| *** paul_glass has quit IRC | 00:06 | |
| *** MentalRay has joined #openstack-dns | 00:07 | |
| *** ericpeterson has quit IRC | 00:17 | |
| *** km has joined #openstack-dns | 00:43 | |
| *** EricGonczer_ has quit IRC | 01:02 | |
| *** ericpeterson has joined #openstack-dns | 01:19 | |
| *** ericpeterson has quit IRC | 01:20 | |
| *** ericpeterson has joined #openstack-dns | 01:20 | |
| *** stanzgy has joined #openstack-dns | 01:38 | |
| *** ericpeterson has quit IRC | 01:57 | |
| *** ericpeterson has joined #openstack-dns | 01:58 | |
| *** stanzgy has quit IRC | 02:01 | |
| *** ericpeterson has quit IRC | 02:23 | |
| *** stanzgy has joined #openstack-dns | 02:27 | |
| *** ericpeterson has joined #openstack-dns | 02:28 | |
| *** ericpeterson has quit IRC | 02:52 | |
| *** richm has quit IRC | 03:06 | |
| *** johnbelamaric has quit IRC | 03:10 | |
| *** kei_yama has quit IRC | 03:49 | |
| *** ericpeterson has joined #openstack-dns | 04:23 | |
| *** crc32 has joined #openstack-dns | 04:27 | |
| *** ericpeterson has quit IRC | 04:27 | |
| *** crc32 has quit IRC | 05:33 | |
| *** ericpeterson has joined #openstack-dns | 06:12 | |
| *** ericpeterson has quit IRC | 06:16 | |
| *** MentalRay has quit IRC | 06:24 | |
| *** jordanP has joined #openstack-dns | 07:24 | |
| *** kodoku has joined #openstack-dns | 07:58 | |
| kodoku | Hi, Requierment.txt for kilo forget pymysql | 07:59 |
|---|---|---|
| *** ericpeterson has joined #openstack-dns | 08:01 | |
| kodoku | And what is pool-target-id when I try powerdns sync ?? | 08:02 |
| kodoku | Kiall I see in irc log you talk about that | 08:04 |
| *** ericpeterson has quit IRC | 08:06 | |
| kodoku | ping :) | 08:25 |
| *** kodoku_ has joined #openstack-dns | 08:37 | |
| *** kodoku has quit IRC | 08:38 | |
| *** jordan_ has joined #openstack-dns | 08:43 | |
| *** jordanP has quit IRC | 08:43 | |
| kodoku_ | omg need docs for pool ^^ | 09:00 |
| *** kodoku has joined #openstack-dns | 09:38 | |
| *** kodoku_ has quit IRC | 09:40 | |
| *** kodoku has quit IRC | 09:46 | |
| *** ericpeterson has joined #openstack-dns | 09:49 | |
| *** kodoku has joined #openstack-dns | 09:50 | |
| *** ericpeterson has quit IRC | 09:55 | |
| *** untriaged-bot has joined #openstack-dns | 10:01 | |
| untriaged-bot | Untriaged bugs so far: | 10:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1442221 | 10:01 |
| openstack | Launchpad bug 1442221 in Designate "Pool manager does not update domain create errors correctly" [Undecided,In progress] - Assigned to Vinod Mangalpally (vinod-mang) | 10:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1442141 | 10:01 |
| openstack | Launchpad bug 1442141 in Designate "designate-api Throws Error on stop()" [Undecided,New] | 10:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1437699 | 10:01 |
| openstack | Launchpad bug 1437699 in Designate "mDNS should Handle Bad File Descriptor" [Medium,New] - Assigned to Ron Rickard (rjrjr) | 10:01 |
| *** untriaged-bot has quit IRC | 10:01 | |
| *** eandersson has joined #openstack-dns | 10:37 | |
| eandersson | edit_managed_records is that for supporting managed records from the api? | 10:39 |
| eandersson | * Added policy check (edit_managed_records), defaulting to rule:admi | 10:39 |
| mugsie | eandersson: yup | 10:46 |
| mugsie | there is a HTTP header you send in with the request | 10:46 |
| mugsie | its not in the client yet | 10:46 |
| mugsie | but if you use curl you are golden | 10:47 |
| eandersson | This should really be re-factored :P https://github.com/openstack/designate/blob/master/designate/notification_handler/base.py | 10:48 |
| eandersson | the _create and _delete functions | 10:48 |
| eandersson | shouldn't use anything directly fromt he configuration :P | 10:48 |
| mugsie | huh | 10:49 |
| mugsie | no we definitly should not | 10:49 |
| eandersson | In the BaseAddressHandler | 10:49 |
| mugsie | -_- | 10:49 |
| mugsie | patches accepted! :D | 10:49 |
| eandersson | =] | 10:49 |
| eandersson | Soon! | 10:49 |
| mugsie | :) | 10:50 |
| eandersson | in _delete | 10:51 |
| eandersson | why would resource_id ever be None? | 10:51 |
| eandersson | or could criterion contain the values needed? | 10:51 |
| eandersson | if so wouldn't resource_id None override that? | 10:52 |
| mugsie | yup | 10:52 |
| mugsie | honestly that area of the code does need a good going over | 10:52 |
| eandersson | I might as well fix it there instead of fixing my code again :p | 10:53 |
| mugsie | I dont use it regularly (if at all) | 10:53 |
| openstackgerrit | Endre Karlson proposed openstack/designate: Switch to API v2 as the new CURRENT https://review.openstack.org/172375 | 11:01 |
| *** kodoku has quit IRC | 11:05 | |
| *** ericpeterson has joined #openstack-dns | 11:06 | |
| *** kodoku has joined #openstack-dns | 11:06 | |
| kodoku | eandersson do you know how sync powerdns database with kilo designate ? | 11:07 |
| kodoku | with pool concept :/ | 11:07 |
| *** ericpeterson has quit IRC | 11:11 | |
| *** eandersson has quit IRC | 11:13 | |
| *** stanzgy has quit IRC | 11:14 | |
| *** eandersson has joined #openstack-dns | 11:20 | |
