Monday, 2015-02-23

*** claudiub has quit IRC00:00
*** ybathia has quit IRC00:32
openstackgerritJim Mankovich proposed openstack/ceilometer-specs: Generic Notification Sensor Meter Plugin Design Spec
*** boris-42 has joined #openstack-ceilometer01:13
*** kbyrne has quit IRC01:23
*** kbyrne has joined #openstack-ceilometer01:25
*** promulo has joined #openstack-ceilometer01:30
*** promulo__ has quit IRC01:34
*** ddieterl_ has joined #openstack-ceilometer02:06
*** ddieterly has quit IRC02:09
*** shengjiemin has quit IRC02:17
*** himangi has quit IRC02:58
*** deepthi has joined #openstack-ceilometer03:37
*** ddieterl_ has quit IRC03:47
*** _nadya_ has joined #openstack-ceilometer04:10
*** rwsu-afk has quit IRC04:19
*** _nadya_ has quit IRC04:34
*** himangi has joined #openstack-ceilometer04:52
*** pcaruana has quit IRC05:45
*** ddieterly has joined #openstack-ceilometer05:49
*** ddieterly has quit IRC05:53
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Imported Translations from Transifex
*** ildikov has quit IRC06:38
*** ddieterly has joined #openstack-ceilometer06:50
*** yatin has joined #openstack-ceilometer06:51
*** ddieterly has quit IRC06:54
*** himangi has quit IRC06:58
*** ildikov has joined #openstack-ceilometer07:32
*** ddieterly has joined #openstack-ceilometer07:50
*** ddieterly has quit IRC07:55
*** theanalyst has quit IRC08:14
*** theanalyst has joined #openstack-ceilometer08:15
*** eglynn has joined #openstack-ceilometer08:16
*** Ala has joined #openstack-ceilometer08:21
openstackgerritIldiko Vancsa proposed openstack/ceilometer: Remove duplications from docco
*** ildikov has quit IRC08:50
*** ddieterly has joined #openstack-ceilometer08:51
*** ddieterly has quit IRC08:56
*** ildikov has joined #openstack-ceilometer09:06
*** yatin has quit IRC09:12
*** yatin has joined #openstack-ceilometer09:25
*** ddieterly has joined #openstack-ceilometer09:52
openstackgerritIldiko Vancsa proposed openstack/ceilometer: Remove duplications from docco
*** ddieterly has quit IRC09:56
*** ParsectiX has joined #openstack-ceilometer10:02
*** claudiub has joined #openstack-ceilometer10:03
openstackgerritMehdi Abaakouk proposed openstack/ceilometer: alarming: add gnocchi alarm rules
openstackgerritMehdi Abaakouk proposed openstack/ceilometer: api: move alarm rules into they directory
*** yatin has quit IRC10:42
*** kbyrne has quit IRC10:49
*** kbyrne has joined #openstack-ceilometer10:49
*** ddieterly has joined #openstack-ceilometer10:53
*** yatin has joined #openstack-ceilometer10:55
*** ddieterly has quit IRC10:57
*** himangi has joined #openstack-ceilometer11:16
*** yatin has quit IRC11:22
*** ofcourseistilllo is now known as straycat_11:28
*** straycat_ is now known as straycat11:29
*** nellysmitt has joined #openstack-ceilometer11:29
*** cdent has joined #openstack-ceilometer11:32
*** yatin has joined #openstack-ceilometer11:40
*** deepthi has quit IRC11:45
*** ddieterly has joined #openstack-ceilometer11:53
*** ddieterly has quit IRC11:58
*** cdent has quit IRC12:18
openstackgerritIldiko Vancsa proposed openstack/ceilometer: Remove duplications from docco
*** exploreshaifali has joined #openstack-ceilometer12:25
*** cdent has joined #openstack-ceilometer12:41
*** yatin has quit IRC12:47
*** zigo has quit IRC12:52
*** ddieterly has joined #openstack-ceilometer12:54
*** zigo has joined #openstack-ceilometer12:56
*** ddieterly has quit IRC12:58
*** yatin has joined #openstack-ceilometer13:00
*** yatin has quit IRC13:02
linuxhermitgood morning13:08
*** ParsectiX has quit IRC13:10
*** ParsectiX has joined #openstack-ceilometer13:16
*** lsmola has quit IRC13:23
* cdent waves at linuxhermit 13:23
cdentCan I get your opinion on something linuxhermit? Look at this:
cdentnote the number of xfails and links to bugs13:24
linuxhermitI like that13:24
cdentMost of those bugs represent opinions on API design, rather than hard and fast rules. So I'm not sure whether to really care. What do you think?13:24
linuxhermitit explains the xfail13:24
linuxhermitwell as an outsider13:24
linuxhermitthose type things drive me crazy13:25
linuxhermitlike sending data with get13:25
* cdent nods13:25
linuxhermitbut I'm not casting stones because there is a whole wide world of ways to do things13:25
linuxhermithere is my concern13:25
linuxhermitwith doing this, will we be diligent enough to come back and fix the tests when the bugs are gone?13:26
linuxhermitif you think yes, then I'm all for it13:26
cdentthat's the main reason I've dedided to actually make bugs (and link them in) rather than just make xfails13:26
linuxhermitcool then I'll do it for mine that works wierd like that13:26
cdentI think it is far more likely the bugs will get fixed in gnocchi (as yet unreleased) than in the main ceilo core13:27
cdentBut I think it is worthwhile being optimistic :)13:27
*** lsmola has joined #openstack-ceilometer13:27
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements
linuxhermitI can go with that13:28
cdenteglynn: you have any thoughts on the above?13:29
*** gordc has joined #openstack-ceilometer13:30
cdentwhat about you gordc have you got an opinion on the stuff that linuxhermit and I were just talking about:
linuxhermitI've got one I'm a touch confused on13:39
