Thursday, 2016-04-07

openstackgerritSteve McLellan proposed openstack/searchlight-specs: WIP Cross-region search spec  https://review.openstack.org/30122700:06
openstackgerritMerged openstack/searchlight: Zero Downtime Re-indexing Unit Tests.  https://review.openstack.org/30020300:12
*** yingjun has joined #openstack-searchlight00:45
*** bpokorny has quit IRC01:02
*** RickA-HP has quit IRC01:11
openstackgerritRick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Functional Tests.  https://review.openstack.org/30199001:20
openstackgerritMerged openstack/searchlight: Listen to instance volume attach/detach events  https://review.openstack.org/30194301:34
openstackgerritLi Yingjun proposed openstack/searchlight: Add region_name to service_credentials for devstack  https://review.openstack.org/30249502:11
openstackgerritMerged openstack/searchlight-ui: Searchlight-ui devstack plugin  https://review.openstack.org/30193502:44
openstackgerritLi Yingjun proposed openstack/searchlight: Hypervisor plugin  https://review.openstack.org/29758602:58
openstackgerritTin Lam proposed openstack/searchlight: Making searchlight-manage python-3 compatible  https://review.openstack.org/30091102:59
*** openstackstatus has quit IRC03:01
*** khushbu_ has joined #openstack-searchlight03:07
*** khushbu_ has quit IRC03:34
*** khushbu_ has joined #openstack-searchlight03:44
*** khushbu_ has quit IRC04:20
*** TravT_ has joined #openstack-searchlight04:41
*** TravT has quit IRC04:43
*** GB21 has joined #openstack-searchlight05:00
*** ankur has quit IRC05:21
*** GB21 has quit IRC06:12
*** GB21 has joined #openstack-searchlight06:12
*** pcaruana has joined #openstack-searchlight06:26
*** GB21 has quit IRC06:26
*** GB21 has joined #openstack-searchlight06:28
*** GB21 has quit IRC06:44
*** GB21 has joined #openstack-searchlight06:44
*** GB21 has quit IRC06:51
*** GB21 has joined #openstack-searchlight06:54
*** GB21 has quit IRC07:02
*** GB21 has joined #openstack-searchlight07:03
*** khushbu_ has joined #openstack-searchlight07:10
*** GB21 has quit IRC07:16
*** khushbu_ has quit IRC07:22
*** GB21 has joined #openstack-searchlight07:29
*** GB21 has quit IRC07:37
*** GB21 has joined #openstack-searchlight08:26
*** openstackstatus has joined #openstack-searchlight08:29
*** ChanServ sets mode: +v openstackstatus08:29
-openstackstatus- NOTICE: jobs depending on npm are now working again08:34
*** khushbu has joined #openstack-searchlight08:54
*** khushbu has quit IRC09:32
*** yingjun has quit IRC09:36
*** yingjun has joined #openstack-searchlight09:36
*** yingjun has quit IRC09:41
*** akanksha_ has joined #openstack-searchlight10:01
*** GB21 has quit IRC10:28
*** GB21 has joined #openstack-searchlight10:31
*** GB21 has quit IRC10:49
*** GB21 has joined #openstack-searchlight10:50
*** khushbu has joined #openstack-searchlight10:57
*** khushbu has quit IRC11:12
*** GB21 has quit IRC11:21
tsufievhi, folks!11:49
tsufievcan SearchLight be used for the following task:11:49
tsufievwhen the user wants to delete a tenant in Horizon, he is warned with a list of entities that are contained in that tenant, and if he confirms Delete operation, not only tenant is deleted, but also all the entities that were assigned to that tenant?11:50
tsufievto me the case 'give me a list of things in a particular tenant and do it fast' looks to be perfectly aligned with SearchLight mission11:51
tsufievbut I'm much less confident about the deletion part11:51
*** GB21 has joined #openstack-searchlight11:52
tsufievsjmc7, TravT_ ^^^11:53