| kodoku | eandersson do you know how sync powerdns database with kilo designate ? with pool concept. Because if I make "designate-manage powerdns sync" ==> usage: designate powerdns sync [-h] pool-target-id | 11:24 |
| mugsie | kodoku: it should be the pool-target-id in the config file | 11:26 |
| kodoku | And when I use designate-manage I have always => No handlers could be found for logger "oslo_config.cfg" (Maybe eventlet bug ?) | 11:26 |
| mugsie | where you configured the db connection string | 11:26 |
| mugsie | you should have a section like [pool_target:f26e0b32-736f-4f0a-831b-039a415c481e] | 11:27 |
| openstackgerrit | Erik Olof Gunnar Andersson proposed openstack/designate: Refactored BaseAddressHandler. https://review.openstack.org/172382 | 11:27 |
| mugsie | thats where to DB connection string is | 11:27 |
| mugsie | so take the UUID from the section title | 11:27 |
| mugsie | and use that in the command | 11:27 |
| kodoku | ok in [pool_manager_cache:memcache] | 11:28 |
| kodoku | ha ok i see | 11:28 |
| kodoku | I configure here powerdns ? | 11:28 |
| mugsie | yeah | 11:29 |
| kodoku | ok but I need [backend:powerdns] too ? | 11:29 |
| mugsie | nope | 11:29 |
| mugsie | not anymore | 11:30 |
| kodoku | ok but in sample I don't see how I write database pdns | 11:30 |
| mugsie | it would be .... one sec | 11:30 |
| kodoku | I have just type =powerdns port=53 host=localhost | 11:30 |
| kodoku | I have option=.... but it's for bind | 11:31 |
| mugsie | options = connect: <connection string> | 11:31 |
| *** EricGonczer_ has joined #openstack-dns | 11:32 | |
| mugsie | sorry, options = connection : string | 11:32 |
| mugsie | you will also need to fill in the IP of your miniDNS server in the masters section | 11:32 |
| mugsie | IP:PORT actually | 11:32 |
| kodoku | pdns is on designate host | 11:33 |
| kodoku | I don't have miniDNS O_o | 11:33 |
| mugsie | you need to run a designate-mdns as well, to get data from Designate -> PDNS | 11:34 |
| mugsie | (actual records etc) | 11:34 |
| *** EricGonczer_ has quit IRC | 11:35 | |
| kodoku | ok for recap | 11:35 |
| *** kodoku_ has joined #openstack-dns | 11:37 | |
| kodoku_ | I conf [pool:79....] + [pool_nameserver:0f.....] + [pool_target:f2222], I launch "manage powerdns sync f2222" + "manage database sync" ==> then i launch central, mdns, api + sink | 11:37 |
| mugsie | yup | 11:38 |
| kodoku_ | ok thx | 11:38 |
| kodoku_ | I'll try that | 11:38 |
| kodoku_ | and for "No handlers could be found for logger "oslo_config.cfg"" ? | 11:38 |
| mugsie | not sure - think that my just be a long outstanding bug | 11:40 |
| *** kodoku has quit IRC | 11:40 | |
| mugsie | should not cause any issues | 11:40 |
| kodoku_ | ok | 11:40 |
| *** mwagner_lap has quit IRC | 11:42 | |
| kodoku_ | mugsie ok works for add records with api but pdns database always empty | 11:51 |
| kodoku_ | so pdns send can't find :/ | 11:51 |
| kodoku_ | records and domains table empty :/ | 11:51 |
| kodoku_ | no error or trace with mdns | 11:51 |
| mugsie | ah, crap | 11:54 |
| kodoku_ | hum wait | 11:54 |
| mugsie | forgot one service - pool-manager | 11:54 |
| kodoku_ | msdn has shutdown | 11:54 |
| kodoku_ | arf | 11:54 |
| mugsie | designate-pool-manager as well | 11:54 |
| kodoku_ | hum | 11:54 |
| kodoku_ | trace | 11:54 |
| kodoku_ | ValueError: Invalid target type: None | 11:55 |
| kodoku_ | need to uncomment [service:pool_manager] ? | 11:56 |
| mugsie | eh, i think so | 11:58 |
| *** EricGonczer_ has joined #openstack-dns | 12:01 | |
| *** EricGonczer_ has quit IRC | 12:01 | |
| kodoku_ | mugsie need cache driver and configure memcache ? | 12:02 |
| mugsie | kodoku_: no, it will do a simple in memory cache if there is no memcache | 12:03 |
| kodoku_ | ok so no need [pool_manager_cache:sqlalchemy] ? | 12:04 |
| kodoku_ | mugsie this is my designate.conf http://pastebin.com/03eickp4 | 12:07 |
| *** ericpeterson has joined #openstack-dns | 12:07 | |
| kodoku_ | maybe you can see if there is wrong conf | 12:07 |
| mugsie | you are missing master in the pool target section | 12:12 |
| mugsie | that should be ip / hostname of miniDNS + port (usually <ip>:5354 ) | 12:13 |
| kodoku_ | ok | 12:13 |
| kodoku_ | always ValueError: Invalid target type: None when I start pool-manager | 12:13 |
| kodoku_ | maybe workers in [service:pool_manager] | 12:16 |
| kodoku_ | mugsie like need workers = 1 ? | 12:17 |
| *** EricGonczer_ has joined #openstack-dns | 12:21 | |
| kodoku_ | hum | 12:22 |
| kodoku_ | ValueError("Invalid target type: %r" % target) | 12:22 |
| kodoku_ | maybe in target def type = powerdns is wrong | 12:22 |
| kodoku_ | maybe type= pdns ? | 12:22 |
| eandersson | Anyone seen this before when using the sink? Remote error: DBError badly formed hexadecimal UUID string (original cause: ValueError: badly formed hexadecimal UUID string) | 12:29 |
| mugsie | kodoku_: humm, let me look - do you have a full stack trace? | 12:30 |
| mugsie | eandersson: yeah - the UUID is not being coverted properly - where are you getting that? | 12:31 |
| *** ericpeterson has quit IRC | 12:31 | |
| eandersson | mugsie: http://paste.openstack.org/show/bsorsUWRjG6kSxQV5pOV/ | 12:33 |
| eandersson | hmm return self.central_api.get_domain(context, domain_id) I think | 12:34 |