linuxhermitso if I have       {"q": { "field": "event_type", "op": "eq", "type": "string" "value": "cookies_chocolate.chip"}}13:39
linuxhermit that I want to send as data13:39
linuxhermitdo I do data: then that or do I need to yamlfy it in someway13:40
linuxhermitor just enclose it as a string?13:40
cdentyou're going to send that as the body?13:40
cdentmaybe this
* cdent is not too savvy on the complex query format13:41
cdentbut that's how you can state a json object in the data key13:41
linuxhermitServerError: {"error_message": {"faultstring": "string indices must be integers, not str", "debuginfo": null, "faultcode": "Server"}}13:42
linuxhermitsomething about that it doesn't like I'll keep digging13:42
*** eglynn has quit IRC13:42
linuxhermitI think it sent the data correctly13:43
linuxhermitI just don't think the server liked what it got13:43
ildikovcdent: that example above is simple query actually:
ildikovlinuxhermit: ^^^13:43
openstackgerritafazekas proposed openstack/ceilometer: requests-aws is runtime dependency
linuxhermitThanks ildikov for the link13:44
cdentildikov: I was meaning complex in the sense that it does in the body13:44
linuxhermitin this case it was more gabbi + dat ain get body13:44
cdentlinuxhermit: you want a - in front of a field13:44
cdentto make it a list13:44
ildikovcdent: in the body?13:44
linuxhermitit's not a list though right13:44
cdenti'll update the paste, just a sec13:44
linuxhermitit's q is a json object13:44
ildikovcdent: hmm, I guess I should check back for context :)13:44
cdentin here the value of q is a list:
* cdent boggles13:46
openstackgerritafazekas proposed openstack/ceilometer: requests-aws is a runtime dependency
linuxhermithmm okay13:47
*** thomasem has joined #openstack-ceilometer13:49
*** eglynn has joined #openstack-ceilometer13:53
gordccdent: whoops. stupid irssi doesn't highlight my name if it's in middle of sentence13:53
*** ddieterly has joined #openstack-ceilometer13:55
gordccdent: what's the question? if it's cool to add a FIXME to testcase?13:58
cdentgordc: if the gripes about the api are even valid in the first place13:59
cdent(e.g. 404 on an intervening path being bad)13:59
*** ddieterly has quit IRC14:00
gordc404 seems to make sense for a bad path. is there a better match?14:00
*** eglynn has quit IRC14:01
cdentgordc: the example in those cases is like /v1/resources being 404 when /v1/resource/generic is 200 with a []14:02
cdentseems like resources should return something, not a 40414:02
cdentso should /v114:03
gordcare you using v1 as just an example? (since v1 doesn't exist anymore)14:03
gordcoh wait. gnocchi14:03
gordci'd probably lean more towards 404 to be honest... having it return 20x and empty makes it seem like it's suppose to be an actionable path (ie. under some conditions it would return something useful)14:05
cdentthe bugs explain it a bit more:
openstackLaunchpad bug 1423990 in Gnocchi "/resource returns 404 should probably be list of resource types" [Undecided,New]14:06
openstackLaunchpad bug 1424627 in Gnocchi "/v1 returns 404, should maybe be json links" [Undecided,New]14:06
cdentthe basic question is: is lack of hypermedia a bug?14:07
gordccdent: ah i see... make it more a 'get_all' response.14:07
gordccdent: yeah i could see it being useful to list all resources.14:08
*** eglynn has joined #openstack-ceilometer14:13
*** eglynn-office has joined #openstack-ceilometer14:14
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
linuxhermitcdent pushed up a new version with the suggestions you made14:16
linuxhermitgordc in one of you doc patches, do you already have a fix for
linuxhermitusing event_type twice instead of event_type and trait_name14:18
* cdent looks14:18
gordclinuxhermit: nope i don't have a fix for that.14:18
linuxhermitcool I'll get one rolling14:18
gordci think you'll need to modify api code for that (i think it's generated by wsme-sphinx or something along that line)14:19
gordclinuxhermit: cool cool14:19
*** exploreshaifali has quit IRC14:21
linuxhermitgordc never mind looks like it's fixed in code but the docs haven't been regened since then14:21
gordclinuxhermit: odd. i guess you have to directly touch the rst files to trigger a refresh of docs14:24
gordcor i pointed you in the wrong direction.14:24
linuxhermitseems to be where the controller is, and in the docs it points to that using the rest-controller:: directive14:26
cdentlinuxhermit I've found it useful on bugs like this one: to provide the gabbi-formatted yaml that tickles the bug. It provides a clear replication strategy and spreads the love.14:29
openstackLaunchpad bug 1424642 in Ceilometer "Wrong Content Type results in Server error" [Undecided,New]14:29
linuxhermitadded cdent14:31
cdentcool thanks14:31
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
linuxhermitthose two commends are also fixed and pushed back up as well14:33
cdenteglynn, eglynn-office, gordc You guys might want to have a look at
cdentSean asks a question I've been asking myself since last May.14:33
gordcwhoops... that was me trying to commit something14:34
gordcbut yes i will look to (sort of have to now)14:34
cdentgordc has a depedendent patchset!14:34
eglynnswamireddy: ^^^ note also14:35
gordchmm..yeah we need a way to catch it. i agree with sdague14:35
cdentI agree with sdague too.14:36
cdentWe shouldn't be dumping the kitchen sink in setup.cfg14:36
