Wednesday, 2015-09-09

« Tuesday, 2015-09-08 Index Thursday, 2015-09-10 »
*** sigmavirus24_awa has quit IRC00:49
*** sigmavirus24_awa has joined #openstack-searchlight00:57
*** ekarlso- has joined #openstack-searchlight05:37
*** ekarlso- has quit IRC05:50
*** sballe has quit IRC06:46
*** lakshmiS has joined #openstack-searchlight08:28
*** TravT_ has joined #openstack-searchlight09:19
*** TravT has quit IRC09:20
ekarlsoanyone around ?10:16
*** sballe has joined #openstack-searchlight11:22
openstackgerritEndre Karlson proposed openstack/searchlight: Add Designate plugin.  https://review.openstack.org/19909912:10
lakshmiShey ekarlso:12:30
*** lakshmiS has quit IRC12:35
*** lakshmiS has joined #openstack-searchlight12:43
*** lakshmiS has quit IRC12:52
sjmc7_ello14:02
rosmaitao/14:03
TravT_hello guys14:03
TravT_Courtesy Searchlight review bash: lakshmiS, nikhil_k, rosmaita, sigmavirus24, TravT, david-lyle, kragniz, sjmc7, ekarlso14:04
*** TravT_ is now known as TravT14:04
ekarlsoTravT_: meh, allmost got the designate stuff sorted14:04
ekarlsoputting up a new review later14:04
TravTekarlso:14:04
TravTok14:04
TravTCurrent reviews: http://tinyurl.com/searchlight-reviews14:04
TravTFirst one to look at: https://review.openstack.org/206268 Add 'sort' parameter14:05
TravTThen: https://review.openstack.org/210759 Allow indexing of single type or index14:05
sjmc7_be cruel to me!14:09
TravTsjmc7_  i'm trying to track down lakshmiS14:10
sjmc7_to be cruel to him too14:10
TravTsee which of his we should review...14:10
TravTwas just looking at this one: https://review.openstack.org/#/c/22014914:10
sjmc7_*shudder*14:10
rosmaitaTravT: i just +2d the sorting patch14:13
TravTrosmaita: cool.14:13
TravTI'll add a +A if nobody posts any objections by the end of this meeting.14:14
rosmaitathat's what i was going to ask14:14
TravTseems to a be a light crowd today. :S14:14
TravTrosmaita: if you can see if the CLI (that second patch) looks good to you, that would be great.14:15
rosmaitawas just looking14:15
rosmaitayou have been pretty tough on that patch!14:16
TravTyeah, sjmc7_ probably hates me14:16
rosmaitai thought it looked good earlier and then you requested like 10 changes14:16
rosmaitai am almost afraid to say it's ok14:17
sjmc7_he's very demanding14:18
TravTwell, i view it as a user interface, so thought i'd be a bit harder on it14:18
sjmc7_:)  it was all good feedback14:18
TravTsjmc7_  the one comment I had yesterday which was poorly worded and you disagreed with i'm okay with possibly improving upon later.14:19
TravTotherwise, i think it works pretty well.14:20
sjmc7_ok. i was having some senior moments yesterday so the misunderstanding may have been mine14:20
TravTand i have to say that the more plugins we get, the more it seems critical.14:20
TravTthat we get that in14:20
TravTrosmaita: nikhil_k_: sigmavirus24_awa: there is a glance bug in glance that would be highly helpful for you to look at14:23
TravThttps://review.openstack.org/#/c/221307/14:23
rosmaitais that the notifications?14:23
sjmc7_yeah. oh, something else occurred - how do other projects handle the log_file config option problem? i.e. nova-api and nova-conductor share a config file14:23
*** sigmavirus24_awa is now known as sigmavirus2414:23
TravTyeah14:23
sjmc7_because we have this now; all the logs end up in the same file14:23
rosmaitai was looking last night, there was a prob with the python 2.7 tests14:23
rosmaitasjmc7_: in the manage patch, do you redirect print to stderr?  not that you have to, just wondering14:24
TravTyeah, i see there is a python 2.7 failure... but if there were any other glaring issues, lakshmiS could address as well14:24
sjmc7_rosmaita - i do not. is that the Done Thing?14:24
rosmaitanot sure14:24
sjmc7_since it's going to be interactive (or ansibled or whatever) it didn't seem too huge a deal14:24
rosmaita(that's why i asked)14:24
TravTsjmc7_: are you talking about how the CLI spits out oslo config message?14:25