| mugsie | looks like the domain_id isnt getting there | 12:34 |
| mugsie | or, it might have the '-' stripped - I have seen it give out about that before | 12:35 |
| *** bustersw_ has joined #openstack-dns | 12:35 | |
| eandersson | maybe I messed up somewhere :P | 12:35 |
| *** EricGonczer_ has quit IRC | 12:35 | |
| eandersson | will add more logging =] | 12:35 |
| kodoku_ | OK I succes to launch pool-manager | 12:35 |
| kodoku_ | i add database for pool-manager | 12:36 |
| kodoku_ | and configure [pool_manager_cache:sqlalchemy] | 12:36 |
| kodoku_ | but when I add records | 12:36 |
| kodoku_ | TRACE : UnexpectedSource: got a response from ('10.121.141.43', 53) instead of (u'juno008.gate', 53) | 12:36 |
| kodoku_ | so need IP and no HOSTNAME in designate.conf ?? | 12:37 |
| mugsie | yeah, unless you have reverse DNS setup, ip is preferable | 12:38 |
| kodoku_ | ok | 12:38 |
| Kiall | actually, dnspython, of all things, behaves TERRIBLY with hostnames. lol | 12:40 |
| Kiall | IPs even with rdns setup should be used :( | 12:40 |
| kodoku_ | ok | 12:41 |
| kodoku_ | I drop and recreate all database | 12:41 |
| kodoku_ | start all service central, api, mdns, pool-cache | 12:42 |
| kodoku_ | all start witout error or trace | 12:42 |
| kodoku_ | I go to add server, domains and records :) | 12:42 |
| kodoku_ | Yes Baby ! | 12:44 |
| *** km has quit IRC | 12:45 | |
| kodoku_ | Kiall mugsie arf, domain is add to domains tables of pdns but when I add record ==> no add to pdns table | 12:46 |
| Kiall | kodoku_: Oh, pdns config.. did you set master=yes ? | 12:47 |
| Kiall | if not, restart it, it should fill it out | 12:47 |
| kodoku_ | hum in options ? | 12:48 |
| kodoku_ | In pool_target I have masters = 10.121.141.43:5354 | 12:48 |
| kodoku_ | arf in pdns config | 12:48 |
| kodoku_ | daemon=yes master=yes | 12:48 |
| Kiall | Okay, and pdns logs should have something about trying to AXFR the zone? | 12:49 |
| kodoku_ | pdns[31304]: AXFR done for 'test.domain', zone committed with serial number 1428669810 | 12:49 |
| kodoku_ | I have 5 times => pdns[31304]: Received NOTIFY for test.domain from 10.121.141.43 but slave support is disabled in the configuration | 12:51 |
| Kiall | Dooh .. Another pdns config option is slave = yes | 12:51 |
| kodoku_ | ok I add it | 12:51 |
| Kiall | We do this during DevStack runs: | 12:52 |
| Kiall | https://github.com/openstack/designate/blob/master/contrib/devstack/lib/designate_plugins/backend-powerdns#L67 | 12:52 |
| kodoku_ | ha so master=no ? | 12:52 |
| Kiall | Yea, I mixed up master vs slave :) | 12:52 |
| Kiall | master = yes won't break it, just means more stuff enabled in pdns than would be strictly necessary | 12:52 |
| kodoku_ | but designate write directly in pdns database no ? | 12:53 |
| kodoku_ | If I add records with designate, I can see this records in records tables of pdns database | 12:54 |
| Kiall | In Kilo, we write only to the domains table and allow pDNS to AXFR the rest of the contnet | 12:54 |
| Kiall | content* | 12:54 |
| kodoku_ | ok works, I add a new record and I can see both in tables | 12:54 |
| kodoku_ | ok | 12:55 |
| kodoku_ | i understand now | 12:55 |
| kodoku_ | Kiall and question for sink. In juno version, eandersson give me a fix write by you. | 12:57 |
| kodoku_ | It's for Fix records add when user launch 10-20 instances from openstack | 12:58 |
| kodoku_ | This fix is commit in kilo version ? | 12:58 |
| Kiall | That's fixed in Kilo, if you see similar issues, yell really really loud so we can fix them before re release Kilo final! | 12:58 |
| Kiall | before we release* | 12:58 |
| kodoku_ | Ok I make in production soon so If i see similar issue, I tell you | 13:00 |
| Kiall | :) | 13:03 |
| *** richm has joined #openstack-dns | 13:14 | |
| eandersson | Almost got it working with Kilo now. :D | 13:16 |
| eandersson | The record is created by the sink, but it is not removed. | 13:16 |
| eandersson | I see delete_record: Calling central's delete_record | 13:16 |
| eandersson | but the record is still there | 13:16 |
| kodoku_ | with sink ? | 13:17 |
| eandersson | yep | 13:17 |
| kodoku_ | ho god | 13:17 |
| kodoku_ | no good for me ^^ | 13:17 |
| eandersson | I use a customized plugin for the sink though :P | 13:17 |
| eandersson | so probably wont affect you | 13:17 |
| kodoku_ | and If new instance with same name is create ? | 13:17 |
| kodoku_ | ok :) | 13:17 |
| kodoku_ | I'll try now | 13:17 |
| eandersson | pretty sure it behaves the same way for you as it did in Juno | 13:18 |
| kodoku_ | and I see in designate.conf.sample => notification_topics = notifications for sink | 13:19 |
| eandersson | Yea, you'd probably want to change that to something like designate-notifications | 13:19 |
| kodoku_ | so now, I doesn't need to add notification_topics = notifications-designate in nova.conf | 13:19 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Moved RecordSet lookup to a custom join query https://review.openstack.org/172415 | 13:19 |
| eandersson | Kiall: When deleting records when something is misconfigured shouldn't it fail? I am not getting any feedback on failures. | 13:23 |
| kodoku_ | ok sink see anything :/ | 13:24 |