cdentbecause even if we do conditionally import the S3Auth thing, it will always get imported because stevedore will go romping along and gather the code, won't it?14:37
*** ddieterly has joined #openstack-ceilometer14:37
cdentgerrit needs +1 buttons on comments14:37
* cdent switches to facebook to review code14:38
*** claudiub has quit IRC14:38
gordccdent: make a script.14:38
cdentif user == "gordc": plus_one()14:38
gordcthat's going to trigger a bug to be created14:40
gordcso apparently that bug is breaking the gate.14:40
cdentI reckon we should consider reverting the rgw stuff?14:40
gordccdent: seems agressive. i do wonder how it passed to begin with...14:42
cdentSomething else must have changed around the same time.14:42
cdentThe way requests is loaded in the gate changed recently.14:42
cdentso that might have had some kind of impact14:43
eglynnthis only landed on what, Saturday?14:43
gordccdent: yep14:43
cdentnearabouts, yeah, eglynn14:43
eglynnTBH I'd be OK with a revert until we have a chance to upack this14:43
eglynnswamireddy: ^^^ we can re-commit once the aws dependency is sorted out14:44
cdentThis problem is present in every other plugin which has external dependencies, isn't it? It just so happens that this particular failure has caused some upset because of timing. Our fix in all other times has been to put the missing library in requirements.txt, has it not?14:46
cdentSo while I agree with sdague in principle, if we to take his word as gospel we'd have a _lot_ of plugins to revert, wouldn't we? (which would be ossum, but that's a different discussion)14:48
gordccdent: this shouldn't be loaded. ceph requirements should only be present if you specifically want to to poll ceph14:48
cdentso what about pysnmp14:48
cdentor tooz (which is only needed is doing coordination)14:48
gordci would say pysnmp shouldn't be there... tooz is actually pretty well integrate across most of the core services so it should be there (even if disabled)14:49
gordcjust my opinion.14:49
cdentor requests itself, which seems to be used by opendaylight14:50
cdentbut not generally14:50
* cdent is just belaboring the point14:50
gordcyes you are!14:50
cdentThe way plugins are done is very wrong.14:50
gordcstop fanning flames.14:50
gordchead back in the sand14:50
* cdent takes the role of burninator aaway from gordc14:50
* cdent stomps and burns14:50
gordcoh that's what you wanted to do? that's cool then.14:51
* gordc gives cdent the flamethrower.14:51
* cdent phwwoooops14:51
cdentokay, I'm done now.14:51
cdentWhat should we actually do, now, for the short term?14:51
gordceglynn: you going to push a revert patch?14:52
eglynngordc: on a call, can do it later if no one else picks it up in the meantime14:52
gordceglynn: sure thing14:52
gordccdent: see an easy fix? or revert? i can wrap the import and log it which would fix startup14:53
cdentDo we have an existing standard for this kind of thing? Do we have a way of configuring plugins such that they exist in setup.cfg but don't actually run?14:54
gordcnot sure if it makes sense though in real life usage. (it'll log like crazy i think)14:55
cdent(entry_points as configuration is...weird)14:55
cdentI think we should a) revert b) use this an opportunity to fix the real issue that's at the base of this.14:55
gordclet's give it 5 minutes of discussion... then i revert.14:55
cdentk, ideas?14:56
cdentIncreasing logging will not be looked upon favorably, so I wouldn't suggest any course that leads that direction...14:56
cdentIs a polling agent only __init__ at startup?14:58
cdentIf so, then a plugin that fails to import something could raise: plugin_base.ExtensionLoadError14:58
cdentand only log at __init__ tie14:59
gordcseems like it.14:59
cdentor, the _extensions method in the agent base could also catch ImportError14:59
cdent(and log once, and be done with it)15:00
cdentthat seems more in keeping with the idea of plugins (they should warn, but not cause the system to die, when they can't work)15:00
gordccdent: what gate are we failing right now? how are patches still passing?15:01
cdentthat's a test that ceilo's gate doesn't run15:02
cdent(I think)15:02
gordcah i see.15:03
gordci think the extenstionloaderror might be a sane solution... seems like that's the very reason the exception was created15:03
*** prad has joined #openstack-ceilometer15:06
cdentgordc it seems our quorum of two has reached a consensus15:08
gordccdent: so we want to revert still to guaruntee unblock?15:08
cdentI'm not sure. I don't understand the nature of the block well enough to know if we can just fix and be happy, or need to do two steps?15:09
gordccdent: let's just revert for now. seems simpler.15:10
cdentas you wish15:10
openstackgerritgordon chung proposed openstack/ceilometer: Revert "Add ceph object storage meters"
openstackgerritafazekas proposed openstack/ceilometer: Revert "Add ceph object storage meters"
cdentuh oh15:17
*** ccrouch has quit IRC15:18
swamireddyeglynn: Please let me know, what was the issue?? I could NOT get it..15:18
swamireddycdent: Hi15:20
cdentHi swamireddy15:20
swamireddycdent: Could you please clarify issue on Revert  "Add ceph object storage meters"
swamireddycdent: I could FIX it asap, if I know the issue...plz15:21
