| *** akanksha_ has joined #openstack-searchlight | 00:26 | |
| *** itisha has joined #openstack-searchlight | 00:47 | |
| *** yingjun has joined #openstack-searchlight | 00:49 | |
| *** bpokorny_ has joined #openstack-searchlight | 01:12 | |
| *** bpokorny has quit IRC | 01:15 | |
| *** bpokorny_ has quit IRC | 01:16 | |
| *** bpokorny has joined #openstack-searchlight | 02:32 | |
| *** akanksha_ has quit IRC | 02:57 | |
| *** itisha has quit IRC | 02:59 | |
| *** bpokorny has quit IRC | 03:01 | |
| *** krotscheck has joined #openstack-searchlight | 03:30 | |
| *** krotscheck_dcm has quit IRC | 03:32 | |
| *** bpokorny has joined #openstack-searchlight | 03:40 | |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 03:41 |
|---|---|---|
| *** bpokorny has quit IRC | 04:08 | |
| *** bpokorny has joined #openstack-searchlight | 04:24 | |
| *** TravT_ has joined #openstack-searchlight | 04:41 | |
| *** TravT has quit IRC | 04:43 | |
| *** bpokorny has quit IRC | 04:47 | |
| *** nikhil has quit IRC | 05:13 | |
| *** nikhil has joined #openstack-searchlight | 05:14 | |
| *** GB21 has joined #openstack-searchlight | 05:22 | |
| *** GB21 has quit IRC | 05:35 | |
| *** pcaruana has quit IRC | 05:49 | |
| *** GB21 has joined #openstack-searchlight | 06:31 | |
| *** pcaruana has joined #openstack-searchlight | 07:41 | |
| *** GB21 has quit IRC | 07:45 | |
| *** briancli1e has quit IRC | 08:12 | |
| *** briancline has joined #openstack-searchlight | 08:13 | |
| *** GB21 has joined #openstack-searchlight | 08:30 | |
| *** yingjun has quit IRC | 09:24 | |
| *** exploreshaifali has joined #openstack-searchlight | 09:36 | |
| *** GB21 has quit IRC | 10:41 | |
| *** GB21 has joined #openstack-searchlight | 11:00 | |
| *** TravT_ has quit IRC | 11:17 | |
| *** TravT has joined #openstack-searchlight | 11:18 | |
| *** dhellmann has quit IRC | 11:22 | |
| *** dhellmann_ has joined #openstack-searchlight | 11:22 | |
| *** dhellmann_ is now known as dhellmann | 11:26 | |
| *** yingjun has joined #openstack-searchlight | 11:57 | |
| *** GB21 has quit IRC | 12:05 | |
| *** exploreshaifali has quit IRC | 12:44 | |
| *** exploreshaifali has joined #openstack-searchlight | 12:56 | |
| *** exploreshaifali has quit IRC | 13:19 | |
| *** yingjun has quit IRC | 13:19 | |
| *** yingjun has joined #openstack-searchlight | 13:20 | |
| *** yingjun has quit IRC | 13:24 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:05 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 14:14 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:15 | |
| sjmc7 | rosmaita: when you have a sec, could i trouble you for your blessing on https://review.openstack.org/#/c/298315? adds some missing documentation i’d like to get in for rc2 | 14:43 |
| * rosmaita looking | 14:44 | |
| sjmc7 | ta | 14:49 |
| rosmaita | sjmc7: may take a bit, i am in 2 concurrent meetings atm | 14:52 |
| sjmc7 | only two? :) no rush | 14:53 |
| *** lei-zh has joined #openstack-searchlight | 14:59 | |
| TravT | Courtesy Searchlight meeting reminder in #openstack-meeting-4: lakshmiS, nikhil_k, rosmaita, TravT, david-lyle, sjmc7, abhijeetm, itisha, GB21, briancline, lei-zh, yingjun, RickA-HP | 15:00 |
| *** yingjun has joined #openstack-searchlight | 15:00 | |
| *** bpokorny has joined #openstack-searchlight | 15:24 | |
| *** bpokorny has quit IRC | 15:28 | |
| *** bpokorny has joined #openstack-searchlight | 15:30 | |
| openstackgerrit | Merged openstack/searchlight: Add missing zero-downtime indexing documentation https://review.openstack.org/298315 | 15:38 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 15:38 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 15:40 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 15:41 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 15:41 | |
| *** bpokorny has quit IRC | 15:48 | |
| *** bpokorny has joined #openstack-searchlight | 15:48 | |
| *** yingjun has quit IRC | 16:01 | |
| *** lakshmiS has joined #openstack-searchlight | 16:02 | |
| *** lei-zh has quit IRC | 16:04 | |
| sjmc7 | lakshmiS: the reason the ‘records’ data isn’t structured is that the structure isn’t very structured | 16:06 |
| sjmc7 | for dns | 16:06 |
| sjmc7 | seems very context dependent | 16:06 |
| lakshmiS | its a text blob of ip's and stuff | 16:07 |
| sjmc7 | it’s bloody confusing, is what it is :) | 16:07 |
| *** pcaruana has quit IRC | 16:08 | |
| sjmc7 | lakshmiS: i’m seeing recordset.create notifications arriving after record.create under the v1 API | 16:13 |