| Kiall | With sink? It depends on the fail I guess | 13:24 |
| kodoku_ | I have consumer in notification queue from designate host | 13:24 |
| Kiall | Sink can be turned on with 50k existing VMs, so throwing a stacktrace as those get deleted isn't really idea | 13:24 |
| Kiall | ideal* | 13:24 |
| eandersson | I see Queueing notification for dns.record.delete | 13:24 |
| eandersson | I see the policy check etc succeed | 13:24 |
| eandersson | but the record is still there | 13:25 |
| kodoku_ | ok so Sink start without error, I see it in queue like consumer, When I start instance I see in queue my instance but sink make anything ...... | 13:28 |
| Kiall | eandersson: okay, could be a bug then :( | 13:28 |
| Kiall | That said.. I added a review for sink support in devstack, including some exercicses for it, and it workd | 13:28 |
| eandersson | I do see the following "Got Timeout while trying to send 'NOTIFY'" in designate-mdns | 13:28 |
| Kiall | worked | 13:28 |
| *** johnbelamaric has joined #openstack-dns | 13:29 | |
| kodoku_ | and my queue is empty.... so designte ACK the msg no ? | 13:30 |
| eandersson | do you see any incoming? | 13:30 |
| kodoku_ | in quueue ? | 13:30 |
| eandersson | yea | 13:30 |
| kodoku_ | yes | 13:30 |
| kodoku_ | ok works after change notification_topics = notification-designate | 13:32 |
| kodoku_ | with notification_topics = notifications not works :/ | 13:32 |
| kodoku_ | but in nova.conf I have notification_topics = notifications,notification-designate | 13:32 |
| eandersson | You need that ont he nova compute nodes | 13:33 |
| eandersson | did you restart the compute nodes after applying the configuration? | 13:33 |
| eandersson | Kiall: Setting record xxx, serial 1428672832: action DELETE, status ERROR _update_record_status | 13:36 |
| eandersson | I do see that | 13:36 |
| *** csoukup has joined #openstack-dns | 13:37 | |
| eandersson | I guess I have to dig into the new code | 13:37 |
| *** kodoku_ has quit IRC | 13:41 | |
| Kiall | eandersson: that suggests that sink did delete it, but the nameserver acted on the deletion. The NOTIFY timeouts are likely the cause | 13:44 |
| eandersson | ok so if I fix the timeouts it should work? | 13:44 |
| eandersson | I just find it odd that it can create, as the same timeouts should happen there as well. | 13:44 |
| Kiall | Yep, it failed to notify the nameserver of a change, so the nameserver never fetched the update zone, so the change (the delete) was marked as failed after a couple of attempts | 13:45 |
| Kiall | Yea - It is odd, since the code is identical for how those notifys etc get sent... Humm | 13:45 |
| eandersson | Is the actual mysql db removal proxied now? | 13:49 |
| eandersson | Since I would assume that the DB entry would be deleted here: self._delete_record_in_storage | 13:49 |
| eandersson | Oh Kiall | 13:51 |
| eandersson | recordset = self.storage.delete_recordset(context, recordset.id) | 13:51 |
| Kiall | That will mark it as action = DELETE, status = PENDING in our DB, then, once it's cleared from the nameserver, actually deleted | 13:51 |
| eandersson | This exists in the delete_recordset | 13:51 |
| eandersson | but there is no such function in delete_record | 13:51 |
| eandersson | only record = self.storage.update_record(context, record) | 13:51 |
| eandersson | gonna check trunk | 13:51 |
| eandersson | https://github.com/openstack/designate/blob/master/designate/central/service.py#L1565 | 13:52 |
| eandersson | and this is delete_recordset | 13:52 |
| eandersson | https://github.com/openstack/designate/blob/master/designate/central/service.py#L1332 | 13:52 |
| eandersson | Maybe I am missing something? :D | 13:53 |
| eandersson | as in missing something obvious (not missing code :P) | 13:53 |
| Kiall | Humm, that certainly looks like an issue.. But, I'm sure our tests cover those .. Which has me wondering a little | 13:53 |
| Kiall | About to hop onto a call, will check after. | 13:54 |
| *** mwagner_lap has joined #openstack-dns | 13:57 | |
| *** ericpeterson has joined #openstack-dns | 13:58 | |
| *** vinod1 has joined #openstack-dns | 13:59 | |
| *** paul_glass has joined #openstack-dns | 14:14 | |
| *** andrewbogott_afk is now known as andrewbogott | 14:20 | |
| *** dvorak is now known as clayton | 14:21 | |
| eandersson | Kiall: I Added self.storage.delete_record(context, record.id) and it works now. =] | 14:22 |
| elarson | I wrote a little script to create random domain names in designate. http://paste.openstack.org/show/202038/ | 14:29 |
| elarson | one thing to note is it uses the randstr package. I wrote that the other day to create small random strings for this sort of thing | 14:30 |
| elarson | just a friendly FYI | 14:31 |
| *** kodoku has joined #openstack-dns | 14:38 | |
| kodoku | Kiall eandersson mugsie Re, I try to automat deploy of designate with puppet and when puppet launch database sync, I have ==> oslo.config.cfg.NoSuchGroupError: no such group: <oslo_config.cfg.OptGroup object at 0x2ec7650> | 14:39 |
| kodoku | Any Ideas ? | 14:39 |
| elarson | kodoku: that sounds like a config option group is missing | 14:40 |
| kodoku | ok I compare with the first deploy | 14:41 |