*** sjmc7_ is now known as sjmc714:25
sjmc7ah, that's better, no underscore14:25
*** TravT is now known as TravT_underscore14:25
sjmc7TravT_ - no.. if you set up searchlight-api and searchlight-listener to use the same config file with log_file specified14:26
* TravT_underscore just wanted to be different14:26
sjmc7they write to the same file14:26
*** TravT_underscore is now known as TravT14:26
sjmc7that sounded more remedial than i meant it14:26
sigmavirus24sjmc7: yeah that's kind of expected, no?14:26
sjmc7yes, it's what you'd expect, but maybe not what you want14:26
sigmavirus24we can have separate config sections for both of those commands so we don't need separate config files14:26
sigmavirus24e..g., [listener] log_file = searchlight-listener.log14:27
sjmc7yeah, that's what i was asking14:27
sjmc7thank you for saying it in words people understand :)14:27
sigmavirus24I mean having the many files like glance does is okay14:27
sjmc7yeah, but it seems like overkill right now14:27
sigmavirus24I just don't know if I like that approach yet as someone who manages ansible templates for all of them14:27
sigmavirus24sjmc7: agreed14:27
sjmc7me neither14:27
sjmc7as someone who maintains ansible templates :)14:27
sigmavirus24lol14:27
sigmavirus24why don't you use openstack-ansible? =P14:27
sjmc7no comment :)14:28
sigmavirus24then I'd manage them for you14:28
sigmavirus24You should at least hang out in #openstack-ansible14:28
TravTi'm all for keeping a single config file for as long as it makes sense14:28
sigmavirus24TravT: yep14:28
sigmavirus24giving different sections for the different commands is good14:28
sjmc7sigmavirus24 - i'm on three internal company chat channels on the topic, i can't bring myself to add more :)14:29
sigmavirus24we can have a default log_file taht's overridden in each section if oslo.cfg supports that stuff14:29
sigmavirus24I forget14:29
rosmaitagot another dumb question ... when a script aborts, is a 0 an appropriate exit code? it's aborting because the user didn't type 'y', but still ...14:29
sigmavirus24rosmaita: depends14:29
sigmavirus24rosmaita: is the user deliberately not doing something?14:29
rosmaitayes14:29
sigmavirus24yeah14:29
sigmavirus24You didn't error out14:29
sigmavirus24So 0 should be fine14:29
rosmaitaok14:29
sjmc7right - it's not an error. although it might be a fat finger14:29
sigmavirus24There may be a more nuianced understanding14:29
* sigmavirus24 can't spell nuianaclvkajaew;lkbjadsf14:29
TravTi think if user said to stop, that non-zero is okay14:30
TravTsigmavirus24: I have trouble with that word too.14:30
TravTnuianaclvkajaew;lkbjadsf14:30
TravTis so hard14:30
TravTi always forget the semi-colon in the middle14:30
sigmavirus24me too14:31
sjmc7i before anclkd except after hggjkt14:31
sjmc7i think 0's ok - it's a conscious decision to exit14:31
sjmc7if we have a separate sections, won't oslo.config still be looking in the DEFAULT section?14:32
sjmc7for the logging?14:32
sjmc7i do enjoy your review comments, rosmaita :)14:32
rosmaitai try14:33
TravTsigmavirus24: not sure if you saw https://review.openstack.org/#/c/221307/14:33
TravTbut we need that for image rbac in searchlight14:33
sigmavirus24TravT: I did see it and rechecked it14:33
sigmavirus24Failing after 7 minutes for that glance job seems ... wrong14:34
sigmavirus24sjmc7: we can look up in listener and if that's not there, fall back to the general/default option14:34
sjmc7ok sigmavirus24. i need to refresh how we initialize that, maybe we do consciously choose the config option14:34
TravTsigmavirus24: rosmaita: lakshmiS has a patch up that requires glance client 1.014:35
TravThttps://review.openstack.org/#/c/219168/14:35
TravTis this something that we should do?14:35
* rosmaita looking14:35
* rosmaita is wondering if that's my fault, i think i did the image-members glanceclient code14:36
sjmc7ah, yes - do we need to bump our requirements to g-c 1.0.0?14:36
sjmc7or cap it <1.0.0 ?14:37