| sjmc7 | which makes the record handling fail with a 404 | 16:14 |
| sjmc7 | this is eerily reminiscent of the problem i had with DHCP ports | 16:14 |
| lakshmiS | i got it in the right order | 16:14 |
| lakshmiS | recordset and then record with v1 | 16:14 |
| sjmc7 | actually, i’m not sure i get a recordset one at all. there’s a zone update that gets sent | 16:17 |
| lakshmiS | so you ran designate record-create ? | 16:17 |
| sjmc7 | yeah | 16:17 |
| sjmc7 | ah, there it is | 16:18 |
| sjmc7 | hmm.. this seems like a potential race condition | 16:19 |
| sjmc7 | it looks like this will work as long as the recordset create is processed in time for the record.create | 16:20 |
| lakshmiS | we will have to lookup recordset instead of relying on notification when record.create arrives then.. | 16:21 |
| lakshmiS | exception condition when its not in es | 16:21 |
| sjmc7 | from the api? | 16:21 |
| lakshmiS | yes | 16:21 |
| sjmc7 | yeah, maybe. naively i guess we could always reindex the recordset from the api | 16:21 |
| sjmc7 | depends how big that race is | 16:25 |
| sjmc7 | because the other danger is that if we pull the recordset from the api and then the recordset event comes in we’ll wipe out the records that way :) | 16:26 |
| lakshmiS | hmm | 16:28 |
| lakshmiS | its going to endup all api lookups :) | 16:29 |
| sjmc7 | kind of feels like for a v1 api call we should go back to the api | 16:29 |
| sjmc7 | except of course we don’t know whether it is v1 | 16:30 |
| lakshmiS | thats the problem with recordset. let me check if they show that in payload for the origin | 16:31 |
| sjmc7 | doubtful; probably the empty record list would be the indicator | 16:32 |
| lakshmiS | yeah nothing there to indicate. for the empty check - have to make sure v2 recordset create allows empty records or not | 16:35 |
| sjmc7 | doesn’t seem to | 16:36 |
| sjmc7 | —records is required by the CLI | 16:36 |
| sjmc7 | so one option might be to ignore recordset create events with empty records, and always index from the API for record.create? | 16:37 |
| sjmc7 | all the record.create handlers would hit the api, v2 notifications wouldn't | 16:37 |
| lakshmiS | yes that would be better | 16:37 |
| sjmc7 | since we’re removing that code anyway once v1 is gone, not too worried about efficiency | 16:38 |
| lakshmiS | its not efficient even for record.update due to api design | 16:39 |
| sjmc7 | v1 looks very fiddly, i can see why they wanted to change it | 16:39 |
| sjmc7 | yeah, i think we could hit the reocrdset API on *any* record event | 16:39 |
| sjmc7 | delete is a bit complicated too | 16:40 |
| sjmc7 | with v1 i don’t see how you delete a recordset at all | 16:41 |
| lakshmiS | v1 doesnt have recordset conept | 16:42 |
| openstackgerrit | Brian Rosmaita proposed openstack/searchlight: Eliminate stray asterisk https://review.openstack.org/300088 | 16:42 |
| lakshmiS | so if users exclusively use v1 api | 16:42 |
| sjmc7 | ah, ok. it just sends a few notifications for good measure | 16:42 |
| lakshmiS | forward compatibility :) | 16:42 |
| sjmc7 | so what happens if i delete all the records in a zone under v1..? | 16:43 |
| sjmc7 | the recordset becomes empty? | 16:43 |
| lakshmiS | you end up with a recordset with type NS and SOA records | 16:43 |
| sjmc7 | ah, those are special? | 16:43 |
| lakshmiS | yes | 16:43 |
| sjmc7 | ok, so let’s just hit the API on all record. events | 16:44 |
| sjmc7 | that seems easier | 16:44 |
| sjmc7 | unless it really is just create that’s the problem | 16:44 |
| lakshmiS | yes to the efficiency(to make sure v1 users not use SL) | 16:44 |
| sjmc7 | users will use both | 16:45 |
| sjmc7 | wait, if i delete a record, i see the associated recordset go away too | 16:51 |
| lakshmiS | did you get a recordset delete event too? | 16:52 |
| sjmc7 | i don’t think so :( | 16:52 |
| lakshmiS | so you verified that recordset is gone from v2 cli? | 16:52 |
| sjmc7 | yeah | 16:53 |
| lakshmiS | damn | 16:53 |
| sjmc7 | it took a while to do it | 16:53 |
| sjmc7 | yeah, no recordset-delete as far as i can see | 16:53 |
| lakshmiS | may be it will come eventually otherwise its a bug in designate | 16:54 |
| sjmc7 | yeah. so what’s a sensible compromise we can make? | 16:54 |
| sjmc7 | option 1 is do nothing :) | 16:54 |
| sjmc7 | mugsie: you around? | 16:54 |
| mugsie | sjmc7: just about to hop on to a meeting - can I ping you in 30? | 16:55 |
| sjmc7 | sure | 16:55 |
| sjmc7 | thanks | 16:55 |
| lakshmiS | even for delete we could lookup the api and remove recordset if its not in designate | 16:56 |
