| *** pk has joined #openstack-dns | 00:05 | |
| *** pk__ has joined #openstack-dns | 00:07 | |
| *** pk has quit IRC | 00:07 | |
| *** sn6i23a has quit IRC | 00:08 | |
| *** rmoe has quit IRC | 00:09 | |
| *** penick has quit IRC | 00:25 | |
| *** rmoe has joined #openstack-dns | 00:28 | |
| *** penick has joined #openstack-dns | 00:29 | |
| *** MentalRay_ has quit IRC | 00:31 | |
| *** pk__ has quit IRC | 00:31 | |
| *** penick has quit IRC | 00:34 | |
| *** Phibs has joined #openstack-dns | 00:52 | |
| Phibs | Can anyone share with me the .SQL needed to modify the default powerDNS schema (as of current) to work with designate? | 00:52 |
|---|---|---|
| Phibs | also, are there any plans to shift to using the new powerDNS httpAPI (JSON) vs direct database access? | 00:52 |
| *** johnbelamaric has quit IRC | 00:53 | |
| Phibs | more importantly is anyone alive :) | 00:58 |
| *** pk has joined #openstack-dns | 01:04 | |
| *** stanzgy has joined #openstack-dns | 01:39 | |
| *** km has joined #openstack-dns | 01:52 | |
| *** pk has quit IRC | 01:52 | |
| *** MentalRay has joined #openstack-dns | 01:57 | |
| *** penick has joined #openstack-dns | 02:34 | |
| *** richm has quit IRC | 02:38 | |
| *** pk has joined #openstack-dns | 02:59 | |
| *** pk has quit IRC | 03:03 | |
| *** pk has joined #openstack-dns | 03:04 | |
| *** fouxm_ has joined #openstack-dns | 03:27 | |
| *** fouxm has quit IRC | 03:28 | |
| *** pk has quit IRC | 03:35 | |
| *** mikehn has quit IRC | 03:38 | |
| *** mikehn has joined #openstack-dns | 03:38 | |
| *** km_ has joined #openstack-dns | 04:03 | |
| *** km has quit IRC | 04:04 | |
| *** pk has joined #openstack-dns | 04:05 | |
| *** km has joined #openstack-dns | 04:30 | |
| *** km_ has quit IRC | 04:32 | |
| *** andrewbogott_afk is now known as andrewbogott | 04:36 | |
| *** andrewbogott is now known as andrewbogott_afk | 04:39 | |
| *** MentalRay has quit IRC | 04:41 | |
| *** MentalRay has joined #openstack-dns | 04:43 | |
| *** pk has quit IRC | 05:09 | |
| *** penick has quit IRC | 05:47 | |
| *** pk has joined #openstack-dns | 06:13 | |
| *** pk has quit IRC | 06:17 | |
| Phibs | anyone alive ? | 06:27 |
| *** MentalRay has quit IRC | 06:27 | |
| Phibs | http://docs.openstack.org/developer/designate/getting-started.html | 06:36 |
| Phibs | doesn't even work :( | 06:36 |
| Phibs | python deps barf | 06:36 |
| stanzgy | Phibs: hi phibs, what's the error message? | 06:51 |
| Phibs | well i've gotten further now with latest pip | 06:51 |
| Phibs | but now I have | 06:52 |
| Phibs | https://www.bsd-unix.net/seitz/screenshots/2015-03-25_0252.png | 06:52 |
| Phibs | (I really only want to be able to upgrade my new powerdns installed schema to support designate) | 06:52 |
| Phibs | so if there is an easier way I'm all for it | 06:52 |
| Phibs | pbr (0.9.0) | 06:53 |
| stanzgy | what the code branch you are using? | 06:54 |
| Phibs | juno stable | 06:54 |
| stanzgy | Maybe you can try the latest master branch if you just wanna a try | 06:55 |
| Phibs | sure | 06:56 |
| Phibs | this one gets to | 06:58 |
| Phibs | distutils.errors.DistutilsError: Could not find suitable distribution for Requirement.parse('traceback2') | 06:58 |
| Phibs | heh | 06:59 |
| stanzgy | oh, let me have a check | 07:00 |
| Phibs | getting a lot of missing stuff but pip install by hand works | 07:01 |
| Phibs | k finished that time ;0 | 07:02 |
| stanzgy | It works good in my environment, maybe you get some thing wrong with your pip/virtualenv | 07:04 |
| Phibs | hmmm | 07:05 |
| Phibs | this still fails: python setup.py develop | 07:05 |
| Phibs | ERROR:root:Error parsing | 07:05 |
| Phibs | hmm | 07:06 |