TravTi also am not sure why is always fails to verify with no message...14:37
rosmaitawell, there are going to be a lot of bugfixes coming14:37
rosmaitadon't think people will want to use 1.0.0 for very long14:37
*** lakshmiS has joined #openstack-searchlight14:38
* rosmaita is pretty sure i didn't expose 'deleted' through the client14:39
TravTlakshmiS o/14:40
TravTjust talking about https://review.openstack.org/#/c/219168/14:40
TravTand glance client 1.014:40
lakshmiSok14:40
TravTit isn't clear to me if that's going to be okay to use, yet or not.14:41
lakshmiSwhat do you mean?14:41
TravTwell, has the rest of openstack adopted it in liberty yet?  will that be a problem?14:41
lakshmiShmm.14:41
lakshmiSin that case we keep it parked14:42
TravTlakshmiS: rosmaita: do we need dual path code based on glance client version?14:43
TravTor am i just being paranoid here14:43
TravT?14:43
TravTlakshmiS: does it not work at all?14:43
lakshmiSnot with client ver 1.0.014:43
TravTso if we have 0.18, it works?14:44
TravTtoday14:44
TravT?14:44
lakshmiShave to check what was my previous version on devstack14:44
lakshmiSTravT: your idea on checking version is not bad14:45
TravTwe do that in horizon.14:46
TravTbut it is api version that is checked14:46
TravTnot the release version check of the package14:46
TravTbut, maybe we need to do that here?14:47
lakshmiSyes14:47
TravTok.14:47
TravTeverybody, as FYI, I have to leave in about 5 minutes.  I tweaked my knee a few weeks back and have a dr apt at the top of the hour.14:48
TravTlakshmiS: does this on have any dependency on glance client version?14:49
TravThttps://review.openstack.org/#/c/220149/14:49
lakshmiSno14:49
TravTok14:50
TravTthen i can try it out when i get back.14:50
TravTok, i do need to drop now.  thanks for some review time.14:53
rosmaitanp, hope your knee is ok14:53
TravTi have a couple patches that i'm going to put up later today, i hope.  but also will go back over the designate patch that i've been harassing ekarlso on14:53
TravTthx14:53
*** lakshmiS has quit IRC15:43
ekarlsosigh16:02
ekarlsomor stuff to code :p16:02
TravTekarlso: re designate patch?16:06
openstackgerritBrian Rosmaita proposed openstack/searchlight: Clarify glance api restart.  https://review.openstack.org/22186016:06
openstackgerritMerged openstack/searchlight: Add 'sort' parameter  https://review.openstack.org/20626816:20
openstackgerritBrian Rosmaita proposed openstack/searchlight: Corrected 'Openstack' references  https://review.openstack.org/22186816:21
ekarlsoso TravT16:34
ekarlsohow do I map parent child ?16:34
TravTekarlso, i put what I think you needed to do in those comments... l16:35
ekarlsokk, i'll try16:35
TravTbut I didn't actually try it out with the comments I gave you.  about half the time I actually code things up before leaving comments.16:36
TravTsee comment on line 40 here:16:37
TravThttps://review.openstack.org/#/c/199099/17/searchlight/elasticsearch/plugins/designate/recordsets.py16:37
sjmc7is this for the relationship inside recordsets?16:37
ekarlsoya sjmc716:37
openstackgerritMerged openstack/searchlight: Clarify glance api restart.  https://review.openstack.org/22186016:37
sjmc7and one Zone always belongs to one Recordset?16:38
TravTsjmc7: that's what ekarlso indicated in an IRC yesterday.16:39
TravT1 zone --> many recordsets16:39
openstackgerritMerged openstack/searchlight: Corrected 'Openstack' references  https://review.openstack.org/22186816:39
sjmc7errr.. yes, sorry16:40
TravTalthough cardinality may be better expressed like this16:40
TravT1 zone <-- many recordsets16:40
sjmc7parent-child is pretty new (newer at least than last time i used e-s in anger) but it seems like a good fit here16:41
TravTyeah, i was thinking in the recordset16:41
TravTwe could add this:16:41
TravT"_parent": {16:41
TravT        "type": “OS::Designate::Zone”16:41
TravT      }16:41
TravTand then in my other comments there16:41