cdentWe're not sure why the problem is only showing up in some tests, but requests-aws was initially in test-requirements.txt but since the rwg stuff runs unconditionally it is always imported15:22
*** claudiub has joined #openstack-ceilometer15:22
cdentsomebody tried to fix this by moving requests-aws to requirements.txt but this wasn't the preferred fix from the qa people. They want the import to be conditional.15:22
cdentSo the fix (which will come along once the gate has unblocked) is to make the rwg raise ExtensionLoadError when it can't import the awsauth module.15:23
cdentDoes that makes sense?15:23
cdentIn the meantime, to unblock things, the revert is happening.15:23
swamireddycdent: I have added the requests-aws pkg in global-requirements.txt and test-requirement.txt only (NOT included in the requirements.txt and requirement-py3.txt)15:24
cdentyes, that's where it is best15:24
cdentbut since the rwg pollster plugin is always loaded, the import fails with an error:15:24
cdentThat error needs to be trapped and transformed intoa ExtensionLoadError15:25
swamireddycdent: Thats right ...I was told that any needed the rgw polling, that use will import the required pkgs (ie requests-aws)15:25
*** rbak has joined #openstack-ceilometer15:26
swamireddyquick fix could be- add the request-aws import to requirement.txt, all will set...15:26
cdentYeah, the code you provided is mostly correct, except for the need for the new exception15:26
cdentThe qa folk not keen on that fix, they want the better longer term fix.15:26
gordcswamireddy: could you try adding a try catch around awsauth requirement in client15:27
swamireddysure, I wil take-up and fix it asap...its not issue...But one fix could be - add the requests-aws to requirement.txt, so import will not fail, and gate will this OK?15:27
cdentswamireddy: They don't want that.15:28
gordcswamireddy: let's avoid that. we have some time to fix it (just a temporary revert to give us time to do clean fix)15:28
swamireddygordc: sure, I can try and update you....but, if we add the requests-aws to requirements.txt, will all set...15:28
cdentBecause it starts the precedence of having all plugins adding stuff to requirements.txt15:28
cdentgordc: who else is active core these days?15:28
gordcsileht: might be back from holiday...15:29
gordccdent: why do you need a core?15:30
cdentI don't personally need a core, but that review does, to punch it through.15:30
*** Ephur has joined #openstack-ceilometer15:31
ildikovcdent: the revert one?15:31
gordclet's take your opinion as pseudo +2 and eglynn already suggested it15:31
cdentildikov: yeah15:31
gordcand ildikov says yes.15:31
gordci making assumptions.15:32
dhellmanngordc, cdent : I still have have +2 on ceilometer, I think15:32
swamireddycdent: gordc: Sure, let me fix by adding an exception and will come back15:32
gordcdhellmann: yep!15:32
cdentWhen we fix it I think it is worthwhile considering trapping ImportError in plugin base as a safety valve. ImportError is an implicit ExtensionLoadError, n'est as?15:32
dhellmanngordc: this one?
* cdent le sighs15:32
ildikovgordc: ildikov has already written the OS Manuals table for Ceph, so she sadly agrees with this particular revert... :S15:33
gordcdhellmann: yes please. disclaimer (all i did was hit revert button, not tested locally but should be good)15:33
swamireddycdent: Agreed..will add the ImportError..15:33
linuxhermitthe dhellmann with the trump15:33
gordcildikov: i think it's ok. we should have it back in cleanly by today/tomorrow15:33
*** rbak has quit IRC15:33
silehtwoot, revert just for a import failure, sniff15:33
cdentswamireddy: I'm meaning in plugin base, not in the plugins themselves15:33
dhellmanngordc: approved15:34
gordcdhellmann: thanks15:34
ildikovgordc: sure, I know, it's just a typical Monday happening... :)15:34
ildikovdhellmann: thanks15:34
cdentI think reverting first asking questions later is a reasonable way to keep things moving, no sileht ?15:34
*** zigo_ has joined #openstack-ceilometer15:35
cdentPresumably we should never have let this through in the first place if we were really paying attention?15:35
dhellmanncdent, gordc, ildikov : is someone working on the fix in ceilometer/agent/ to trap the import error, or should I?15:35
* dhellmann notes he has to disappear for the oslo meeting in ~20 minutes15:35
* cdent doesn't know15:36
*** rbak has joined #openstack-ceilometer15:36
swamireddycdent: gordc: Shall I await the fix the plugin base?15:37
gordcdhellmann: i think swamireddy was going to look at it? i think there's some existing infrastructure that we can leverage... revert was just to let us fix it without rushing.15:37
*** alexpilotti has joined #openstack-ceilometer15:37
gordcswamireddy: sure15:37
gordcdhellmann: i guess not. so if you want to look at it... or i can. either or.15:38
dhellmanngordc: I'm looking for a place to put a test. The fix is < 1 line.15:38
*** exploreshaifali has joined #openstack-ceilometer15:38
gordcdhellmann: :) well that's better than what i had planned15:38
dhellmanngordc: actually, logging turned it into 3-4 lines :-)15:39
*** yeungp has joined #openstack-ceilometer15:39
ildikovdhellmann: :)15:39
gordcdhellmann: you disappoint me. :P15:40
*** yeungp has quit IRC15:40
*** yeungp has joined #openstack-ceilometer15:40
dhellmanngordc: heh15:42