*** jhesketh has joined #openstack-searchlight11:54
*** ChanServ changes topic to "OpenStack Searchlight - https://wiki.openstack.org/wiki/Searchlight"11:54
*** khushbu_ has joined #openstack-searchlight12:02
*** jhesketh has quit IRC12:07
*** jhesketh has joined #openstack-searchlight12:07
*** GB21 has quit IRC12:16
*** yingjun has joined #openstack-searchlight12:29
*** yingjun_ has joined #openstack-searchlight12:31
*** yingjun has quit IRC12:31
*** sigmavirus24_awa is now known as sigmavirus2412:47
*** khushbu_ has quit IRC12:49
*** khushbu_ has joined #openstack-searchlight12:52
*** yingjun_ has quit IRC12:54
*** yingjun has joined #openstack-searchlight12:55
*** yingjun has quit IRC12:59
*** TravT_ has quit IRC13:10
*** TravT has joined #openstack-searchlight13:11
*** khushbu_ has quit IRC13:25
openstackgerritRick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Functional Tests.  https://review.openstack.org/30199013:32
*** khushbu_ has joined #openstack-searchlight13:35
openstackgerritRick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Functional Tests.  https://review.openstack.org/30199013:52
sjmc7hi tsufiev14:07
sjmc7we can list all the things belonging to a tenant that we’re able to index. deletions would have to go to each service’s API but part of the actions registry work that matt borland and travis are working on would allow that kind of context sensitive action14:09
*** RickA-HP has joined #openstack-searchlight14:25
tsufievsjmc7, hi!14:27
tsufievyes, I was thinking about the registry stuff14:28
*** yingjun has joined #openstack-searchlight14:28
TravThey tsufiev14:28
TravTsjmc7 has it right14:28
TravTthat is one of the target use cases.14:29
sjmc7that’s going on my resume14:29
TravTbut even without the registered batch actions, you can see the things14:29
tsufievTravT, can the context-sensitive action be triggered programmatically?14:29
sjmc7deletion is interesting because there are some restrictions around ordering and i don’t know how we capture that14:29
TravTtsufiev, have you tried out searchlight at all yet?14:30
tsufievnow I'm ashamed :(14:30
tsufievstill in the backlog14:30
TravTa simple devstack install would let you try it14:30
TravTwe actually even have a ui plugin for it now14:31
TravTthat is devstack installable14:31
TravTi'd also be happy to do a hangout with you if that's easier14:31
tsufievokay, I think the preparations are no more serious than pulling out a patch for code review14:31
tsufievI'll do it14:31
TravTwe have a local conf sample14:32
TravTspeaking of which sjmc7, this can be approved now: https://review.openstack.org/#/c/301901/2/devstack/local.conf14:32
sjmc7okey dokey. all the changes in the ui repo are there to support it?14:33
TravTyep14:33
TravTdavid and li approved it last night14:33
TravTtusfiev https://github.com/openstack/searchlight/tree/master/devstack14:33
sjmc7ok. will test it shortly and +214:33
TravTthere is a sample local.conf there.14:33
tsufievTravT, thank you, going to look at it today-tomorrow14:34
TravTthe only thing it is missing is this line: https://review.openstack.org/#/c/301901/2/devstack/local.conf14:34
TravTsounds good.14:34
TravTtsufiev, if you also want to try out the swift searching, that takes a couple extra manual steps14:34
TravTto get the middleware installed on swift for indexing14:35
tsufievno, I'll omit Swift for the first time14:35
TravTsounds good14:35
TravTsjmc7, regarding the local.conf patch above14:39
sjmc7yes…?14:41
TravTnevermind for now.14:43
TravTi'm gonna bring it up briefly in the meeting14:43
*** khushbu_ has quit IRC14:43
sjmc7ok. regarding your note on https://review.openstack.org/#/c/298570/ - what changes are you expecting for a potential rc3?14:43