TravTit seems pretty easy to pass the parent ID through in both notifications and initial indexing16:42
ekarlsoTravT: not using a mapping ?16:42
ekarlsosorry i'm very vague when it comes to ES16:42
TravTso, in the recordsets mapping16:43
sjmc7just checking the python-es library support for it16:43
TravTi checked that yesterday16:43
TravTthat's what those comments here are based on: https://review.openstack.org/#/c/199099/17/searchlight/elasticsearch/plugins/designate/recordsets.py16:43
sjmc7i know, just for my own edification :)  yeah, that seems sensible16:43
TravTbut good to have your view too16:43
openstackgerritBrian Rosmaita proposed openstack/searchlight: Corrected RST markup  https://review.openstack.org/22187516:43
sjmc7ekarlso - yes, in the mapping you'd define the _parent16:43
TravTekarlso: in the mapping, you define in the child who the parent is16:44
TravTthe each time you index a child document you have to specify the id of the parent16:44
sjmc7and the recordset indexing would need to add the parent id as travis indicated in the comments16:44
TravTsince we use the OS ID's for the doc ID and you already have the zone_id, you can pass that in16:44
TravTas the parent id16:44
TravTfor initial indexing and in notification indexing16:44
sjmc7one hting to watch for - deleting a parent does NOT delete all children16:45
ekarlsobut uh16:45
sjmc7needs a delete_by_query16:45
ekarlsoTravT: in the examples in "mapping": {} has the diff types it seems and then the properties inside the type?16:45
TravTlink?16:46
ekarlsobased on hat https://www.elastic.co/guide/en/elasticsearch/guide/current/parent-child-mapping.html16:46
sjmc7ekarlso - this would be oustide 'properties'16:46
ekarlsooh, nvm16:46
sjmc7{"recordset": {"properties": {...}, "_parent": {"type"...16:46
ekarlsouh?16:47
ekarlsoaint it in the mappings section ?16:47
sjmc7yes16:47
sjmc7at the top level in the dict that get_mappings returns16:47
ekarlsohttp://paste.ubuntu.com/12321965/ < at that ?16:47
sjmc7err... no16:48
sjmc7the "mappings" is done for you by searchlight16:48
ekarlsoahh16:48
sjmc7so it's {"mappings": {"recordset": get_mapping()16:48
sjmc7https://github.com/openstack/searchlight/blob/master/searchlight/elasticsearch/plugins/base.py#L5416:49
ekarlso?should w16:49
TravTekarlso i think it is this (gimme sec to make paste)16:49
ekarlso http://paste.ubuntu.com/12321987/ sjmc7 ?16:50
sjmc7ekarlso - nope :)16:50
TravTyou don't need the zone wrapping that16:50
ekarlsorofl :p16:51
TravTand it is sibling to properties16:51
sjmc7at the same level as "properties" goes the "_parent" bit16:51
sjmc7you're thinking too SQL :)16:51
ekarlsosjmc7: http://paste.ubuntu.com/12322004/ ..16:52
TravTyes16:53
sjmc7there you go16:53
TravTso, yesterday i was also playing with nesting these.16:54
ekarlsothen just pass parent=... in engine.index() ?16:54
sjmc7right16:54
sjmc7just the zone id16:54
TravTyes, using the zone_id16:54
sjmc7i'll stop talking now :)16:54
TravTsjmc7:16:54
TravTi was looking at nesting16:54
TravTbut based on what ekarlso said that there could thousands of recordsets16:55
TravTupdating individually, it seemed that maybe parent child made more sense16:55
sjmc7yeah, nesting's not really for this16:55
TravTES docs say nested queries are still faster than parent child16:55
TravT"but they are working on that"16:55
TravT:S16:55
sjmc7it's all relative :)16:55
sjmc7i need to look at what kind of joining is possible16:56
TravThttps://www.elastic.co/guide/en/elasticsearch/guide/current/has-child.html16:56
ekarlsoso based on that ..16:56
TravThttps://www.elastic.co/guide/en/elasticsearch/guide/current/has-parent.html16:56
ekarlso2015-09-09 16:56:07.707 ERROR searchlight.cmd.manage [-] Failed to setup index extension recordset: ('25 document(s) failed to index.', [{u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'4bfed609-f762-45d6-932b-716d08d9fd25', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:56
sjmc7there you go :)16:56
