeynutI tried setting them to 0 and 1 ... I think I've seen -1 as well but didn't try that (yet)16:20
*** lisaclark1 has quit IRC16:26
*** kebray has joined #openstack-barbican16:26
*** jorge_munoz has quit IRC16:31
*** lisaclark1 has joined #openstack-barbican16:32
*** jorge_munoz has joined #openstack-barbican16:33
*** jorge_munoz has quit IRC16:36
*** jorge_munoz has joined #openstack-barbican16:38
openstackgerritOpenStack Proposal Bot proposed openstack/barbican: Updated from global requirements  https://review.openstack.org/14862516:39
hockeynutjaosorior I'm focusing on getting the logging (all of it) from the functional tests to show up when we run in the gate.  It all works great locally (since we use nose locally) but in the gate it runs via testr16:47
rm_workdumb question -- can you change the gate to use nose instead of testr? :P16:47
rm_worktestr is so frustrating <_<16:47
reaperhulkno, we can't.16:49
reaperhulkand you should just test with pytest locally and let testr suck in the gate ;)16:49
reaperhulk(pytest is also good for debugging testr problems actually)16:49
rm_workdarn, i thought we had some control over what our gate jobs did16:50
reaperhulkwe totally do, but we're not going to change it16:50
rm_work:(16:50
reaperhulkopenstack politics16:50
rm_work:( :( :(16:51
*** atiwari has quit IRC16:56
openstackgerritMerged openstack/barbican-specs: Workflow documentation is now in infra-manual  https://review.openstack.org/13931016:58
*** jorge_munoz has quit IRC16:58
*** jorge_munoz has joined #openstack-barbican17:12
*** lisaclark1 has quit IRC17:37
*** bdpayne has joined #openstack-barbican17:38
openstackgerritMerged openstack/barbican: Updated from global requirements  https://review.openstack.org/14862517:42
*** chlong has quit IRC18:08
*** jkf has joined #openstack-barbican18:14
*** gyee has quit IRC18:30
*** lisaclark1 has joined #openstack-barbican18:31
*** jorge_munoz has quit IRC18:33
*** tkelsey has quit IRC18:43
*** lisaclark1 has quit IRC18:49
*** lisaclark1 has joined #openstack-barbican18:53
*** jkf has quit IRC18:57
*** jkf has joined #openstack-barbican18:59
*** lisaclark1 has quit IRC19:05
*** kebray has quit IRC19:08
*** lisaclark1 has joined #openstack-barbican19:10
*** kebray has joined #openstack-barbican19:17
*** crc32 has quit IRC19:26
*** david-lyle has quit IRC19:37
*** openstackstatus has joined #openstack-barbican19:44
*** ChanServ sets mode: +v openstackstatus19:44
*** bdpayne has quit IRC19:51
-openstackstatus- NOTICE: Gerrit restarted to address likely memory leak leading to server slowness. Sorry if you were caught in the restart19:53
*** atiwari has joined #openstack-barbican20:00
*** paul_glass has quit IRC20:01
*** paul_glass has joined #openstack-barbican20:06
*** gyee has joined #openstack-barbican20:10
*** bdpayne has joined #openstack-barbican20:13
*** lisaclark1 has quit IRC20:31
openstackgerritMerged openstack/barbican: Return the actual name value for entities  https://review.openstack.org/14549520:33
*** lisaclark1 has joined #openstack-barbican20:35
*** lisaclark1 has quit IRC20:40
openstackgerritMerged openstack/barbican: Add data model entities for OrderRetryTask  https://review.openstack.org/14716020:58
*** lisaclark1 has joined #openstack-barbican21:00
*** lisaclark1 has quit IRC21:04
*** SheenaG11 has quit IRC21:14
*** SheenaG1 has joined #openstack-barbican21:20
*** SheenaG11 has joined #openstack-barbican21:23
*** SheenaG1 has quit IRC21:25
*** lisaclark1 has joined #openstack-barbican21:39
*** samueldmq is now known as samueldmq-away21:46
*** crc32 has joined #openstack-barbican21:47
*** ayoung has quit IRC21:47
openstackgerritSteve Heyman proposed openstack/barbican: ** DO NOT MERGE **  https://review.openstack.org/14660821:49
*** ayoung has joined #openstack-barbican21:53
*** dave-mccowan has quit IRC22:07
*** kebray has quit IRC22:17
*** ryanpetrello has quit IRC22:44
rm_workwoodster_: which one is the content-types spec?22:45
rm_workI thought it merged already, maybe there is a different one? or maybe I am nuts22:45
rm_worktoo many reviews happening simultaneously for me to track anymore <_<22:45
woodster_rm_work: https://review.openstack.org/#/c/14507322:46
rm_workwoodster_: just commented with a question of my own on the per-secret CR22:46
*** lisaclark1 has quit IRC22:47
*** rellerreller has quit IRC22:48
woodster_rm_work, cool, I'll take a look shortly.22:48
*** paul_glass has quit IRC22:51
openstackgerritBrianna Poulos proposed openstack/castellan: Copy cinder.keymgr to castellan  https://review.openstack.org/14874222:52
*** alee has quit IRC22:53
rm_workAlright!22:53
*** gyee has quit IRC22:53
*** gyee has joined #openstack-barbican22:54
*** rtom has joined #openstack-barbican22:57
*** ryanpetrello has joined #openstack-barbican22:58
*** gyee has quit IRC23:06
*** atiwari has quit IRC23:14
*** dimtruck is now known as zz_dimtruck23:19
*** jamielennox|away is now known as jamielennox23:36
*** chlong has joined #openstack-barbican23:41
*** lisaclark1 has joined #openstack-barbican23:43
*** lisaclark1 has quit IRC23:43
*** lisaclark1 has joined #openstack-barbican23:43
*** chlong has quit IRC23:45
*** chlong has joined #openstack-barbican23:46
*** rtom has quit IRC23:49
*** jkf has quit IRC23:54
*** ayoung has quit IRC23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!