| elarson | kodoku: can you tell where in the code the error is coming from? that might help narrow it down | 14:41 |
| * elarson also wonders if best practice is to pass a group object or the name when referencing groups in the config as it looks like object pattern doesn't really help users debug the config. | 14:43 | |
| kodoku | File "/usr/lib/python2.7/site-packages/stevedore/driver.py", line 50, in _default_on_load_failure raise err | 14:44 |
| kodoku | and designate.conf is same of original | 14:44 |
| kodoku | elarson full trace http://pastebin.com/AhxhdY72 | 14:45 |
| kodoku | I have stevedore (1.3.0) | 14:47 |
| *** EricGonczer_ has joined #openstack-dns | 14:49 | |
| *** andrewbogott is now known as andrewbogott_afk | 14:51 | |
| timsim | kodoku: What does the config look like? | 14:52 |
| kodoku | I up designate.conf | 14:53 |
| *** ericpeterson has quit IRC | 14:53 | |
| kodoku | timsim http://pastebin.com/usWD5AbT | 14:54 |
| timsim | Usually when I see that error, it spells out a specific config section it's looking for. Not an Oslo config object. | 14:57 |
| kodoku | :/ | 14:59 |
| kodoku | when I launch designate-manage I have this issue | 15:00 |
| kodoku | without option | 15:00 |
| kodoku | even designate-manage --help Traceback :/ | 15:01 |
| elarson | kodoku: if you're running the database sync, do you have a [storage:sqlalchemy] section? | 15:03 |
| elarson | could be a misspelling too | 15:03 |
| kodoku | yes | 15:03 |
| *** andrewbogott_afk is now known as andrewbogott | 15:03 | |
| kodoku | I have link my conf | 15:03 |
| kodoku | http://pastebin.com/usWD5AbT | 15:03 |
| kodoku | But I think is not a conf issue | 15:04 |
| kodoku | Because "designate-manage --help" make a trace | 15:04 |
| kodoku | Maybe a module not install or bug in setup.py install :/ | 15:04 |
| timsim | Yeah I guess try and run setup.py again? And make sure that the config is where it's expecting it | 15:05 |
| kodoku | python setup.py install works without errors | 15:06 |
| kodoku | and same issue | 15:06 |
| timsim | I think you can specify the config directory to designate-manage, try that | 15:06 |
| timsim | Well wait, if you're on Juno, a bunch of that config doesn't need to be there. | 15:06 |
| kodoku | I'am on kilo version | 15:07 |
| timsim | Oh, well nvm then | 15:07 |
| kodoku | I reinstall on fresh system | 15:08 |
| eandersson | Kiall: https://review.openstack.org/#/c/172459/ Let me know if I am completely wrong about this. =] | 15:09 |
| vinod1 | eandersson - that looks right - did you hit this issue with the v1 api? | 15:13 |
| Kiall | I really think the gate jobs are going to fail with that... But just off that call and being pushed to get some more bits+pieces done today before I'm gone Monday :( | 15:13 |
| Kiall | If anything, I delete the delete_recordset was an extra call that should be removed! | 15:14 |
| eandersson | You have the same for create_record though | 15:15 |
| eandersson | ecord = self.storage.create_record(...) | 15:15 |
| Kiall | the actual delete from the DB get's covered from here: https://github.com/openstack/designate/blob/master/designate/central/service.py#L2138 | 15:15 |
| *** MentalRay has joined #openstack-dns | 15:15 | |
| Kiall | (as does the delete_recordset call) | 15:15 |
| eandersson | That is how I thought that it worked at first. | 15:15 |
| eandersson | The delete_recordset code threw me off. | 15:16 |
| eandersson | I really see it as an issue with managed records not being deleted though. | 15:16 |
| eandersson | As the way it is right now, we need to go into the database and delete those everytime. | 15:16 |
| kodoku | arf always same bug T_T | 15:17 |
| eandersson | It should at the very least report something to the sink even if it doesn't throw an exception. | 15:17 |
| Kiall | eandersson: actually, are you using increment_serial = False in your sink handler? | 15:19 |
| eandersson | nope | 15:19 |
| Kiall | Can you throw the logs on paste.openstack.org? I can't look straight away, but I'll try look over the next few hours. | 15:20 |
| eandersson | In the actual DNS creation process my code is almost identical to trunk. | 15:20 |
| Kiall | Oh - custom handler? Okay, what about the stock ones? can you test that out? | 15:20 |
| Kiall | Or is the only change the one you submitted as a refactor? | 15:20 |
| eandersson | Pretty much. | 15:20 |
| kodoku | If you have time to look : http://paste.openstack.org/show/202081/ | 15:21 |
| eandersson | I just re-use BaseAddressHandler with the addition of the refactored code. | 15:21 |
| elarson | kodoku: it looks like oslo.messaging is assuming a config group exists. can you confirm what version you're using? | 15:29 |
| kodoku | elarson oslo.messaging (1.8.0) | 15:32 |
| *** jmcbride has joined #openstack-dns | 15:34 | |
| eandersson | Kiall: I'll set up a local environment tonight or tomorrow and reproduce it. These logs are way too messy. | 15:41 |
| -openstackstatus- NOTICE: gerrit has been restarted to address a hung event stream. change events between 15:00 and 15:43 utc which were lost will need to be rechecked or have approval workflow votes reapplied for zuul to act on them | 15:44 | |