ekarlso[searchlight]/[OS::Designate::RecordSet]/[4bfed609-f762-45d6-932b-716d08d9fd25]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'bf4d86c1-4b4b-4e01-b170-5dcf2d32492d', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[bf4d86c1-4b4b-4e01-b170-5dcf2d32492d]]'}}, {u'index': {u'status':16:56
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'994595dc-852a-4ca2-868a-c24e541fe095', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[994595dc-852a-4ca2-868a-c24e541fe095]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'7ae46726-4dcb-4e1e-97ed-ff7215b2ae85', u'_index':16:56
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[7ae46726-4dcb-4e1e-97ed-ff7215b2ae85]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'a5f11a2a-93c3-4106-82e9-09ccc2b84486', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:56
ekarlso[searchlight]/[OS::Designate::RecordSet]/[a5f11a2a-93c3-4106-82e9-09ccc2b84486]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'57314230-b8ea-48d7-afcb-709f91c0f859', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[57314230-b8ea-48d7-afcb-709f91c0f859]]'}}, {u'index': {u'status':16:56
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'18eb658d-85e1-4064-9c20-8b5ee125917f', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[18eb658d-85e1-4064-9c20-8b5ee125917f]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'27f0f416-8c59-4c33-9e1d-9b9216a6a4a2', u'_index':16:56
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[27f0f416-8c59-4c33-9e1d-9b9216a6a4a2]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'51bb6541-495a-4469-a533-54ba8488ddc8', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:56
ekarlso[searchlight]/[OS::Designate::RecordSet]/[51bb6541-495a-4469-a533-54ba8488ddc8]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'791d80de-cddb-4b2e-8816-6b9e9ad5db5a', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[791d80de-cddb-4b2e-8816-6b9e9ad5db5a]]'}}, {u'index': {u'status':16:57
TravTaugh16:57
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'7aae70d1-cee7-4998-98a2-acaa7b48afaf', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[7aae70d1-cee7-4998-98a2-acaa7b48afaf]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'8a907d6d-e8c5-45c1-bb69-6725685d8d7a', u'_index':16:57
sjmc7hahahahaha16:57
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[8a907d6d-e8c5-45c1-bb69-6725685d8d7a]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'aaa9883f-3bb4-44bb-9ed2-2db6b5dfce9c', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:57
ekarlso[searchlight]/[OS::Designate::RecordSet]/[aaa9883f-3bb4-44bb-9ed2-2db6b5dfce9c]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'cf8edf10-2c2e-4d4e-a084-58600e29a0c1', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[cf8edf10-2c2e-4d4e-a084-58600e29a0c1]]'}}, {u'index': {u'status':16:57
sjmc7stop, man!16:57
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'cfe40861-cdb5-4abc-b67b-bdcd2e66b40f', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[cfe40861-cdb5-4abc-b67b-bdcd2e66b40f]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'd8491fb8-4fa4-4b7b-9a25-51562068f3cb', u'_index':16:57
TravTmake it stop!16:57
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[d8491fb8-4fa4-4b7b-9a25-51562068f3cb]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'eded1c15-392e-4375-a577-aa783f078792', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:57
ekarlso[searchlight]/[OS::Designate::RecordSet]/[eded1c15-392e-4375-a577-aa783f078792]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'eeedd854-8082-4e81-8e5c-30a8fe216068', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[eeedd854-8082-4e81-8e5c-30a8fe216068]]'}}, {u'index': {u'status':16:57
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'fbaa6f73-6390-4da1-9819-3b908a18a123', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[fbaa6f73-6390-4da1-9819-3b908a18a123]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'77d5357b-d742-4fc4-8748-5dfd3646d5eb', u'_index':16:57