| sjmc7 | i don’t know how instantaneous it is | 16:56 |
| lakshmiS | yeah time dependent | 16:56 |
| sjmc7 | it actually seems to take a while until the record disappears from the api | 16:56 |
| sjmc7 | for delete, if the record list is empty, the set can be deleted | 16:57 |
| lakshmiS | which is ok as long as designate guarantees notification but that's something we cannot address | 16:57 |
| sjmc7 | for update, i think what you have is fine | 16:57 |
| sjmc7 | designate seems to guarantee NO recordset.delete notification for record deletes | 16:57 |
| sjmc7 | so i think that’d be ok | 16:58 |
| sjmc7 | for creates… | 16:58 |
| sjmc7 | i’m leaning towards go to the api | 16:58 |
| lakshmiS | hope you don't get that delete recordset notification by lunch :) | 16:58 |
| sjmc7 | ignore recordset.create with empty records | 16:58 |
| sjmc7 | if we do, we’ll ignore it | 16:58 |
| sjmc7 | it’ll be a silent 404 | 16:58 |
| lakshmiS | api makes sense for create record | 16:59 |
| sjmc7 | i think the other two are fine, but delete needs to remove the recordset if the list of records is empty after filtering | 17:03 |
| sjmc7 | at this point in time it’s gonna be a bit of a best-effort | 17:04 |
| lakshmiS | yeah no other way | 17:07 |
| sjmc7 | does that make sense, to do that? | 17:07 |
| sjmc7 | so create: ignore recordset.create events with empty ‘record’ lists | 17:07 |
| sjmc7 | and index from the api | 17:07 |
| lakshmiS | delete can make an extra check for recordset | 17:08 |
| sjmc7 | update: leave as you have it from the api | 17:08 |
| sjmc7 | delete: leave as you have it except that if ‘records’ is empty, delete the recordset | 17:08 |
| lakshmiS | i will try to comment as much as i can. god help someone who wants to understand it :) | 17:09 |
| sjmc7 | :) | 17:09 |
| sjmc7 | well, we can drop it all in newton :) | 17:09 |
| lakshmiS | hopefully | 17:09 |
| sjmc7 | ok. i’ll add comments on the review as a note | 17:11 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 17:17 | |
| openstackgerrit | Travis Tripp proposed openstack/searchlight-ui: Fix issues when magic search facet uses range queries https://review.openstack.org/300107 | 17:38 |
| -openstackstatus- NOTICE: Job failures for missing traceroute packages are in the process of being fixed now, ETA 30 minutes to effectiveness for new jobs | 17:49 | |
| openstackgerrit | Travis Tripp proposed openstack/searchlight-ui: Fix issues when magic search facet uses range queries https://review.openstack.org/300107 | 18:01 |
| TravT | sjmc7: ^ that one is ready for testing. I'll go look at your comments on the parent patch. | 18:02 |
| mugsie | sjmc7: sorry - got distracted | 18:09 |
| mugsie | the recordset.delete event question? (on the deletion of the last record in it) seems like a bug | 18:11 |
| mugsie | we should emit that notification | 18:11 |
| sjmc7 | mugsie: yeah, probably. when’s the v1 api going away though? | 18:12 |
| mugsie | first commit of Octa if I have my way | 18:13 |
| sjmc7 | ok. well, given we have to get RC 2 done today we’ll go with this | 18:14 |
| sjmc7 | we can tweak it in newton | 18:14 |
| openstackgerrit | Travis Tripp proposed openstack/searchlight-ui: Searchlight UI Query Error Handling https://review.openstack.org/299566 | 18:19 |
| openstackgerrit | Travis Tripp proposed openstack/searchlight-ui: Fix issues when magic search facet uses range queries https://review.openstack.org/300107 | 18:20 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 18:20 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 18:22 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 18:24 | |
| *** bpokorny has quit IRC | 18:24 | |
| sjmc7 | lakshmiS: can we disable the functional test CI jobs? we’re getting an unacceptably high failure rate | 18:38 |
| lakshmiS | seems like elasticsearch connection is timing out. have seen that happen on busy systems. i wanted to check if we can increase es timeout setting | 18:39 |
| *** bpokorny has joined #openstack-searchlight | 18:40 | |
| *** bpokorny has quit IRC | 18:40 | |
| sjmc7 | no, there’re a lot of memory errors | 18:40 |
| sjmc7 | timeouts are a result of crashes | 18:40 |
| *** bpokorny has joined #openstack-searchlight | 18:41 | |
| lakshmiS | can we try incrasing heap space?. temporarily we can disable for now, but it will be tough to test since it only fails on jenkins | 18:44 |
| sjmc7 | i’d like to disable it if possible | 18:44 |
| sjmc7 | we have to get RC2 done today | 18:44 |