| stanzgy | do you run 'pip install -r test-requirements.txt -r requirements.txt' successfully? | 07:06 |
| Phibs | yea | 07:06 |
| Phibs | ok | 07:08 |
| Phibs | its working now ;) | 07:08 |
| Phibs | maaaany python deps ;0 | 07:08 |
| stanzgy | :D | 07:10 |
| Phibs | will it update an existing schema for me and not change what's there ? | 07:13 |
| Phibs | b/c I used what is here: https://doc.powerdns.com/md/authoritative/backend-generic-mypgsql/ | 07:13 |
| stanzgy | If you use designate-manage command to manage your db schema before, there will be a migrate_version table in your pdns db and it should update your exsiting to latest. | 07:16 |
| Phibs | hmm | 07:17 |
| Phibs | do you know if openstack plans on supporting the powerdns http API vs direct database ? | 07:17 |
| stanzgy | designate-manage decides which scripts to run according to your current db migration version. | 07:18 |
| stanzgy | direct database | 07:18 |
| Phibs | right, so if it was never run before, but I've created the powerdns schema already... | 07:18 |
| Phibs | Would be nice if designate used https://doc.powerdns.com/md/httpapi/api_spec/ | 07:19 |
| stanzgy | AFAIK there is no plan to support pdns http api | 07:20 |
| Phibs | this is bad then :) | 07:20 |
| Phibs | I wonder if my company can write something | 07:20 |
| Phibs | what is your relation to the designate project? | 07:21 |
| stanzgy | you can write a wishlist or a blueprint and the community will discuss about it | 07:21 |
| Phibs | nod | 07:22 |
| Phibs | thanks for your help | 07:22 |
| stanzgy | I'm a openstack developer and new to designate too :) | 07:22 |
| Phibs | hehe | 07:23 |
| Phibs | I will see what we can do, I think using an API is much better | 07:23 |
| Phibs | esp since that allows you to use any powerdns backend, and designate does not know :) | 07:23 |
| Phibs | thanks again, tis late here :) | 07:24 |
| stanzgy | you are welcome | 07:24 |
| *** chlong has quit IRC | 07:52 | |
| *** jordanP has joined #openstack-dns | 07:59 | |
| *** pk has joined #openstack-dns | 08:01 | |
| *** pk has quit IRC | 08:05 | |
| ekarlso | heya guys | 08:19 |
| ekarlso | Phibs: what's uå ? | 08:21 |
| *** MentalRay has joined #openstack-dns | 08:26 | |
| *** km has quit IRC | 08:45 | |
| *** MentalRay has quit IRC | 08:47 | |
| *** pk has joined #openstack-dns | 09:50 | |
| *** pk has quit IRC | 09:54 | |
| openstackgerrit | Cyril Roelandt proposed openstack/python-designateclient: Enable Python 3 https://review.openstack.org/167552 | 10:42 |
| *** stanzgy has quit IRC | 10:54 | |
| *** untriaged-bot has joined #openstack-dns | 11:02 | |
| untriaged-bot | Untriaged bugs so far: | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1434189 | 11:02 |
| openstack | Launchpad bug 1434189 in Designate "Delete zone process can be interrupted" [Undecided,In progress] - Assigned to Tim Simmons (tim-simmons-t) | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432842 | 11:02 |
| openstack | Launchpad bug 1432842 in Designate "Property setters dont work." [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432850 | 11:02 |
| openstack | Launchpad bug 1432850 in Designate "Add periodic refresh task" [High,New] - Assigned to Endre Karlson (endre-karlson) | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1425668 | 11:02 |
| openstack | Launchpad bug 1425668 in Designate "Poor error message when using same database for designate and the pool manager cache" [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432839 | 11:02 |
| openstack | Launchpad bug 1432839 in Designate "Calling find_domain in any v1 operation is sub-optimal" [Medium,New] - Assigned to Endre Karlson (endre-karlson) | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1436224 | 11:02 |