sjmc7the goggles do nothing!16:57
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[77d5357b-d742-4fc4-8748-5dfd3646d5eb]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'fe050b20-67ac-434b-8350-df4f35d6782f', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:57
ekarlso[searchlight]/[OS::Designate::RecordSet]/[fe050b20-67ac-434b-8350-df4f35d6782f]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'f39f3c99-62d3-43cb-81a7-7deaed9b8fa9', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[f39f3c99-62d3-43cb-81a7-7deaed9b8fa9]]'}}, {u'index': {u'status':16:57
ekarlso400, u'_type': u'OS::Designate::RecordSet', u'_id': u'f9feb535-ccdc-465a-8002-6b94b548330f', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[f9feb535-ccdc-465a-8002-6b94b548330f]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'6bcb0730-af8f-4b3f-b949-40a6611bc41b', u'_index':16:57
ekarlsou'searchlight', u'error': u'RoutingMissingException[routing is required for [searchlight]/[OS::Designate::RecordSet]/[6bcb0730-af8f-4b3f-b949-40a6611bc41b]]'}}, {u'index': {u'status': 400, u'_type': u'OS::Designate::RecordSet', u'_id': u'537610b9-65f5-4ada-9b9a-848087d12c65', u'_index': u'searchlight', u'error': u'RoutingMissingException[routing is required for16:57
TravTi'm getting motion sick16:57
ekarlso[searchlight]/[OS::Designate::RecordSet]/[537610b9-65f5-4ada-9b9a-848087d12c65]]'}}])16:57
ekarlsooh feck16:57
ekarlsolol16:57
ekarlsodidn't expect it to be that big :P16:57
ekarlsowha tthe crap does that mean? : p16:57
TravTi think you should stop tailing logs to irc16:57
sjmc7i wish someone's irc handle was 'designate'16:57
ekarlso:p16:57
ekarlso+1 TravT :p16:58
TravTekarlos16:58
TravTekarlso: you might need to delete your mappings manually16:58
TravTand try again16:58
ekarlsoyeah, I did http DELETE http://localhost:9200/_all16:58
ekarlsothen sync16:58
TravThmm, ok16:58
ekarlsobut I wonder if the mapping is wrong ?16:59
ekarlsoohmm17:00
ekarlsoin the indexing do I need to add like a _parent thing in the payload ?17:00
TravTyes17:00
sjmc7no, just a parent_id in the indexing. the parent does have to be indexed already though17:00
TravTFor initial indexing, I am thinking that we could add _parent field to the objects returned from get_objects().17:01
TravTsearchlight/elasticsearch/plugins/base.py17:01
TravTUses the bulk API, which means we should be able to just add that field and it will work correctly.17:01
TravThttps://www.elastic.co/guide/en/elasticsearch/reference/current/docs-bulk.html#bulk-parent17:01
TravTplease note the *should*17:01
ekarlsooh no, the kid woke up..17:01
ekarlsowhy can't he ever sleep when expected : p17:02
sjmc7:D17:02
sjmc7first one, i will presume :)17:02
ekarlsomeh, yeh :p17:02
TravTekarlso: i usually put the kids to bed and then somehow find myself back to the computer screen17:03
TravTseems you must do similar17:03
ekarlsoTravT: :P17:13
ekarlsoTravT: any idea on the ^ boom ?17:13
TravThey, ekarlso, did you add the _payload like I mentioned above?17:14
ekarlso_parent you mean to _payload ?17:14
TravTyes sorry17:16
ekarlsoyeah, in recordset I did17:16
ekarlsoit should be _parent: <zone id> no ?17:16
TravTi *think* so.17:16
sjmc7ekarlso - i will have a little play now17:17
ekarlsomeh :p17:17
ekarlsoI can put up the latest changeset for ya sjmc717:17
ekarlsoso u can see17:17
sjmc7ok17:17
openstackgerritEndre Karlson proposed openstack/searchlight: Add Designate plugin.  https://review.openstack.org/19909917:19
ekarlsodone17:20
sjmc7ok. i need to play with parent-child a bit, but i have a suspicion the ordering will be important17:20
ekarlso:/17:21
TravTmmm17:21
sjmc7and i think the reason you get a routing error is that it's trying to figure out which shard the parent lives on17:21
sjmc7but that's a wild semi-educated guess17:21
TravTi'm wondering if zone has already been populated17:21
TravTare recordsets getting populated prior to zones when you hit this error?17:22