*** sigmavirus24 is now known as sigmavirus24_awa14:55
*** sigmavirus24_awa is now known as sigmavirus2414:55
TravTCourtesy Searchlight meeting reminder in #openstack-meeting-4: lakshmiS, nikhil_k, rosmaita, TravT, david-lyle, sjmc7, abhijeetm, itisha, GB21, briancline, lei-zh, yingjun, RickA-HP15:00
*** khushbu has joined #openstack-searchlight15:08
*** GB21 has joined #openstack-searchlight15:42
*** pcaruana has quit IRC15:53
*** lei-zh1 has joined #openstack-searchlight16:01
*** yingjun has quit IRC16:03
openstackgerritLei Zhang proposed openstack/searchlight: Disable index refresh during data re-sync  https://review.openstack.org/30199616:04
openstackgerritSteve McLellan proposed openstack/searchlight: Remove port.create.end handler from nova  https://review.openstack.org/30291816:06
sjmc7TravT: 302918 is the patch for https://bugs.launchpad.net/searchlight/+bug/156752516:06
openstackLaunchpad bug 1567525 in OpenStack Search (Searchlight) "port.create.end events are handled incorrectly by nova" [Critical,New]16:06
sjmc7i have to run an errand i’ve been putting off for weeks, be back in a bit16:06
*** lei-zh1 has left #openstack-searchlight16:07
*** lei-zh1 has joined #openstack-searchlight16:09
*** lei-zh1 has left #openstack-searchlight16:14
*** RickA-HP has quit IRC16:18
openstackgerritOpenStack Proposal Bot proposed openstack/searchlight: Updated from global requirements  https://review.openstack.org/30288816:24
*** bpokorny has joined #openstack-searchlight16:29
*** david-lyle has quit IRC16:36
*** lakshmiS has joined #openstack-searchlight16:39
openstackgerritOpenStack Proposal Bot proposed openstack/searchlight: Updated from global requirements  https://review.openstack.org/30288816:46
*** david-lyle has joined #openstack-searchlight17:06
*** pcaruana has joined #openstack-searchlight17:08
*** david-lyle has quit IRC17:11
*** david-lyle has joined #openstack-searchlight17:17
*** briancline has quit IRC17:18
TravTlakshmiS: sjmc7 plugin output17:20
TravThttps://bugs.launchpad.net/searchlight/+bug/155777817:20
openstackLaunchpad bug 1557778 in OpenStack Search (Searchlight) "Plugin info api does not expose parent child relationships." [Undecided,New] - Assigned to Lakshmi N Sampath (lakshmi-sampath)17:20
sjmc7lakshmiS’s comment is similar to what i had but includes grandchildren17:20
lakshmiSso what structure is useful for horizon to read?17:20
sjmc7lakshmiS: you will also need the parent field mappings17:20
*** briancline has joined #openstack-searchlight17:20
sjmc7zone_id, container_id etc17:20
lakshmiSsjmc7: the structure i had is complete tree and you had a flat strcuture17:21
sjmc7right17:21
lakshmiSits not restriced by one level17:21
sjmc7yeah, my slight preference was direct descendants but i’m not too fussed17:21
TravTwell, i think just immediate parents and children might make more sense17:21
sjmc7there’s some duplication with complete trees17:21
sjmc7i.e container will exist twice17:21
TravTbecause you have to combine them anyway17:21
sjmc7unless any children don’t exist at the top level17:22
sjmc7which i think will complicate parsing17:22
lakshmiSthats right. there is no top level child17:22
TravTlet's pretend OS::Swift::Account had another child17:22
lakshmiSit only exists once17:22
TravTOS::Swift::Container wouldn't know about it's siblings17:22
sjmc7“children” would need to be a list in that model, also17:23
TravTso i think client has to build a tree structure by parsing everything anway?17:23
lakshmiSyes it need to be a list for that17:23
sjmc7TravT: the obvious client use is to enumerate all plugins as you can now17:23
*** david-lyle has quit IRC17:23
sjmc7and i think keeping that is useful without requiring tree parsing17:23
sjmc7i.e. i don’t agree with introducing breaking behavior17:24