| openstackgerrit | Vinod Mangalpally proposed openstack/designate: Show admin endpoint in list of versions https://review.openstack.org/172488 | 15:46 |
| elarson | kodoku: fwiw, I have oslo.messaging==1.8.1 | 15:47 |
| mugsie | vinod1: I did have admin api in the list but got overrule by Kiall | 15:50 |
| mugsie | not sure why he didn't want it | 15:50 |
| Kiall | Becauage / is a URL you can hit unauthenticated, and we shouldn't expose the /admin to non admins | 15:51 |
| Kiall | Because* | 15:51 |
| vinod1 | but we document the admin endpoint here - http://docs.openstack.org/developer/designate/rest/admin/quotas.html | 15:52 |
| mugsie | Kiall: thats tenous | 15:52 |
| elarson | this was actually confusing me just yesterday as I needed to up some quotas | 15:53 |
| mugsie | because admin apis are documented, and can even be in service catalogs | 15:53 |
| Kiall | vinod1: yea, but users who can't get it shouldn't see it returned as a possible API | 15:53 |
| *** ericpeterson has joined #openstack-dns | 15:53 | |
| Kiall | e.g. every public openstack deployments hides the "adminURL" catalog property from end users.. If we expose it at / .. there all going to go fork it. | 15:54 |
| Kiall | (or just get annoyed and leave it ;)) | 15:54 |
| mugsie | i really dont think they will | 15:56 |
| mugsie | I actually doubt anyone will care. That said it would be very easy to have a policy check there | 15:56 |
| *** ericpeterson has quit IRC | 15:58 | |
| openstackgerrit | Eric Larson proposed openstack/designate: Add the Hook Point API https://review.openstack.org/164748 | 15:59 |
| *** jordan_ has quit IRC | 16:00 | |
| vinod1 | Is something like this policy.check('admin', context, target) valid - if so kiall can we add that as an additional check before displaying the endpoint? | 16:00 |
| elarson | +1 | 16:00 |
| *** untriaged-bot has joined #openstack-dns | 16:01 | |
| untriaged-bot | Untriaged bugs so far: | 16:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1442221 | 16:01 |
| openstack | Launchpad bug 1442221 in Designate "Pool manager does not update domain create errors correctly" [Undecided,In progress] - Assigned to Vinod Mangalpally (vinod-mang) | 16:01 |
| Kiall | vinod1: so, we don't have the KS middleware in front of that endpoint.. | 16:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1442141 | 16:01 |
| openstack | Launchpad bug 1442141 in Designate "designate-api Throws Error on stop()" [Undecided,New] - Assigned to Graham Hayes (grahamhayes) | 16:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1442699 | 16:01 |
| openstack | Launchpad bug 1442699 in Designate "Add admin endpoint to the list of versions" [Undecided,In progress] - Assigned to Vinod Mangalpally (vinod-mang) | 16:01 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1437699 | 16:01 |
| openstack | Launchpad bug 1437699 in Designate "mDNS should Handle Bad File Descriptor" [Medium,New] - Assigned to Ron Rickard (rjrjr) | 16:01 |
| *** untriaged-bot has quit IRC | 16:01 | |
| Kiall | we could add it, with "delay_auth_decision" set to False to unauthenticated users can still hit it | 16:02 |
| Kiall | s/to/so/ | 16:02 |
| *** jmcbride has quit IRC | 16:04 | |
| mugsie | if we are putting in policy there, we should be adding it to all endpoints | 16:04 |
| *** kodoku has quit IRC | 16:04 | |
| mugsie | so people can limit the v1 as well | 16:04 |
| mugsie | or even v2 if they feel like it | 16:04 |
| vinod1 | first off is this issue - (displaying admin in versions) important for rc1? | 16:08 |
| Kiall | mugsie: humm, yea.. | 16:08 |
| Kiall | vinod1: ehh.. I'm not sure. If we have a reasonable way to do it, then it's probably a good thing to do.. | 16:08 |
| Kiall | I think we can just drop the KS middleware in front on the "versions" app in api-paste.ini, the tricky bit is getting only that 1 "app" to use delay_auth_decision=False | 16:09 |
| Kiall | Actually, I think that can be done easily too .. let me see | 16:10 |
| openstackgerrit | Kiall Mac Innes proposed openstack/designate: Throw KS middleware in front of versions app https://review.openstack.org/172496 | 16:15 |
| Kiall | vinod1 / mugsie: in theory .. ^ gives us a context we can work with.. Not sure how the context middleware will handle empty headers etc etc tho | 16:15 |
| *** jmcbride has joined #openstack-dns | 16:15 | |
| Kiall | (I have no intent of that merging, just writing it was easier than explaing it ;)) | 16:16 |
| Kiall | lol .. and delay_auth_decision = False should be set to True ;) | 16:16 |
| *** ericpeterson has joined #openstack-dns | 16:17 | |
| vinod1 | with all these changes - the change will no longer be simple and will require more testing | 16:18 |
| Kiall | ++ | 16:19 |
| *** jmcbride1 has joined #openstack-dns | 16:19 | |
| *** jmcbride has quit IRC | 16:20 | |
| Kiall | "admin endpoint not exposed in versions list for admin users" is a bug, but probably not critical enough to force in now.. and one we can fix post release in the stable branch. | 16:20 |
| Kiall | mugsie: thoughts ^ ? | 16:22 |
| mugsie | yeah. I think we should do it, but this close is asking for trouble | 16:23 |