silehtcdent, yes I'm ok with that but a bit sad when it occurs15:44
cdentlinuxhermit++; making tests and reporting bugs for the win15:45
cdentsileht: true enough15:45
gordcdhellmann: cdent: since i have you, trying to add ceilometermiddleware to devstack but getting this:
* cdent looks15:46
gordcdo i actually have to add ceilometermiddleware to global-requirements.txt (even though it isn't going to be in any requirements.txt)15:46
linuxhermitcdent: just trying to help15:46
linuxhermitwill submit fixes for code once I feel more comfortable with it15:46
gordci should mention i put it in projects.txt (just not requirements.txt)15:47
cdentgordc: what's the review associated with that?15:48
cdentthat's what I thought but I failed to get gerrit to let me type in a url.15:49
* cdent shivers at gerrit15:49
cdentgordc: you've still got to install the middleware15:50
cdentpresumably in lib/swift15:50
gordcerr add a pip install ceilometermiddleware?15:51
cdentneither ceilometer nor swift will15:51
cdentprobably here:15:51
*** zhenq has joined #openstack-ceilometer15:51
cdentpip_install ceilometermiddleware15:51
cdentassume it is on pypi?15:52
gordccdent: yep15:52
openstackgerritDoug Hellmann proposed openstack/ceilometer: Protect agent startup from import errors in plugins
gordci'll do that. wasn't sure if i needed to do it explicitly or not. i guess i do15:52
dhellmanngordc, cdent, ildikov : ^^15:52
gordcdhellmann: wicked!15:53
ildikovdhellmann: thanks, looking15:53
cdentwe should look around and see if there are similar places in the code where stevedore is being used15:54
cdent(that's probably the biggie, though)15:54
dhellmanncdent: good idea -- you, gordc, and ildikov should feel free to take that patch over if you need to extend it with more cases15:54
gordci wonder if we should kill the revert... too many decisions to make on monday morning!15:55
ildikovdhellmann: sure, thanks much for the quick fix15:55
cdentgordc: most of the time you don't have to install python modules in devstack because it is already in some projects requirements but since the middleware thingie is standalone, you have to make special effort15:55
ildikovgordc: how long is the queue?15:56
cdentgordc: did we determine how wedged the wedge is?15:56
swamireddygordc: Sorry...I would have taken care..15:56
dhellmanngordc: no, leave the revert, it will probably land before my change and it should still be testing the import itself explicitly15:56
cdentI thought there was some fear that things wouldn't merge without a revert?15:56
gordcdhellmann: cdent: sounds good.15:56
gordcswamireddy: i guess if you can reapply your patch on top of dhellmann's patch we can verify if it solves your case or not15:57
openstackgerritMehdi Abaakouk proposed openstack/ceilometer: alarming: add gnocchi alarm rules
swamireddygordc: will test locally and confirm..15:58
gordcswamireddy: thanks! sorry for reverting. was just simpler.15:59
gordci'll re-review right away after you've confirmed.16:00
swamireddygordc: sure....16:00
gordccdent: damn, i see the pip_install used everywhere in devstack... thanks for the pointer.16:01
gordcildikov: did you leave off the 'non-metric' meters from docs?16:08
ildikovgordc: I'm afraid I don't really understand your question :)16:08
ildikovgordc: it contains all meters16:09
gordcactually never i see the meters i was trying to block... i was going to ask about vpnaas, lbaas..16:09
ildikovgordc: I can create a deprecated column if we disable them in Liberty by default16:10
ildikovgordc: but for now I sadly have to include all in the docs16:10
gordcildikov: that's fine. i don't know if we ever formalised "this is deprecreated"16:11
swamireddygordc: cdent: I apologize for this glitch on Monday morning..:(16:11
gordci'm wondering how we can mark them as 'this is an event'16:11
cdentswamireddy: I think its a good thing actually: It allowed us to find a quite important bug in the very good of the polling agents.16:11
gordcswamireddy: no worries. i approved it... and i break stuff all time.16:12
ildikovgordc: I guess deperecation itself means keeping them for two cycles16:12
cdentBreaking stuff is the quickest way to finding and fixing things.16:12
gordci find it healthy to break stuff.16:12
ildikovgordc: so here we can keep the "support" for collecting them if someone really wants16:12
* cdent breaks gordc's arm16:13
gordcildikov: yeah i think that's fine. do you think there's a way to hint "you should really use events"16:13
ildikovgordc: I added a note about it somewhere IIRC16:14
gordccdent: i've seen people type/drum/eat/etc... with their feet. one arm is nothing16:14
gordcildikov:  oh cool!16:14
ildikovgordc: also in the dev docs I included one into the new meters section16:14
* cdent breaks gordc's other arm and both legs16:14
ildikovgordc: I will double check the OS Manuals side16:14
ildikovgordc: the other one I'm sure, but it's not merged yet anyway16:14
gordccdent: that's cruel.16:15
* cdent is the burninator16:15
ildikovcdent: you should not break gordc's arms, because you have to help me to maintain the docs then :)16:18
gordcildikov: type with eyes.16:18
* cdent takes it all back16:19
ildikovgordc: sometimes I wish I could :)16:19
ildikovgordc: I checked the source of the api reference today16:19