| sjmc7 | and we’ll incur the wrath of infra | 18:45 |
| lakshmiS | that may be a problem | 18:45 |
| lakshmiS | hopefully they approve soon. i will try | 18:45 |
| sjmc7 | we can try decreasing memory usage too. lemme take a look at what we’re using now | 18:45 |
| sjmc7 | the problem is i think there are two errors; that the JVM runs out, and that the machine runs out | 18:46 |
| sjmc7 | looks like we’ve got it set pretty low. i’ll increase it and see if it helps. i don’t know how much memory the VMs have | 18:47 |
| sjmc7 | lakshmiS: i’ll put a patch up to increase the heap space for now | 18:49 |
| sjmc7 | and make a note to do it more efficiently | 18:49 |
| lakshmiS | ok | 18:50 |
| lakshmiS | its easy to try out in our code anyway, but i will keep the patch up ready incase... | 18:51 |
| openstackgerrit | Steve McLellan proposed openstack/searchlight: Increase heap for elasticsearch in func tests https://review.openstack.org/300150 | 18:51 |
| lakshmiS | approved | 18:53 |
| sjmc7 | we should be able to just run one for all the tests with different base index names i think | 18:53 |
| sjmc7 | but i don’t want to try to do that now | 18:53 |
| lakshmiS | yeah we could | 18:55 |
| sjmc7 | it’s tricky cos of the multiprocessing | 18:55 |
| openstackgerrit | Merged openstack/searchlight: Fix _is_multiple_alias_exception signature https://review.openstack.org/298947 | 18:58 |
| *** openstackgerrit has quit IRC | 19:03 | |
| *** openstackgerrit has joined #openstack-searchlight | 19:05 | |
| openstackgerrit | Merged openstack/searchlight: Increase heap for elasticsearch in func tests https://review.openstack.org/300150 | 19:05 |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 19:17 |
| TravT | sjmc7: are you going to cherry pick this one to stable/mitaka as well? | 19:29 |
| TravT | https://review.openstack.org/#/c/300150/ | 19:29 |
| sjmc7 | no | 19:29 |
| sjmc7 | i wasn’t planning to | 19:29 |
| sjmc7 | i guess it doesn’t hurt though in case we have to do backports in future | 19:29 |
| TravT | yeah, i'm not seeing a downside | 19:30 |
| sjmc7 | ok. i was trying to reduce what we’re pushing but it doesn’t do any harm. unless it makes nodepool machines run out of memory | 19:30 |
| sjmc7 | next time, don’t change infra the week of the release :) | 19:31 |
| TravT | what could possibly go wrong! | 19:31 |
| sjmc7 | https://review.openstack.org/#/c/300181/ | 19:31 |
| openstackgerrit | Merged openstack/searchlight: Eliminate stray asterisk https://review.openstack.org/300088 | 19:31 |
| TravT | cool | 19:31 |
| openstackgerrit | Merged openstack/searchlight: Disable oslo_config file discovery in tests https://review.openstack.org/298830 | 19:32 |
| sjmc7 | it seems to have sorted it out, fingers crossed | 19:32 |
| TravT | when in doubt, increase memory | 19:32 |
| sjmc7 | looks like they recommend 8GB of RAM for nodepool machines so a few hundred MB shouldn’t tip the balance | 19:32 |
| lakshmiS | if fails we will increase again :) | 19:33 |
| TravT | i got a sweet pop-up ad offering to download more memory to my system the other day. maybe we could do that for the nodepool machines. :-P | 19:33 |
| sjmc7 | TravT: looking at https://review.openstack.org/#/c/300107 (facet qs) | 19:34 |
| sjmc7 | to save me some time, what was the before/after? | 19:34 |
| sjmc7 | oh wait, they’re in the commit message | 19:35 |
| sjmc7 | haha | 19:35 |
| TravT | yeah | 19:35 |
| TravT | i can also screenshare show you as well | 19:35 |
| sjmc7 | nah, that’s ok | 19:35 |
| sjmc7 | what was the problem before? | 19:35 |
| TravT | turned out to be two things really | 19:36 |
| TravT | first of all, the facets were all doing a term query, so for things like min_ram or size or date, etc | 19:36 |
| sjmc7 | ok | 19:36 |
| TravT | it would only match if you put in exact match | 19:36 |
| TravT | you couldn't do like min_ram <512 | 19:37 |
| TravT | it would just not match | 19:37 |
| sjmc7 | got it | 19:37 |
| TravT | next thing was that I found when testing >= <= that due to how magic search sends facets through, it was only grabbing part of the input | 19:37 |
| TravT | and if you have say a search phrase (for like description) or something that had an = sign in it, it would only grab part | 19:38 |
| TravT | magic search sends facets through in URL query parameter status, but it doesn't url encode things first | 19:38 |
| sjmc7 | sigh | 19:38 |
| TravT | yeah, | 19:38 |
| sjmc7 | nothing you tell me about magicsearch makes me like it :) | 19:39 |