| vinod1 | kiall you mentioned some other changes before rc1 - what changes would these be? | 16:24 |
| Kiall | The only blocker is making the decision in V2 being stable | 16:24 |
| Kiall | (that I know of) | 16:25 |
| Kiall | https://review.openstack.org/#/c/172375/ <-- review for that is here, I'd like to get "Votes" in the form of +/-2 by EOD Monday! | 16:25 |
| Kiall | (My -2 is just to block in going in for now ;)) | 16:25 |
| eandersson | Kiall: So it is not a bug then? | 16:57 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Moved RecordSet lookup to a custom join query https://review.openstack.org/172415 | 16:57 |
| *** andrewbogott is now known as andrewbogott_afk | 17:01 | |
| *** mwagner_lap has quit IRC | 17:02 | |
| eandersson | I still find it odd that it can create, but not delete. That could lead to a lof of unwated records in the db. =] | 17:02 |
| Kiall | eandersson: re your patch, Gerrit was broke when you submitted it so the tests didn't get triggered.. The "recheck no bug" common is how you get Gerrit to re-trigger the tests :) | 17:10 |
| Kiall | comment* | 17:10 |
| eandersson | ah | 17:11 |
| eandersson | but the tests did indeed fail like you said | 17:11 |
| eandersson | All failing with RecordNotFound | 17:12 |
| Kiall | re "that it can create, but not delete. That could lead to a lof of unwated records in the db.".. It does delete, the SQL DELETE query is just deferred until the record has been purged from the nameservers | 17:13 |
| eandersson | ah ok | 17:13 |
| eandersson | that makes a lot more sense | 17:13 |
| *** jmcbride1 has quit IRC | 17:13 | |
| eandersson | def wai then | 17:13 |
| eandersson | thanks Kiall | 17:13 |
| Kiall | The failing NOTIFY is preventing the nameserver from updating, so it never get's removed | 17:13 |
| Kiall | (or some other issue is preventing it.. That might be a red herring) | 17:14 |
| eandersson | I guess I am just hung up on the lack of support for Managed records in the API. | 17:14 |
| eandersson | As I am afraid of something like this happening in production. | 17:14 |
| Kiall | eandersson: someone.. mugsie maybe? added a review for that yesterday.. I have no clue if we merged it tho.. lol. | 17:14 |
| eandersson | Would be a lot of work to manually go through the databases | 17:14 |
| mugsie | Kiall: afaik we did | 17:14 |
| *** MentalRay has quit IRC | 17:14 | |
| Kiall | (It's been 2 days of ear glued to a phone) | 17:14 |
| Kiall | We did.. https://github.com/openstack/designate/commit/a4f3ad90b9f7d9a98b47c5adcf949d5f45c3ded4 | 17:14 |
| Kiall | ^ let's you, client support is still needed tho | 17:15 |
| eandersson | ah yea saw that code | 17:15 |
| eandersson | that is awesome | 17:15 |
| eandersson | I can use it for my sync script | 17:15 |
| mugsie | eandersson: so you can delete from the API, by adding X-Designate-Edit-Managed-Records:True as a header | 17:15 |
| eandersson | perfect | 17:15 |
| Kiall | Maybe? Depends exactly what you're tying to do! Ie.g. you can't create managed records using it, that would have been a much bigger API change (exposing the managed_* fields so you can set them) | 17:16 |
| mugsie | vinod1 timsim elarson Kiall thoughts on putting https://review.openstack.org/172415 in RC1 ? | 17:16 |
| eandersson | I might have to track deleted records though and periodically follow up to make sure that they have been deleted | 17:16 |
| Kiall | eandersson: SELECT * FROM records WHERE action = 'DELETE' AND satus = 'ERROR'; | 17:17 |
| Kiall | ^ that'll get all the ones that failed to delete | 17:17 |
| timsim | mugsie: I'd like to. But that's a pretty big change. I don't know if that'll fly. Seems like we could probably get it tested properly. | 17:17 |
| eandersson | awh that is nice | 17:17 |
| Kiall | and pool manager will keep trying to delete them | 17:17 |
| vinod1 | just looking at the timings - with the patch v1 is taking more time? | 17:17 |
| eandersson | perfect | 17:17 |
| Kiall | once the underlying issue is fixed, they should get deleted | 17:17 |
| eandersson | sorry a little out of touch with the new changes | 17:17 |
| eandersson | setting up a new dev environment where I can test out the changes properly | 17:17 |
| *** jmcbride has joined #openstack-dns | 17:18 | |
| mugsie | vinod1: nope, the number you are interested in is the last one | 17:18 |
| Kiall | vinod1 / mugsie the numbers in the comment are wrong ;) | 17:18 |
| Kiall | The V1 comment is comparing a limited V1 API call to a unlimited V2 API call ;) | 17:19 |
| mugsie | v1 isnt limted | 17:19 |
| Kiall | scratch that.. | 17:19 |
| Kiall | The V1 comment is comparing an unlimited V1 API call to a limited V2 API call ;) | 17:19 |
| Kiall | time http http://127.0.0.1:9001/v1/domains/3b366791-35f8-42d3-8078-f9672669f04d/records > /dev/null | 17:19 |
| Kiall | vs | 17:19 |
| Kiall | time http http://127.0.0.1:9001/v2/zones/3b366791-35f8-42d3-8078-f9672669f04d/recordsets\?limit\=1700 > /dev/null | 17:19 |
| Kiall | v1 vs v2 | 17:19 |
| mugsie | the limit is there to make sure we got all the records | 17:19 |
| Kiall | no limit, vs limit=1700 | 17:19 |
| mugsie | the deafult limit is there now | 17:20 |
| mugsie | "On Zone with 1700 Records" | 17:20 |