| openstack | Launchpad bug 1436224 in Designate "Kilo beta 3 unit test errors in Debian: AttributeError: 'ValidationError' object has no attribute 'relative_path'" [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1436210 | 11:02 |
| openstack | Launchpad bug 1436210 in Designate "Multiple serial update notifications are not buffered " [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1433645 | 11:02 |
| openstack | Launchpad bug 1433645 in Designate "Floating IP Reverse DNS API does not respect async operations" [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1430976 | 11:02 |
| openstack | Launchpad bug 1430976 in Designate "Pool Manager creates/deletes can go into an infinite loop" [Undecided,Fix committed] - Assigned to Tim Simmons (tim-simmons-t) | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1436261 | 11:02 |
| openstack | Launchpad bug 1436261 in Designate "Requires jsonschema >= 2.4.0 (not 2.2.0)" [Undecided,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1435583 | 11:02 |
| openstack | Launchpad bug 1435583 in Designate "designatedashboard packaging is not very simple" [Undecided,In progress] - Assigned to Eric Peterson (ericpeterson-l) | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1425117 | 11:02 |
| openstack | Launchpad bug 1425117 in Designate "Designate does not work with postgres" [High,New] | 11:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1435568 | 11:02 |
| openstack | Launchpad bug 1435568 in Designate "pep8 gate taking longer than 30 mins on occasion" [Undecided,New] | 11:02 |
| *** untriaged-bot has quit IRC | 11:02 | |
| *** MentalRay has joined #openstack-dns | 11:11 | |
| *** MentalRay has quit IRC | 11:12 | |
| *** pk__ has joined #openstack-dns | 11:39 | |
| openstackgerrit | Endre Karlson proposed openstack/designate: Fix handling of wildcards in api / storage https://review.openstack.org/166806 | 11:39 |
| *** pk__ has quit IRC | 11:43 | |
| *** chlong has joined #openstack-dns | 11:51 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Remove usage of relative_path in ValidationError Objects https://review.openstack.org/167570 | 11:55 |
| *** openstackgerrit has quit IRC | 12:06 | |
| *** openstackgerrit has joined #openstack-dns | 12:06 | |
| *** ChanServ sets mode: +v openstackgerrit | 12:06 | |
| *** richm has joined #openstack-dns | 12:10 | |
| openstackgerrit | Kiall Mac Innes proposed openstack/designate: Support mDNS response over 65k https://review.openstack.org/167268 | 12:10 |
| *** richm has quit IRC | 12:11 | |
| *** richm has joined #openstack-dns | 12:11 | |
| *** jmcbride has joined #openstack-dns | 12:56 | |
| openstackgerrit | Kiall Mac Innes proposed openstack/designate: Unwind mDNS XFRMixin from NotifyEndpoint and Handler https://review.openstack.org/166627 | 12:59 |
| openstackgerrit | Endre Karlson proposed openstack/designate: Allow for duplicate domains in different pools https://review.openstack.org/167582 | 13:00 |
| *** jbratton has quit IRC | 13:10 | |
| *** johnbelamaric has joined #openstack-dns | 13:19 | |
| openstackgerrit | Endre Karlson proposed openstack/designate: Allow for duplicate domains in different pools https://review.openstack.org/167582 | 13:22 |
| *** pk__ has joined #openstack-dns | 13:28 | |
| *** andrewbogott_afk is now known as andrewbogott | 13:30 | |
| *** pk__ has quit IRC | 13:32 | |
| *** openstackgerrit has quit IRC | 13:36 | |
| *** openstackgerrit has joined #openstack-dns | 13:36 | |