*** TravT_ has joined #openstack-searchlight17:34
*** TravT has quit IRC17:35
*** TravT_ is now known as TravT17:35
openstackgerritMerged openstack/searchlight: Allow indexing of single type or index  https://review.openstack.org/21075917:46
openstackgerritMerged openstack/searchlight: Corrected RST markup  https://review.openstack.org/22187517:50
sjmc7TravT and ekarlso... one other thing; reindexing the parent may not maintain the parent/child relationship17:55
TravTdid i miss some other conversation?17:57
TravTmy internet got really flaky the last 40 mins17:57
sjmc7no, but just reading more on this - https://www.elastic.co/guide/en/elasticsearch/guide/current/indexing-parent-child.html17:57
sjmc7because of the strnage routing that's required, you have to provide the parent id on all single-document requests (i.e. for a recordset) so that e-s can find it on the right shard17:57
sjmc7reindexing the parent *should* be ok because the id doesn't change17:58
TravTjust read it...17:59
sjmc7ekarlso, i can reproduce the error you're getting17:59
sjmc7curl 192.168.235.128:9200/test/child/117:59
TravTwe aren't trying to change parents17:59
TravTso, re-index shouldn't be a problem.17:59
sjmc7raises the error because e-s knows it needs to be told where to index17:59
sjmc7TravT no, but we do reindex parent documents18:00
sjmc7that's going to be ok because the id won't change, so red herring18:00
sjmc7this is ok: curl 192.168.235.128:9200/test/child/1?parent=AU-zMCb316T5c0I1tBXU18:00
sjmc7so i think the indexing request was missing the parent id18:00
sjmc7having read it, the order's unimportant; e-s doesn't care if the parent actually exists18:01
sjmc7we need a patch to plugins/base.py to support bulk indexing documents with parents i think, TravT/ekarlso. i'll get something up shortly18:08
TravTso, then what I'm thinking is that we could consider landing designate without the parent child relationship18:09
TravTand add that in a follow on patch18:09
sjmc7also an option18:10
TravTjust need to reverify the rbac.18:10
TravTlooks like ekarlso made changes based on my comments there18:10
sjmc7yep18:11
TravTmy brain needs some food.18:11
TravThttps://teszmillan.files.wordpress.com/2011/11/brain_drain.jpg18:11
TravTso, i'm gonna go eat some lunch and be back shortly.18:12
openstackgerritSteve McLellan proposed openstack/searchlight: Add '_parent' to bulk indexing if needed  https://review.openstack.org/22190318:31
sjmc7ekarlso - try rebasing on top of that ^18:31
sjmc7and implementing get_parent_id_field to return 'zone_id'18:39
ekarlsouh sjmc720:21
ekarlsosjmc7: does that work on notifications too ?20:21
sjmc7notifications you're idnexing one by one20:21
sjmc7so the problem isn't there; you're setting parent=zone_id in the index call20:22
ekarlsoblows up still sjmc720:25
ekarlsosame error actually20:25
sjmc7GRRR!20:26
sjmc7ok, i'll take another look20:26
ekarlso:d20:29
sjmc7ekarlso, did you implement get_parent_id_field in the rec ordset plugin?20:43
ekarlsosjmc7: ya!20:43
sjmc7i'd prefer it if you hadn't :)20:44
ekarlsouh ?:p20:44
sjmc7then there'd be a good reason it didn't work :)20:44
sjmc7i'll take a look in a sec, just got designate running in devstack again20:44
ekarlso:p20:47
sjmc7ekarlso, worked for me after adding get_parent_id_field to recordset.py21:26
ekarlsorly ?21:27
ekarlsosjmc7: can u up a new PS then ?21:27
sjmc7errr.. wait a minute.21:27
ekarlso:P21:27
sjmc7no, my recordsets all went away :)21:27
sjmc7one moment please!21:27
sjmc7ekarlso - 015-09-09 16:30:43.176 ERROR searchlight.cmd.manage [-] Failed to setup index extension recordset: No module named ipdb21:30
ekarlsooh haha21:31
ekarlsooops :D21:31
ekarlsodebug!21:31
sjmc7:)21:31
sjmc7worked though, after reomving that21:31
ekarlsobtw, you logging just a LOG.warn in debugmode when something splats is REALLY annoying21:31
sjmc7yeah... we need to fix that21:32
sjmc7it should explode21:32