lakshmiSi am ok either way or we could include both based on input17:24
openstackgerritOpenStack Proposal Bot proposed openstack/searchlight: Updated from global requirements  https://review.openstack.org/30288817:24
TravTsjmc7 i can't tell if you are agreeing with me or not17:25
sjmc7i’m stating my preference: i think all plugins must exist at the top level17:25
lakshmiSi am actually inclined to do both since client can request in what is useful for them and we cant always predict that17:26
TravToh, okay we were talking about a couple of different things17:26
sjmc7whether we include multiple levels of grandchildren i prefer not, but not too bothered17:26
TravTi missed that aspect of Lakshmi's17:26
lakshmiSso what is the primary use case of this output?17:26
TravTi missed that lakshmi's moved to complete tree structure17:26
*** akanksha_ has quit IRC17:27
TravTi'll talk UI only for a moment17:27
sjmc7ease of use should be the main concern17:27
TravT1) I would like to come up with a way for a general purpose way that people can do parent / child restriction queries17:28
TravTthey can do this today in kind of a not so great way.17:28
TravTbut really can only do has parent17:29
TravTby restricting on the parent id field17:29
TravT2) i think there may be a way we could make the resource type structure navigable / learnable17:30
TravTi was talking with piet and another ux person the other day17:31
TravTabout how to make things usable and learnable17:31
sjmc7to people or machines?17:31
TravTpeople17:31
sjmc7i think maybe it doesn’t make sense to do this if we don’t know what we want17:31
sjmc7or rather, until we know what we want17:31
TravTam i not pointing out why i want it?17:31
sjmc7you are pointing out you want something, but it’s much bigger than just the parent relationships17:32
TravTanyway17:32
TravTumm17:32
sjmc7sorry, finish up17:32
sjmc7i shouldn’t presume & interrupt17:32
TravTso let me go back to 1) then17:33
TravTi simply can not in the ui even think about creating a way to do parent child queries17:33
TravTbecause all it has is an id17:34
TravTno knowledge of even what that id type is17:34
sjmc7yep. so i think we can solve that with exposing the parent/child relationships17:34
TravTwhich also relates to this: https://blueprints.launchpad.net/searchlight/+spec/disoverable-resource-type-mapping17:34
sjmc7if it knows that RecordSet:’zone_id’ refers to Zone:id17:35
TravTon #2)17:35
sjmc7yes - so that one the scope is much larger17:35
sjmc7it’s more like exposing a subset of the mapping17:35
TravTone way of learning is being able to explore structure17:35
sjmc7maybe we can solve 1) bearing 2) in mind17:36
lakshmiSmy argument is that it is so easy to create output in both tree and flat; lets do it and leave the user/ui to use it the way it fits to the scenario17:36
sjmc7well, wait - if we want to move to a point where the info includes that Server:image.id refers to OS::Glance::Image we can do the parent-child in the same way17:37
sjmc7so perhaps not use “children” as the key, but “metadata” or something17:37
sjmc7the fact it’s parent child is sort of a special case of that17:37
TravTyeah, that's also something I was thinking when I added that bp above17:38
sjmc7but i do think we can add the parent-child links in first, and expand it later17:38
sjmc7just bear the later stage in mind17:38
TravTso from a learning perspective17:39
lakshmiSthe expansion can include a lot more fields for interlinking17:39
TravTa computer can easily assemble a tree from flat with just parent and child17:39
TravTa person can't17:39
sjmc7our trees are not going to be more than 3 levels deep, i wouldn’t think17:40
TravTalso, what we do with rbac?17:40