ildikovgordc: that is a killer one, I mean the OS Manuals version not ours16:20
gordci haven't seen it.16:20
ildikovgordc: I suggest to keep it this way...16:21
ildikovgordc: looks ugly in the sense of more xml tags and less text16:21
gordcbut tbh, anything api is madness in openstack (cdent's problem)16:21
*** devlaps has joined #openstack-ceilometer16:21
ildikovgordc: I wonder if there is a chance to generate it somehow16:22
gordclike what we have in our dev docs?16:22
cdentyou're not talking about the API wadl are you?16:22
ildikovgordc: got you, as for the API problem16:22
gordci just zone out when people say api16:22
ildikovgordc: yeap, somehow like, the OS Manuals version uses wadl format16:22
ildikovgordc: but the file does not look like a generated one though16:23
*** zhenq has quit IRC16:25
*** zhenq has joined #openstack-ceilometer16:25
*** amalagon has quit IRC16:27
cdentildikov: I made the same conclusion. I think that's hand (or close to) generated WADL.16:28
*** jdandrea has quit IRC16:28
*** jdandrea has joined #openstack-ceilometer16:29
ildikovcdent: I didn't find any suggestions in the README file, where there usually is16:29
ildikovcdent: and also the generated files contain a not at the beginning that yous hould not edit it manually as those are generated files16:29
ildikovcdent: so I'm almost sure16:30
cdentI wonder what from then. It's all very opaque16:30
ildikovcdent: and finally, if there would be a script that generate those files, then it would be up to date and it's not...16:30
*** zhenq has quit IRC16:32
*** zhenq has joined #openstack-ceilometer16:32
*** Erethon has left #openstack-ceilometer16:33
gordcildikov: do you have a link where you mention "store this as events" i'm thinking we should explicitly mark it in each table too16:36
gordci think i remember that in operator thread it mentioned the meters are confusing (or soemthing about description is not helpful)16:37
*** abhi- has joined #openstack-ceilometer16:37
ildikovgordc: they meant the libvirt ones there as I remember, they mentioned that they had to check the libvirt api doc for info...16:38
gordcah right16:38
ildikovgordc: I tried to polish them a bit, it would need some more for sure16:38
gordcthat sounds familiar16:38
ildikovgordc: but marking those meters are good idea anyway16:38
*** ddieterly has quit IRC16:39
ildikovgordc: that is why I tried to convince Saba as for the disc meters, but then I rewrote those descriptions in the OS Manuals version...16:39
gordcyeah, i think we can polish it but i can't think of a better description "it's trying to explain it as something it isnt"16:39
ildikovgordc: for the event ones for sure16:39
gordcok. maybe we can mark them after you move everything over16:40
ildikovgordc: amybe we could have an additional column as "event type metric" or smth like...16:40
*** fnaval has joined #openstack-ceilometer16:41
gordcildikov:  i don't know if we need another column. we can probably highlight column or add asterisk and mention it in bottom of each table?16:41
ildikovgordc: see the note at the bottom of the page16:41
*** devlaps has quit IRC16:42
ildikovgordc: well, maybe we can play with the background color or add a maker after the name of each one16:42
gordccool cool. note make sense.16:43
ildikovgordc: we have time to decide it until all the tables get merged and then it can be marked in a single patch16:43
gordcthat sounds like a good plan.16:44
ildikovgordc: cool, I never know if what I wrote means as much sense in English as I would like it to...16:44
ildikovgordc: cool, agreed then :)16:44
gordcmakes sense to me. although i sort of make up my own sentences when people talk to me.16:44
ildikovgordc: I usually invent new expressions in English that are "similar" to the original ones... :)16:46
ildikovgordc: ok, I need to run now, I'm close to starve to death16:46
gordci think we all do... and then we add it to :).. or maybe just me.16:46
gordccool cool. ttyl16:47
ildikovgordc: I will check back to your OS Manuals patches after16:47
ildikovgordc: LOL :), it's good to think I can be the part of building a new dictionary/language :)16:47
*** devlaps has joined #openstack-ceilometer16:48
ildikovgordc: np, laters :)16:48
abhi-gordc: regarding the ceph/rgw requests-aws import failure, wouldn't requests-aws be required in requirements.txt when we require a build with metering of radosgw enabled?16:53
gordcabhi-: the general approach we've taken (although probably not always followed) is to have driver related libs part of test-reqs only.16:54
gordcif ops/deployers want to actually use said driver, than the library is loaded (manually)16:55
abhi-ah ok16:55
*** ildikov has quit IRC16:55
*** rwsu has joined #openstack-ceilometer17:05
swamireddygordc: ceilometer-acentral is working with the patch...(without the patch it failed with   -- ceilometer ImportError: No module named awsauth)17:17
*** _cjones_ has joined #openstack-ceilometer17:18
swamireddygordc: For this testing I have removed all requests-aws pkg (pip and dpkg pkgs both removed)...and performed the testing.. ceilometer-acentral working (not seen the ImportError)17:19
* cdent goes to see waves17:20
swamireddygordc: Shall I re-submit the patch with same commit id? or it should be separate and new patch?17:21