| TravT | me more so than you | 19:39 |
| TravT | other than it makes the facets discoverable for less technical users | 19:39 |
| sjmc7 | i think by now you could’ve written it :) | 19:39 |
| TravT | yeah, in retrospect... | 19:39 |
| TravT | that actually was what I woke up debating | 19:39 |
| sjmc7 | can you approve 300181 for the stable branch? | 19:40 |
| sjmc7 | hitting the same memory error merging the cherrypicks | 19:40 |
| sjmc7 | i think after that the designate one’s the last patch for rc2, gonna test again now | 19:41 |
| TravT | yeah, that's why i asked about backporting the memory fix | 19:41 |
| sjmc7 | i’ve +2ed the UI ones, both look fine | 19:41 |
| sjmc7 | lakshmiS: aside from the test failures the designate one’s finished? | 19:42 |
| sjmc7 | i feel like i’ve spent a day in travis’ shoes, reviewing patches :) | 19:42 |
| TravT | hehe... i could crank out a couple more ui fixes for you if you want. | 19:43 |
| lakshmiS | ah aah. i have added functional tests for zones. debating whether to add for recordset/record or not coz that will take more time | 19:43 |
| TravT | ;) | 19:43 |
| sjmc7 | lakshmiS: given we didn’t have any before, i’m ok with not adding them now | 19:43 |
| sjmc7 | we really should be trying to finish up | 19:43 |
| lakshmiS | yeah i will lave the records to another patch for which a bug is already there | 19:44 |
| lakshmiS | s/lave/leave | 19:44 |
| TravT | lakshmiS: i really want to get the backports done and kick off a fresh stack from stable/mitaka | 19:44 |
| TravT | so I can put up the release tags | 19:44 |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 19:44 |
| lakshmiS | ok dnoe | 19:44 |
| lakshmiS | done | 19:44 |
| lakshmiS | lets wait for gate gods | 19:45 |
| TravT | lakshmiS: can you please look at the UI patches as well? | 19:45 |
| lakshmiS | will do that now | 19:45 |
| lakshmiS | one last change. my editor ends up with tab instead of spaces... | 19:46 |
| TravT | ok. | 19:46 |
| sjmc7 | the test files are still full of tabs, lakshmiS | 19:53 |
| lakshmiS | yeah fixing the same | 19:54 |
| -openstackstatus- NOTICE: Any jobs which erroneously failed on missing traceroute packages should be safe to recheck now | 19:55 | |
| sjmc7 | also the create and update record handlers are identical i think now? | 19:57 |
| *** briancline has quit IRC | 19:58 | |
| *** briancline has joined #openstack-searchlight | 19:58 | |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 19:58 |
| lakshmiS | sjmc7: i left them separate to have the comments to explain the reason | 20:00 |
| sjmc7 | hmm.. ok | 20:00 |
| sjmc7 | i’m not a big fan of duplicated functions though | 20:01 |
| lakshmiS | if another commit is required i can optimize the comments ;( | 20:05 |
| TravT | i just found something a little odd with the search panel. | 20:07 |
| *** briancline has quit IRC | 20:10 | |
| TravT | false alarm. | 20:14 |
| openstackgerrit | Merged openstack/searchlight-ui: Searchlight UI Query Error Handling https://review.openstack.org/299566 | 20:15 |
| *** briancline has joined #openstack-searchlight | 20:17 | |
| *** briancline has quit IRC | 20:26 | |
| openstackgerrit | Rick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Unit Tests. https://review.openstack.org/300203 | 20:31 |
| *** bpokorny has quit IRC | 20:32 | |
| TravT | lakshmiS: the new dns recordsets i'm creating don't seem to have a project id | 20:34 |
| *** bpokorny has joined #openstack-searchlight | 20:34 | |
| sjmc7 | TravT: through the v1 or v2 api? | 20:35 |
| TravT | whatever openstack client uses | 20:35 |
| sjmc7 | i see project_ids | 20:35 |
| sjmc7 | from both | 20:35 |
| lakshmiS | yeah just check that | 20:35 |
| lakshmiS | checked | 20:36 |
| TravT | let me see debug breakpoint to see what comes in via notification. | 20:36 |
| TravT | s/see/set | 20:36 |
| sjmc7 | it’s not explicitly in the mapping but there’s a dynamic template that catches it | 20:36 |
| sjmc7 | be aware that debugging it is REALLY annoying | 20:37 |
| sjmc7 | because it sends several updates at weird intervals | 20:37 |
| sjmc7 | i kept getitng bounced around threads | 20:37 |
| sjmc7 | keep an eye on which thread is handling which events | 20:37 |
| TravT | ugh | 20:37 |
| sjmc7 | yeah | 20:37 |
| TravT | so, i'm running the resources script | 20:38 |
| TravT | https://review.openstack.org/#/c/230697/14/sample-scripts/resources.sh | 20:38 |
| TravT | which executes line 188 | 20:38 |
| TravT | resources.sh -n designate -t OS::Designate::Zone | 20:39 |