TravTis there a case where you can see a child, but not a parent?17:40
sjmc7yeah, potentially17:41
lakshmiShmm yes17:41
sjmc7another reason i don’t like a full tree17:41
lakshmiSatleast not all the fields in parent17:41
sjmc7it just seems to complicate stuff17:41
lakshmiSplugin info is only for structure right?17:41
lakshmiSrbac will still gets applied irrespecitve of that17:41
lakshmiSi dont see a reason for complication17:42
sjmc7what if Snapshots are visible, Volumes are not?17:42
lakshmiSor maybe i am not understanding your point17:42
lakshmiSso what are we including in plugin info for volume17:44
lakshmiSother than index17:44
sjmc7currently, you wouldn’t see Volume at all17:44
lakshmiSah i see17:44
sjmc7http://paste.openstack.org/show/F2UwFQ7GOWQvAMYy6IWN/17:45
sjmc7that’s what i mean with using metadata17:45
sjmc7that could be expanded to include other relationships17:45
lakshmiSthe scope has definetly expanded for plugin info17:45
sjmc7and you could flip that for zone->recordset17:46
sjmc7i think the only thing i’d be really firm on is that moving to a full tree is a breaking change17:46
sjmc7and i would be very loathe to do that17:46
lakshmiSit depends on what fields you show in metadata17:47
TravTsjmc7: another option is to show that info in facets17:47
sjmc7yep17:47
sjmc7this is true17:47
lakshmiSif its only parentd id field, thats not something useful to user since they cant retrieve it17:47
sjmc7not sure i understand17:48
sjmc7lakshmiS:17:48
sjmc7i think also we’re going to go round in circles forever :)  if we can agree on the requirements at least, that might be a better start17:48
lakshmiSi mean from an rbac perspective even if volumes are not visible to the user, including parent id of volume in snapshot is not going to reveal anything that rbac would disallow17:49
lakshmiSit is going to be field name and not any value of the field17:49
TravTwell, i have two things i know i'd like to do in the UI now. Easily restrict by parent (which is indirectly possible by guessing parent id field)17:50
TravTand be able to know how to handle resource id linkages17:51
sjmc7ok17:51
sjmc7so, given that, and that we don’t necessarily need to solve it in one step, that should influence the design17:51
*** lakshmiS has quit IRC17:52
*** lakshmiS_ has joined #openstack-searchlight17:52
sjmc7perhaps we add explict parent/children, and then other linkages later; perhaps they’re similar enough that the separation isn’t so forced17:53
TravTyou know17:54
TravTi'm thinking i could make the has parent in ui easier with just the metadata in facet...17:55
*** david-lyle has joined #openstack-searchlight17:56
*** david-lyle has quit IRC17:56
lakshmiS_That's better than metadata in plugin info17:56
sjmc7problem solved :)17:57
*** david-lyle has joined #openstack-searchlight17:58
sjmc7so now the decision is how to indicate in the facet output that a field refers to another type, and that (in some cases) that type is the parent of this one17:59
TravTi think so, with one question on top of that17:59
TravTare facets too limiting18:00
sjmc7:D18:00
TravTIOW, can i have resource IDs in fields that are not listed as facets?18:00
TravTmy brain made an escape attempt due to lack of food energy18:00
sjmc7yes. but arguably that shouldn’t be the case18:00
TravTi gotta run for lunch18:00
sjmc7yeah, me too. i think putting those links in facets is reasonable18:01
sjmc7what you will NOT get there is the ‘child’ links18:01
TravTthat is very true...18:01
sjmc7though i guess you could indicate that Zone.“id” refers to RecordSet18:01
TravTin my limited use with the ui, i haven't needed that yet18:01
sjmc7you might if your ‘details’ included children18:02