gordcswamireddy: cool stuff. i think technically i can revert my revert (once it merges)17:21
gordci'm not sure you can recommit with same id.17:21
gordcyou can try ... i feel like git will yell at you though.17:21
swamireddygordc: Yep, git will allow the reverts17:22
gordcmaybe when my revert merges you can just revert it again.17:22
gordcwhatever works.17:22
swamireddygordc:  Both should not revert and I try to revert may cause the problem...17:23
gordcoh, yeah i'm not doing anything... my suggestion was that when this merges in 30mins (
gordcyou can hit the revert and it will technically revert the revert... making it a positive again?17:24
gordci don't really have a preference.17:25
*** ddieterly has joined #openstack-ceilometer17:28
*** himangi has quit IRC17:31
gordcswamireddy: i created this:
openstackLaunchpad bug 1424746 in Ceilometer "central agent will throw error on startup" [Medium,Triaged]17:32
gordcwith dhellmann patch your code with work, it'll just show an error at startup which might be nice to hide... don't know if you want to fix it on your end.17:32
swamireddygordc: Cool.. I can submit the patch with this bug id...17:33
swamireddygordc: fix in rgw_clint code??17:34
gordcswamireddy: i'm actually not sure where the proper fix is. i just noticed the error when i looked at the logs:
gordcbut yeah, you'd probably need to fix it in client...17:36
*** zhenq has quit IRC17:37
gordcand now that i look at error, another side issue that might exist is that we're creating a client for each pollster (we might want to share it. ie make it class level client)17:37
gordcbut one fix at a time is cool17:37
gordcor both if easy.17:37
swamireddygordc: Thats a reason Abhi was asking to include the requests-aws in requirements.txt..17:37
gordcwhich reason? the error?17:38
*** amalagon_ has joined #openstack-ceilometer17:38
swamireddygordc: errors are comming in as Failed to import extension for rgw.usage: .. even I the acentral issue fixed..17:39
gordcwell you can do a try: catch: on import awsauth and raise it as something less aggressive.17:40
swamireddygordc: I could think it as - some where in conf, rgw is True, then only use the rgw polling...17:40
gordci think dhellmann's fix is more to avoid breaking agent completely17:40
gordcswamireddy: yes, that could work as well17:41
swamireddygordc: Not sure, if ceilometer has this type of support ATM...17:41
gordcwhat support?17:41
swamireddygordc: Yes, agree. Dhallmann´s fix to avoid the breaking agent..17:42
swamireddygordc: If user don´t need the swift or ceph polling, that can be turned-off at conf file level itself...17:42
swamireddygordc:  This gives the top level nob to on or off..17:43
*** Ala has quit IRC17:45
gordchmm.. yeah you'd probably need to code in support for that.17:45
gordci'm ok with the bug being address separately to be honest.17:45
gordcbbl getting called away from computer.17:45
*** exploreshaifali has quit IRC17:45
*** zul has quit IRC17:51
linuxhermitcdent: should I consider moving on to another part of gabbi testing?17:51
linuxhermitprior to that existing on getting merged, or wait a few until more discussion has occurred (which is the way I'm learning)17:52
*** zul has joined #openstack-ceilometer17:53
cdentlinuxhermit: feel free to move on or if you want another thing I tend to do is just create a yaml file for the sake of finding bugs, but not to actually commit anything, and then report the bugs17:57
*** ccrouch has joined #openstack-ceilometer17:57
linuxhermithmm okay17:58
cdentIt can be an interesting diversion to learn stuff.18:02
*** ybathia has joined #openstack-ceilometer18:02
*** ildikov has joined #openstack-ceilometer18:07
*** eglynn-office is now known as eglynn-officeafk18:09
*** david-lyle_afk is now known as david-lyle18:09
*** _cjones_ has quit IRC18:10
*** zhenq has joined #openstack-ceilometer18:12
*** eglynn has quit IRC18:13
*** _cjones_ has joined #openstack-ceilometer18:15
*** zul has quit IRC18:25
*** zul has joined #openstack-ceilometer18:25
*** sbfox has joined #openstack-ceilometer18:26
swamireddygordc: I plan to hold my patch and see if I can fix the import  issue also.. (like Failed to import extension for rgw.usage: No module named awsauth)..18:31
swamireddygordc:  I see the same issue might have  been there for other driver libs, not sure how they have fixed it...18:32
*** harlowja has joined #openstack-ceilometer18:33
*** zul has quit IRC18:42
*** ddieterly has quit IRC18:43
*** ddieterly has joined #openstack-ceilometer18:44
*** zul has joined #openstack-ceilometer18:44
*** abhi- has quit IRC18:51
*** zul has quit IRC19:04
*** zul has joined #openstack-ceilometer19:04
cdentgordc: I was surprised too19:08
cdentIn fact I think when I first wrote the code you've just changed I made the same mistake and eglynn-officeafk caught it.19:09
gordccdent: i could've sworn i've tried this before... probably some 'everything is an object' bs19:09
cdentIt's okay with non reference types19:09
cdentx = y = 119:09
cdentbut x = y = [] or x = y = {} will get funky19:09
*** ddieterly has quit IRC19:10
cdentit's the same issues as def foo(x=[])19:10
gordccdent: ah right.19:10
*** ddieterly has joined #openstack-ceilometer19:10
* cdent seeks food19:11