| *** briancline has joined #openstack-searchlight | 20:43 | |
| TravT | i just re-ran it 5 times. | 20:46 |
| openstackgerrit | Rick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Unit Tests. https://review.openstack.org/300203 | 20:47 |
| lakshmiS | so do you see a tenant_id in notification event payload ? | 20:47 |
| TravT | it isn't hitting the breakpoint... | 20:48 |
| TravT | let me dump it out | 20:48 |
| TravT | oh, this is interesting | 20:49 |
| TravT | try adding all_projects to your query | 20:49 |
| TravT | the couple of records that are created automatically when you create a zone have project | 20:51 |
| TravT | the record i create with designate record-create "$zone_id" --name www."$zone_name" --type A --data 192.0.2.1 | 20:51 |
| TravT | doesn't have it | 20:51 |
| TravT | i don't have the object dumped yet | 20:52 |
| lakshmiS | that is strange. same cli command you showed creates a tenant_id for me | 20:52 |
| TravT | but i'm almost wondering if line 175 overwrites it | 20:52 |
| TravT | https://review.openstack.org/#/c/299733/4/searchlight/elasticsearch/plugins/designate/notification_handlers.py | 20:52 |
| lakshmiS | designate record-create 6837da17-ffd3-46ca-9fb3-f82804a26c9f --name www.domain1-example.com. --type A --data 192.0.2.1 | 20:52 |
| lakshmiS | event_type:dns.record.create | 20:52 |
| lakshmiS | event_type:{u'status': u'PENDING', u'zone_id': u'6837da17-ffd3-46ca-9fb3-f82804a26c9f', u'managed': False, u'managed_resource_id': None, u'managed_resource_type': None, u'tenant_id': u'80264096ac454d3d904002491fafe2ec', u'created_at': u'2016-03-30T04:23:48.000000', u'managed_extra': None, u'updated_at': None, u'managed_plugin_type': None, u'version': 1, u'managed_plugin_name': None, u'managed_tenant_id': None, u'action': u'CREATE', | 20:52 |
| lakshmiS | u'hash': u'bc16c7fc29e740c99f580efa9fc0d247', u'managed_resource_region': None, u'recordset_id': u'7f36cd3a-7ad8-49ec-b5c2-253eb495fd32', u'data': u'192.0.2.1', u'id': u'55af5e51-9550-471b-8fbd-791f8f2e53f1', u'serial': 1459311828, u'description': None} | 20:52 |
| TravT | hmmm | 20:52 |
| TravT | maybe something changed in designate? | 20:53 |
| TravT | did they land code to move to project_id recently? | 20:53 |
| TravT | let me dump the data | 20:53 |
| lakshmiS | has to be very recent in that case. my devstack is 2 days old | 20:53 |
| lakshmiS | TravT: ui patch 300107 looks good. when you use date fields, i guess it only checks if its a number and not necessarily any date format? | 20:56 |
| TravT | there is something wrong with my pycharms i think. | 20:57 |
| TravT | it is not using the code i see in my editor | 20:57 |
| TravT | will restart it | 20:57 |
| sjmc7 | i’m only getting project id, not tenant | 20:58 |
| lakshmiS | in event payload? | 21:00 |
| sjmc7 | indexed | 21:00 |
| lakshmiS | thats correct | 21:01 |
| sjmc7 | yeah | 21:01 |
| lakshmiS | api gives project_id and event gives tenant_id | 21:01 |
| TravT | lakshmiS: the date formatting for the ui patch doesn't do anything special | 21:01 |
| TravT | it allows same syntax ES allows for query strings. | 21:02 |
| lakshmiS | TravT: checked with date =0 and no error. is there a date format to be followed? | 21:02 |
| sjmc7 | wait, what? we get both? | 21:02 |
| TravT | okay, if API provides project_id, then the line i mentioned about would be a problem, right? | 21:03 |
| TravT | well, maybe not | 21:03 |
| sjmc7 | we may need to alter the serialization functions to write both | 21:03 |
| TravT | def _serialize(self, obj): | 21:03 |
| TravT | obj['project_id'] = obj.pop('tenant_id') | 21:03 |
| sjmc7 | no… | 21:04 |
| sjmc7 | i can do that as a separate patch in a sec | 21:04 |
| sjmc7 | i will say though that i’m not seeing tenant_id anywhere | 21:05 |
| lakshmiS | all my notification event payloads has only tenant_id. no project_id | 21:06 |
| sjmc7 | my devstack was up to date as of march 23rd | 21:06 |
| lakshmiS | http://paste.openstack.org/show/492669/ | 21:06 |
| lakshmiS | mine is 2 days old | 21:06 |
| sjmc7 | urgh, maybe v1 is tenant_id? | 21:07 |
| TravT | which means it is newton... | 21:07 |
| TravT | but i don't see any patches on designate that mention project id | 21:07 |
| lakshmiS | nope even v2 is tenant_id | 21:07 |
| sjmc7 | we’re not directly indexing v1 records though, we’re going to the api, so that’s fine | 21:07 |
| sjmc7 | argh | 21:07 |
| lakshmiS | see the link | 21:07 |
| sjmc7 | wait, we are intentionally turning it into project id | 21:08 |