sjmc7“show me all subnets belonging to this network"18:02
TravTbut it definitely seems like you would want to say, show me zones with x record18:02
TravTyeah18:02
TravTokay, let's come back on this after a little bit18:04
khushbusjmc7: I have some queries related to this bug https://bugs.launchpad.net/searchlight/+bug/152608018:05
openstackLaunchpad bug 1526080 in OpenStack Search (Searchlight) "'TypeError: format requires a mapping' from some error messages" [Medium,Incomplete] - Assigned to khushbu (khushbuparakh)18:05
*** TravT has quit IRC18:06
sjmc7khushbu: shoot, although i am trying to squeeze in some food :)18:08
sjmc7it’s possible this has been fixed by other changes since it was filed18:09
sjmc7lot of water under the bridge since december18:10
khushbuso I am trying out to find the repo n I got this https://github.com/openstack/searchlight/blob/master/searchlight/elasticsearch/plugins/nova/servers_notification_handler.py with some error logs so we are facing type error in this file or some other18:10
sjmc7i think this may have been fixed18:12
sjmc7see https://github.com/openstack/searchlight/commit/faba34ddb88f05ed12478a07d04b074a76452195 (scroll down to servers_notification_handler.py)18:12
sjmc7some of the log messages were altered as part of that patch18:12
khushbuok18:18
sjmc7thanks for looking at it, sorry it was a red herring :(18:18
khushbuSo one should mark it as complete18:19
sjmc7yeah, i’ll close it18:19
khushbuThanks for help18:20
*** TravT has joined #openstack-searchlight18:42
*** GB21 has quit IRC19:03
*** nikhil has quit IRC19:07
*** nikhil has joined #openstack-searchlight19:11
*** david-lyle_ has joined #openstack-searchlight19:13
*** david-lyle has quit IRC19:15
*** sigmavirus24 is now known as sigmavirus24_awa19:17
*** david-lyle_ has quit IRC19:19
*** bpokorny_ has joined #openstack-searchlight19:23
*** briancline has quit IRC19:24
*** briancline has joined #openstack-searchlight19:25
*** bpokorny has quit IRC19:26
*** david-lyle has joined #openstack-searchlight19:29
*** david-lyle has quit IRC19:31
*** david-lyle has joined #openstack-searchlight19:31
*** khushbu has quit IRC19:32
*** david-lyle has quit IRC19:37
*** TravT has quit IRC19:39
*** bpokorny_ has quit IRC19:40
*** bpokorny has joined #openstack-searchlight19:41
*** bpokorny has quit IRC19:52
*** bpokorny has joined #openstack-searchlight19:53
*** lakshmiS has joined #openstack-searchlight20:07
*** bpokorny has quit IRC20:16
*** lakshmiS_ has quit IRC20:54
*** david-lyle has joined #openstack-searchlight20:56
*** pcaruana has quit IRC21:01
*** bpokorny has joined #openstack-searchlight21:02
*** bpokorny has quit IRC21:25
*** bpokorny has joined #openstack-searchlight21:26
*** bpokorny_ has joined #openstack-searchlight22:01
*** bpokorny_ has quit IRC22:01
*** bpokorny_ has joined #openstack-searchlight22:02
*** bpokorny has quit IRC22:04
*** briancline has quit IRC22:14
*** RickA-HP has joined #openstack-searchlight22:17
*** briancline has joined #openstack-searchlight22:17
*** briancline has quit IRC22:25
*** briancline has joined #openstack-searchlight22:27
*** lakshmiS has quit IRC22:46
*** briancline has quit IRC22:46
*** briancline has joined #openstack-searchlight22:52
*** briancline has quit IRC23:02
*** briancline has joined #openstack-searchlight23:07
*** briancline has quit IRC23:22
*** briancline has joined #openstack-searchlight23:25
*** briancline has quit IRC23:34
*** briancline has joined #openstack-searchlight23:38
*** bpokorny has joined #openstack-searchlight23:40
*** bpokorny_ has quit IRC23:40
*** bpokorny has quit IRC23:46
*** bpokorny has joined #openstack-searchlight23:47

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