| *** ChanServ sets mode: +v openstackgerrit | 13:36 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add API_v2 Adapter Objects https://review.openstack.org/164327 | 13:53 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Replace API_v2 views with Adapters https://review.openstack.org/166983 | 13:53 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Decorator to log RPCAPI Calls https://review.openstack.org/167352 | 13:53 |
| openstackgerrit | Graham Hayes proposed openstack/designate: WIP: Logging Cleanup https://review.openstack.org/167353 | 13:53 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add __str__() methods to DesignateObjects https://review.openstack.org/142218 | 13:53 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Enhanced Logging to APIs https://review.openstack.org/167351 | 13:53 |
| *** mugsie has quit IRC | 14:00 | |
| *** mugsie has joined #openstack-dns | 14:00 | |
| *** paul_glass has joined #openstack-dns | 14:03 | |
| *** jmcbride has quit IRC | 14:03 | |
| *** uschreiber_ has joined #openstack-dns | 14:08 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Schema to RRData Objects https://review.openstack.org/167600 | 14:11 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Remove usage of relative_path in ValidationError Objects https://review.openstack.org/167570 | 14:15 |
| *** mugsie has quit IRC | 14:23 | |
| *** mugsie has joined #openstack-dns | 14:24 | |
| *** ericpeterson has joined #openstack-dns | 14:28 | |
| *** jschwarz has joined #openstack-dns | 14:40 | |
| *** pk__ has joined #openstack-dns | 14:43 | |
| *** mugsie has quit IRC | 14:44 | |
| *** pk__ has quit IRC | 14:48 | |
| Kiall | timsim: about? | 14:48 |
| *** mugsie has joined #openstack-dns | 14:48 | |
| timsim | Yep | 14:48 |
| Kiall | So - Agents, targets vs nameservers, was that the only issue you had with pool manager for that? | 14:49 |
| Kiall | (I'm just cleaning up the review for that now..) | 14:49 |
| *** MentalRay has joined #openstack-dns | 14:50 | |
| timsim | Yep, the only issue was that I needed to be able to Notify/Create/Delete one place, and poll another. | 14:50 |
| timsim | Kiall: Quick question on https://review.openstack.org/#/c/167268/3/designate/dnsutils.py: Is def _xl_to_wire supposed to be embedded in _call_? | 14:51 |
| Kiall | lol.. no | 14:52 |
| Kiall | serves me right for not building a 66k response to test -_- | 14:52 |
| timsim | Yeah I can't make it work :/ | 14:52 |
| timsim | Eventually you get this error: http://paste.openstack.org/show/196603/ | 14:53 |
| goncalo | [A4 | 14:53 |
| timsim | But that was like 7 minutes after the initial request. | 14:53 |
| timsim | The zone does have 40k records, but still. | 14:53 |
| *** mugsie has quit IRC | 14:54 | |
| *** mugsie has joined #openstack-dns | 14:54 | |
| Kiall | timsim: ah, unindent a level and give it a go? | 14:55 |
| Kiall | timsim: yea, endre filed a bug for that yesterday.. Buried a few layers down, we're doing a billion SQL queries for an AXFR | 14:57 |
| timsim | I should probably delete some of these records...still chugging | 14:59 |
| Kiall | Yea.. We're literally doing worst possible case data lookup from the DB .. Suffied to say that needs fixing before rc1 ;) | 15:01 |
| timsim | Ah, I see it. Is ekarlso working on that? | 15:02 |
| Kiall | No, he just spotted it during some stress testing yesterday | 15:02 |
| timsim | Alright, I can probably take it. | 15:02 |
| Kiall | For mDNS (where the issue hits realy hard) - We really just need to implement a new storage method that does a targeted join cross rrsets and rr's, returning the raw rows rather than objects (it's read only - we've no need to go build up the obj's for mDNS) and have mDNS use that.. | 15:03 |
| Kiall | the other cases went unnoticed because we do page at a time .. | 15:04 |
| timsim | Kiall: New error http://paste.openstack.org/show/196606/ | 15:04 |