| sjmc7 | i’m really confused now | 21:08 |
| sjmc7 | TravT: are you having a problem searching? | 21:09 |
| sjmc7 | or did you just look at the raw notification payload? | 21:09 |
| TravT | when i do search, i don't see a project id | 21:09 |
| TravT | and the only way i see the created recordset is to turn on all_projects | 21:10 |
| lakshmiS | here's what i think | 21:11 |
| TravT | but something is really horky with my env | 21:11 |
| TravT | i'm putting log statements in that don't show | 21:11 |
| lakshmiS | for initial indexing, api provides project_id | 21:11 |
| TravT | and debug doesn't hit breakpoint | 21:11 |
| TravT | but i don't see extra processes | 21:11 |
| lakshmiS | and https://github.com/openstack/searchlight/blob/master/searchlight/elasticsearch/plugins/designate/recordsets.py#L93 saves it | 21:11 |
| sjmc7 | api gives project id, events we turn teannt id into project it | 21:11 |
| sjmc7 | one sec, testing it myself | 21:11 |
| sjmc7 | i am seeing records without all_projects | 21:12 |
| lakshmiS | one issue i see is tenant_id is not set when we create recordset for v1 record create event | 21:13 |
| sjmc7 | we’re going to the api for create | 21:14 |
| lakshmiS | ah yes. enough on this now ;) | 21:14 |
| TravT | lakshmiS: which type has a facet named date? | 21:14 |
| lakshmiS | TravT: i was checking just the query for designate created_at date field | 21:15 |
| TravT | so, that last patch only changes magic search facets | 21:15 |
| TravT | if you are mucking around with full text search | 21:16 |
| TravT | that's unrelated | 21:16 |
| lakshmiS | yeah was a side question | 21:16 |
| lakshmiS | :) | 21:16 |
| TravT | date you can do things like | 21:16 |
| TravT | date: [now TO now-1m/d] | 21:17 |
| sjmc7 | i think we need to pull the trigger on these or decide not to; we are running the risk of not getting the release tagged at all | 21:17 |
| lakshmiS | i need to patch it for another thing i found. parent linking was not happening for v1 api | 21:18 |
| TravT | lakshmiS: if you don't see anything else with this | 21:18 |
| TravT | https://review.openstack.org/#/c/300107/ | 21:18 |
| TravT | can you pull the trigger? | 21:18 |
| lakshmiS | sure | 21:18 |
| TravT | i do need to submit the release tag | 21:18 |
| TravT | sjmc7: so you don't see issues with project id? | 21:19 |
| sjmc7 | no. | 21:19 |
| TravT | i don't trust my env right now. | 21:19 |
| sjmc7 | and it’s been like that for ages | 21:19 |
| TravT | i don't see any phantom python processes | 21:19 |
| sjmc7 | the notifications do not include project_id, that is true | 21:19 |
| TravT | but it definitely is not running my code at the moment | 21:19 |
| sjmc7 | but the code makes the switcheroo | 21:19 |
| sjmc7 | good catch on the parent lakshmiS :( | 21:20 |
| sjmc7 | i’ll wait for an updated patch | 21:20 |
| openstackgerrit | Merged openstack/searchlight-ui: Fix issues when magic search facet uses range queries https://review.openstack.org/300107 | 21:25 |
| sjmc7 | that’s everything for the ui repo, TravT | 21:25 |
| TravT | yep, going to submit release request now | 21:26 |
| TravT | i killed off all my pycharms | 21:26 |
| TravT | and it is now re-indexing itself | 21:26 |
| TravT | so hoping designate is happier afterwards | 21:26 |
| TravT | https://review.openstack.org/#/c/300222/ | 21:29 |
| TravT | sjmc7^ | 21:29 |
| sjmc7 | yay! | 21:30 |
| TravT | \o/ | 21:30 |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 21:34 |
| sjmc7 | lakshmiS: doesn’t the parent id come from the parent_field? | 21:37 |
| lakshmiS | i was just thinking about that. | 21:38 |
| lakshmiS | why is it even mapped currenlty | 21:38 |
| sjmc7 | ? | 21:38 |
| lakshmiS | https://github.com/openstack/searchlight/blob/master/searchlight/elasticsearch/plugins/designate/recordsets.py#L100 is that required? | 21:38 |
| sjmc7 | no, i don’t think so | 21:39 |
| lakshmiS | hmm, i thought i missed it so added same :) | 21:39 |
| lakshmiS | so no effect | 21:40 |
| sjmc7 | if adding it does no harm then i’m going to +2 it , we’re really out of time | 21:40 |
| lakshmiS | yup | 21:40 |
| sjmc7 | we’ve got to revert it :( | 21:45 |
| *** TravT has quit IRC | 21:46 | |
| lakshmiS | so we have to remove it from indexing too then? | 21:47 |
| sjmc7 | checking | 21:47 |
| *** TravT has joined #openstack-searchlight | 21:48 | |
| sjmc7 | lakshmiS: looks like it’s leaving junk there too, might as well take it out | 21:48 |
| lakshmiS | yeah its there in source | 21:49 |
| lakshmiS | patch coming through... | 21:49 |