| mugsie | (it was actually a little less) | 17:20 |
| Kiall | Ah.. Okay.. It's still comparing the V1 vs V2 API rather than V1 before, V1 after ;) | 17:20 |
| mugsie | ah | 17:20 |
| mugsie | it is | 17:20 |
| mugsie | huh | 17:20 |
| mugsie | woops | 17:20 |
| Kiall | ;) | 17:20 |
| Kiall | Also .. Is that on your desktop env? You changed that back from SQLite, right? Just in case ;) | 17:21 |
| mugsie | it is my desktop - running MySQL 5.6 | 17:21 |
| mugsie | (yes i did ;) ) | 17:22 |
| mugsie | designate➜ designate git:(master) ✗ time http http://127.0.0.1:9001/v1/domains/3b366791-35f8-42d3-8078-f9672669f04d/records > /dev/null ae1-2:grahamhayes-hp-project1 | 17:22 |
| mugsie | http > /dev/null 0.19s user 0.04s system 2% cpu 9.576 total | 17:22 |
| mugsie | thats without | 17:22 |
| Kiall | K - So the Q is .. rc1 potential? It's certainly something we should fixup in Kilo.. But might be something for 2015.1.1 rather than 2015.1.0 .. Humm.. If we leave it for 2015.1.1, we can likely make it work for all the resources which have related stuff to load.. | 17:29 |
| vinod1 | +1 for 2015.1.1 | 17:32 |
| mugsie | i would err towards .1 - apart from the 55-60% time difference + the reduced load + reduced DB traffic. | 17:33 |
| mugsie | it is a huge amount of time | 17:33 |
| mugsie | and we go from N+1 (where N = number of recordsets) DB calls to 2 | 17:34 |
| mugsie | or 3 if we have a marker (but that would be N+2 right now) | 17:34 |
| vinod1 | mugsie are you saying you want the change in 2015.1.0? | 17:36 |
| mugsie | I would like it. but it would require very in depth review | 17:37 |
| mugsie | (it fixes a bug that was opened against Icehouse) | 17:38 |
| mugsie | sorry, against Juno | 17:38 |
| mugsie | but I can understand why we wouldnt want to merge it | 17:38 |
| *** MentalRay has joined #openstack-dns | 17:41 | |
| *** photocyte has joined #openstack-dns | 17:49 | |
| *** boris-42 has joined #openstack-dns | 17:53 | |
| *** mwagner_lap has joined #openstack-dns | 18:02 | |
| eandersson | If you have two sites, is it possible two have one designate per site and somehow still share a sub-domain? | 18:11 |
| eandersson | or what is the most redundent setup you could do with two sinks? | 18:12 |
| *** andrewbogott_afk is now known as andrewbogott | 18:12 | |
| *** vinod1 has quit IRC | 18:18 | |
| *** vinod1 has joined #openstack-dns | 18:20 | |
| *** vinod1 has quit IRC | 18:21 | |
| *** vinod1 has joined #openstack-dns | 18:21 | |
| *** vinod2 has joined #openstack-dns | 18:25 | |
| *** vinod1 has quit IRC | 18:25 | |
| *** crc32 has joined #openstack-dns | 18:33 | |
| *** photocyte has quit IRC | 18:38 | |
| *** MentalRay has quit IRC | 18:54 | |
| eandersson | have a good weekend guys! | 18:58 |
| *** eandersson has quit IRC | 18:58 | |
| *** johnbelamaric1 has joined #openstack-dns | 19:02 | |
| *** johnbelamaric has quit IRC | 19:02 | |
| *** EricGonc_ has joined #openstack-dns | 19:08 | |
| *** EricGonczer_ has quit IRC | 19:11 | |
| *** MentalRay has joined #openstack-dns | 19:12 | |
| *** johnbelamaric1 has quit IRC | 19:12 | |
| *** crc32 has quit IRC | 19:13 | |
| *** photocyte has joined #openstack-dns | 19:19 | |
| *** crc32 has joined #openstack-dns | 19:20 | |
| *** MentalRay has quit IRC | 19:25 | |
| *** johnbelamaric has joined #openstack-dns | 19:28 | |
| *** crc32 has quit IRC | 19:47 | |
| *** johnbelamaric has quit IRC | 19:48 | |
| *** vinod2 has quit IRC | 19:48 | |
| *** MentalRay has joined #openstack-dns | 19:49 | |
| *** EricGonc_ has quit IRC | 19:51 | |
| *** johnbelamaric has joined #openstack-dns | 19:52 | |
| *** EricGonczer_ has joined #openstack-dns | 19:53 | |
| *** csoukup has quit IRC | 19:56 | |
| *** EricGonc_ has joined #openstack-dns | 20:03 | |
| *** vinod1 has joined #openstack-dns | 20:06 | |
| *** EricGonczer_ has quit IRC | 20:06 | |
| *** johnbelamaric has quit IRC | 20:08 | |
| *** johnbelamaric has joined #openstack-dns | 20:08 | |
| *** boris-42 has quit IRC | 20:38 | |
| openstackgerrit | Eric Larson proposed openstack/designate-specs: Add hook-point-api specification and liberty specs https://review.openstack.org/171689 | 20:39 |
| *** vinod1 has quit IRC | 20:49 | |
| *** MentalRay has quit IRC | 20:51 | |
| *** bustersw_ has quit IRC | 20:53 | |
| openstackgerrit | Eric Larson proposed openstack/designate: Add the Hook Point API https://review.openstack.org/164748 | 21:17 |
| *** jmcbride has quit IRC | 21:17 | |
| *** photocyte has quit IRC | 21:30 | |
| *** johnbelamaric has quit IRC | 21:31 | |
| *** MentalRay has joined #openstack-dns | 21:39 | |
| *** EricGonczer_ has joined #openstack-dns | 21:55 | |
| *** EricGonc_ has quit IRC | 21:57 | |
| *** paul_glass has quit IRC | 22:07 | |
| openstackgerrit | Eric Larson proposed openstack/designate: Add the Hook Point API https://review.openstack.org/164748 | 22:28 |
| *** EricGonczer_ has quit IRC | 22:44 | |
| *** ericpeterson has quit IRC | 22:45 | |
| *** ericpeterson has joined #openstack-dns | 22:56 | |
| *** andrewbogott is now known as andrewbogott_afk | 23:11 | |
| *** ericpeterson has quit IRC | 23:15 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!