*** zul has quit IRC19:11
*** zul has joined #openstack-ceilometer19:11
*** cdent has quit IRC19:14
*** eglynn has joined #openstack-ceilometer19:15
*** alexpilotti has quit IRC19:24
gordceglynn: i don't know if it's worth blocking this:
gordclooking at dhellmann's patch, it seems like it'll resolve the broken client issue.19:25
eglynngordc: it shouldn't be a problem thought to simply resubmit swamireddy's patch?19:25
gordceglynn: yeah, just asking if we want to avoid a "revert a revert" patch.19:26
gordci'm just slowly watching it move up the merge list since i don't really have an opinion on it.19:26
eglynngordc: TBH I'm not 100% sure whether simply revoking the +2/+A will stop it in its tracks19:27
gordceglynn: true. i'm pretty sure i've done it before but it's so rare that i'm not really sure to be honest.19:28
gordci guess that answers the question19:28
gordclet it be.19:28
eglynnyeap, think so19:29
dhellmanneglynn, gordc : you could submit a change to the patch to kick it out of the queue (tweak the commit message in gerrit) but I think ultimately you do want the revert to land and for a new patch to be submitted that deals with the import error explicitly. My patch is just meant to prevent things from breaking when badly behaved plugins are present.19:39
eglynndhellmann: yes, that makes sense, thanks!19:39
*** zul has quit IRC19:48
*** openstackgerrit has quit IRC19:51
*** openstackgerrit has joined #openstack-ceilometer19:52
openstackgerritMerged openstack/ceilometer: Revert "Add ceph object storage meters"
*** zul has joined #openstack-ceilometer20:01
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements
*** drjones has joined #openstack-ceilometer20:10
*** _cjones_ has quit IRC20:10
openstackgerritgordon chung proposed openstack/ceilometer: fix inline multiple assignment
*** promulo has quit IRC20:19
*** eglynn has quit IRC20:23
*** ddieterly has quit IRC20:32
*** ddieterly has joined #openstack-ceilometer20:33
*** harlowja has quit IRC20:34
*** harlowja_ has joined #openstack-ceilometer20:34
*** promulo has joined #openstack-ceilometer20:34
*** ddieterly has quit IRC20:37
*** exploreshaifali has joined #openstack-ceilometer20:38
*** ddieterly has joined #openstack-ceilometer20:38
*** zul has quit IRC20:40
*** zul has joined #openstack-ceilometer20:40
*** ddieterly has quit IRC20:46
*** thomasem has quit IRC20:47
ccrouchso who has got a summit session proposal that I can vote for?21:00
openstackgerritMerged openstack/ceilometer: Protect agent startup from import errors in plugins
*** ybathia has quit IRC21:04
*** ddieterly has joined #openstack-ceilometer21:14
*** thomasem has joined #openstack-ceilometer21:20
*** amalagon_ has quit IRC21:22
*** amalagon has joined #openstack-ceilometer21:23
*** amalagon has quit IRC21:27
*** amalagon has joined #openstack-ceilometer21:41
*** sbfox has quit IRC21:41
*** zhenq has quit IRC21:45
*** karimb has joined #openstack-ceilometer21:45
*** karimb has quit IRC21:48
*** ybathia has joined #openstack-ceilometer21:49
*** exploreshaifali has quit IRC21:51
*** yeungp has quit IRC21:56
*** yeungp has joined #openstack-ceilometer21:56
*** yeungp has quit IRC21:57
*** yeungp has joined #openstack-ceilometer21:57
*** sbfox has joined #openstack-ceilometer21:57
*** yeungp has quit IRC21:58
*** yeungp has joined #openstack-ceilometer21:58
*** ddieterly has quit IRC22:01
*** ddieterly has joined #openstack-ceilometer22:03
gordcccrouch: see last few minutes of meeting22:05
*** nellysmitt has quit IRC22:16
*** nellysmitt has joined #openstack-ceilometer22:16
*** nellysmitt has quit IRC22:17
*** zhenq has joined #openstack-ceilometer22:19
*** harlowja_ has quit IRC22:29
*** harlowja has joined #openstack-ceilometer22:30
ccrouchgordc: perfect, thanks22:34
ccrouchbuttons pushed!22:34
gordcccrouch: :)22:34
ccrouchi think there is like 25mins left?22:34
ccrouchfor all you late voters out there?22:34
*** yeungp has quit IRC22:37
*** yeungp has joined #openstack-ceilometer22:37
gordcccrouch: any good talks to promote?22:39
ccrouchi'm only telling because you asked :-)22:42
ccrouchit will hopefully be good :-) Its got good potential :-)22:43
*** yeungp has quit IRC22:44
gordcccrouch: :) i appreciate the undersell.22:45
gordci'm sure you'll get a bunch of interest if it's containers/kolla related.22:46
ccrouchi fear we are at "peak docker" in the hype cycle22:47
ccrouchpeople are getting a little inured to the topic22:48
*** zhenq has quit IRC22:48
gordcno time for research. just give me keywords.22:49
ccrouchgordc: I recommend "doomed"22:53
ccrouchas a good keyword22:53
*** thomasem has quit IRC22:54
gordci'm pretty sure i've promoted termie's talk more than my own.22:55
ccrouchhow can you *not* vote for a talk like22:55
ccrouchit should be keynote :-)22:55
gordci wish. :)22:56
*** thomasem has joined #openstack-ceilometer23:03
*** thomasem has quit IRC23:03
*** harlowja has quit IRC23:04
*** rwsu is now known as rwsu-afk23:04
*** thomasem has joined #openstack-ceilometer23:04
*** devlaps has quit IRC23:14
*** devlaps has joined #openstack-ceilometer23:14
*** harlowja has joined #openstack-ceilometer23:16
*** promulo__ has joined #openstack-ceilometer23:30
*** promulo has quit IRC23:31
*** prad has quit IRC23:36
*** alexpilotti has joined #openstack-ceilometer23:36
*** gordc has quit IRC23:37
*** ddieterly has quit IRC23:50

Generated by 2.14.0 by Marius Gedminas - find it at!