| sjmc7 | ok | 21:49 |
| sjmc7 | i’ll do a quick sanity test then i’ll +2 | 21:49 |
| openstackgerrit | Lakshmi N Sampath proposed openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 21:50 |
| lakshmiS | ok. i really have to leave now. will be back in 15 mins. | 21:50 |
| TravT | i'm pulling it as well | 21:50 |
| sjmc7 | ok, i’m comfortable with it | 21:53 |
| TravT | let's do it! | 21:54 |
| sjmc7 | i’ve +2ed, can cherypick it once it merges | 21:54 |
| TravT | are there any missing cherry picks... | 21:55 |
| sjmc7 | checking | 21:55 |
| sjmc7 | although honestly it’s a little hard to tell | 21:55 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 21:56 | |
| TravT | will try to correlate from here | 21:57 |
| sjmc7 | one missing, i think | 21:57 |
| sjmc7 | https://review.openstack.org/#/c/298830/ | 21:58 |
| sjmc7 | picked it to https://review.openstack.org/#/c/300228/ | 21:59 |
| sjmc7 | looks like that’s everything | 21:59 |
| lakshmiS | back | 22:07 |
| openstackgerrit | Merged openstack/searchlight: Backward compatibility with designate v1 api https://review.openstack.org/299733 | 22:08 |
| *** akanksha_ has joined #openstack-searchlight | 22:10 | |
| TravT | ok, i'm double checking this | 22:10 |
| TravT | https://launchpad.net/searchlight/+milestone/mitaka-rc2 | 22:10 |
| TravT | once the cherry picks land, can you guys also kick off a devstack on stable/mitaka? | 22:11 |
| TravT | just make sure we don't have any surprises | 22:11 |
| sjmc7 | ok, but i have to go quite soon | 22:12 |
| lakshmiS | will check it ou | 22:12 |
| lakshmiS | need to get tempest working soon | 22:14 |
| TravT | ok, everything listed in rc2 has backports... just waiting for designate | 22:15 |
| TravT | to finish gate jobs | 22:15 |
| openstackgerrit | Rick Aulino proposed openstack/searchlight: Zero Downtime Re-indexing Unit Tests. https://review.openstack.org/300203 | 22:16 |
| TravT | gate jobs done. | 22:17 |
| lakshmiS | ok installing now | 22:19 |
| *** bpokorny_ has joined #openstack-searchlight | 22:28 | |
| *** bpokorny_ has quit IRC | 22:30 | |
| *** bpokorny_ has joined #openstack-searchlight | 22:30 | |
| *** bpokorny has quit IRC | 22:32 | |
| sjmc7 | tested initial indexing, some searches, created some stuff and seen it appear | 22:32 |
| sjmc7 | i’m happy | 22:32 |
| sjmc7 | i mean, not happy, but i think it works | 22:32 |
| TravT | cool | 22:32 |
| TravT | i'm still watching it scroll | 22:32 |
| TravT | i have the release request ready to submit | 22:33 |
| sjmc7 | i had to cheat a bit since i’m already late, but i’m running the mitaka code | 22:33 |
| sjmc7 | against mitaka services | 22:33 |
| TravT | Well, fantastic work sjmc7! | 22:33 |
| TravT | it has been quite a haul | 22:34 |
| TravT | i went ahead and pulled the trigger on the release request patch. at least get it up there. | 22:36 |
| TravT | https://review.openstack.org/#/c/300235/ | 22:36 |
| TravT | i could put a WIP -1 until we finish this devstack verification | 22:36 |
| sjmc7 | ok, i have to scoot. i’ll be back a later for a bit | 22:38 |
| TravT | cool | 22:38 |
| TravT | i have to go help coach son's football in 20 minutes as well | 22:38 |
| openstackgerrit | Travis Tripp proposed openstack/searchlight: Simple Script for Generating Resources https://review.openstack.org/230697 | 22:42 |
| *** bpokorny_ has quit IRC | 22:44 | |
| *** bpokorny has joined #openstack-searchlight | 22:44 | |
| TravT | lakshmiS | 22:49 |
| TravT | i just hit this error again: | 22:49 |
| TravT | 2016-03-31 22:49:28.603 | ovs-ofctl: br-tun is not a bridge or a socket | 22:49 |
| TravT | 2016-03-31 22:49:28.626 | ovs-ofctl: br-tun is not a bridge or a socket | 22:49 |
| TravT | in devstack (not a searchlight error) | 22:50 |
| TravT | how did you fix that? | 22:50 |
| lakshmiS | didnt really fix it. removed nuetron | 22:50 |
| TravT | oh geez | 22:50 |
| lakshmiS | but i didnt get that error 2 days back when i installed | 22:50 |
| lakshmiS | so looks like its broken again then | 22:51 |
| lakshmiS | my stack is still installing. lets see if i hit it today | 22:51 |
| TravT | ok. | 22:51 |
| TravT | well, i actually have to leave | 22:51 |
| TravT | i'll check back in later | 22:51 |
| lakshmiS | ok | 22:52 |
| *** bpokorny_ has joined #openstack-searchlight | 23:41 | |
| *** bpokorny has quit IRC | 23:41 | |
| *** bpokorny_ has quit IRC | 23:41 | |
| *** bpokorny has joined #openstack-searchlight | 23:42 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!