ekarlsoyou get literally no trace of what's going on except going in and adding LOG.exception...21:32
sjmc7you can now do index sync --type OS::Designate::RecordSet21:32
ekarlsoya, I know :P21:32
sjmc7but yes, i agree21:32
sjmc7you can now do index sync --type OS::Designate::RecordSet21:32
sjmc7gah21:32
sjmc7+    def get_parent_id_field(self):21:32
sjmc7+        return "zone_id"21:32
sjmc7+21:32
sjmc7that's the change to recordsets.py, on top of 149395821:33
ekarlsoyeah I thought I had done that ? :/21:33
sjmc7i don't want to submit a patchset because then i can't review it :)21:33
ekarlsoah!!!21:33
ekarlsoget_parent_id_fields...21:34
ekarlsovs field21:34
sjmc7ah :)21:34
ekarlsothat's what I get for getting up at 0500 and going to bed at 010021:34
ekarlsomeh :P21:34
sjmc7and spinning :)21:34
ekarlsolet's see if my setup works21:34
ekarlsowell, spinning 45 mins of interval spinning and 30 minutes of interval training actually :p21:34
ekarlsowas a zombie at the office today21:35
sjmc7it's just.. wrong21:35
ekarlsoit's nice :p21:35
ekarlsobreak tmrw and interval runniong on friday :P21:35
ekarlsopraise you sjmc7 that worked!21:36
sjmc7hurrah21:36
ekarlso^ gogo21:36
sjmc7if you +1 my patch i'll get it approved so we can get designate in21:37
openstackgerritEndre Karlson proposed openstack/searchlight: Add Designate plugin.  https://review.openstack.org/19909921:37
ekarlsoehm, ^21:37
ekarlsodone sjmc721:39
sjmc7thanks. now go to bed :)21:40
ekarlsohaha, nah21:40
ekarlsogot 1 patch to fix first htne some other stuff :/21:40
ekarlsobesides, tmrw is sleep until 8 ish day :P21:41
ekarlsoluxure!21:41
sjmc7:)21:41
ekarlsosjmc7: fixing up v2 for designate..21:43
ekarlsoand filing dandy ux tickets..21:43
ekarlsov2 horizon that is21:43
sjmc7what what what21:43
ekarlsosjmc7: support for the Designate V2 api in horizon21:44
sjmc7a lot of the javascript folks are at HP21:44
ekarlsoTravT: you got time for a horizon q ?22:00
TravTekarlso: yeah, what's up?22:02
ekarlsoTravT: so, I have a wizardcontroller in angular and a stepcontroller22:03
ekarlsoI've defined ctrl.zone = {} on the wizctrl and ctrl.zone = {type: 'PRIMARY'} on the step one22:03
ekarlsobut when I do submit the data is empty ?22:03
TravTdo you happen to have a patchset i can look at?22:04
ekarlsouh, there's github :P22:04
TravTi can try to look at that.22:05
ekarlsohttps://github.com/ekarlso/designate-dashboard/blob/master/designatedashboard/static/designatedashboard/zones/steps/create-zone/create-zone.controller.js22:05
ekarlsoI've defined PRIMARY there as a base for the inputs in the form22:06
ekarlsothen when hitting submit() in https://github.com/ekarlso/designate-dashboard/blob/master/designatedashboard/static/designatedashboard/zones/actions/create.wizard.controller.js ctrl.zone is {}22:06
ekarlsowhich makes me kinda :/22:08
TravTok, is this in a state that I can pull it and try it out?22:10
ekarlsoshould be yeah22:11
ekarlsono manual with it though ;)22:12
ekarlsojust link in the enabled foles22:12
ekarlsos/foles/files22:12
TravTok, i think i'll have to pull another horizon as well.22:13
ekarlsosjmc7: when do you think we should look at adding options for plugins ?22:27
ekarlsoTravT: did hipchat blow up for you ?22:38
TravTno... too many windows.22:41
sjmc7ekarlso - soon. now :)  i'm going to do some related config stuff now, so i'll maybe take a look at plugin confs while i'm doing it23:02
ekarlsosjmc7: meh, I was actually doing it :p23:07
ekarlsoif you want I can put it up tmrw and save you it :p23:07
sjmc7ah. then go for it :)23:07
sjmc7nah, take a stab at it23:07
sjmc7but get some sleep!23:08
ekarlsoyap, that's next up haha :P23:08
ekarlsoseeya :)23:08
sjmc7g night23:09
*** david-lyle has quit IRC23:11
*** david-lyle has joined #openstack-searchlight23:12
*** david-lyle has quit IRC23:22
*** david-lyle has joined #openstack-searchlight23:23
*** sballe has quit IRC23:36
« Tuesday, 2015-09-08 Index Thursday, 2015-09-10 »

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