| *** jmcbride has joined #openstack-dns | 15:04 | |
| Kiall | Grr, I was worried about that :( | 15:05 |
| *** openstackgerrit has quit IRC | 15:08 | |
| *** openstackgerrit has joined #openstack-dns | 15:09 | |
| *** ChanServ sets mode: +v openstackgerrit | 15:09 | |
| *** uschreiber_ has quit IRC | 15:09 | |
| Kiall | timsim: can you post a SQL dump of that by any chance? | 15:16 |
| timsim | Sure, give me a sec | 15:16 |
| Kiall | (looking at the way to fix it, but hard to tell) | 15:16 |
| timsim | https://gist.github.com/TimSimmons/e308f470e7b0a2439153#file-log-sql | 15:20 |
| timsim | Kiall: ^ | 15:20 |
| Kiall | timsim: can you give this a quick go? http://paste.openstack.org/show/Q7UTgZ2qOnyDZw6OCf36/ | 15:23 |
| Kiall | totally untested - brb - coffee calling | 15:23 |
| timsim | Sure | 15:23 |
| *** jmcbride has quit IRC | 15:25 | |
| timsim | Kiall: In "f.read(655....)" what is f? | 15:28 |
| timsim | L331 in the paste | 15:28 |
| *** johnbelamaric has quit IRC | 15:31 | |
| *** johnbelamaric has joined #openstack-dns | 15:31 | |
| Kiall | whoops was meant to be response | 15:33 |
| timsim | That's what i figured. It's working now | 15:33 |
| timsim | Well | 15:33 |
| timsim | MiniDNS is chugging away on 40k recors | 15:33 |
| Kiall | It's still slow as hell, but does the AXFR succeed? | 15:33 |
| timsim | Don't know yet | 15:33 |
| timsim | Well then again, I'm not sure I will know, since the dig timed out | 15:34 |
| timsim | I guess I'll see if there aren't errors in the miniDNS log... | 15:34 |
| timsim | I should really delete some of these records. | 15:34 |
| Kiall | Before you do - I meant a SQL dump if possible, rather than the sql logs ;) | 15:35 |
| Kiall | (only if it's dummy / test data ofc..) | 15:35 |
| timsim | Oh yeah, sure. | 15:36 |
| timsim | Kiall: http://964700e4a3d9dbf5b5ba-7a27b8c5d9fcdc26d383a194ab4f0ebe.r14.cf2.rackcdn.com/bigrecordemp.sql | 15:39 |
| timsim | Most of the records are in an error state. | 15:39 |
| Kiall | I'd imagine so :) | 15:39 |
| ekarlso | found a fix for the select issue ? | 15:40 |
| Kiall | did you say you'd look at the 75 quadrillion (give or take a query or two) AXFR issue BTW? If not, I'll ask endre to look into it tomorrow.. He's gone for the day ;) | 15:41 |
| Kiall | lol - and there he is. | 15:41 |
| timsim | I can do it, sure. | 15:42 |
| ekarlso | Kiall: ... popping in before heading out again :| | 15:43 |
| timsim | Kiall: I don't think I've done the diff right. Where does the call to xl_wire go? http://paste.openstack.org/show/196646/? I | 15:44 |
| *** pk__ has joined #openstack-dns | 15:44 | |
| *** richm has quit IRC | 15:44 | |
| openstackgerrit | Endre Karlson proposed openstack/designate: Allow for duplicate domains in different pools https://review.openstack.org/167582 | 15:45 |
| *** richm has joined #openstack-dns | 15:46 | |
| *** richm has left #openstack-dns | 15:47 | |
| *** richm has joined #openstack-dns | 15:48 | |
| *** pk__ has quit IRC | 15:49 | |
| Kiall | timsim: heh, that would be because I did my diff based on a tree without the xl patch .. http://paste.openstack.org/show/196649/ | 15:50 |
| Kiall | (again, untested, my stack is setup with a different patchset for testing ;)) | 15:50 |
| timsim | Alright, trying again. | 15:52 |
| Kiall | Also - I'm assuming you've told dig to use a stupidly large timeout? | 15:52 |
| Kiall | +time=9999999 or something ;) | 15:52 |
| *** rmoe has quit IRC | 15:54 | |
| timsim | Oh yeah I was totally doing that ;) | 15:55 |
| mugsie | :) | 15:56 |
| *** jschwarz has quit IRC | 15:58 | |
| mugsie | timsim: https://review.openstack.org/#/c/166983/ / https://review.openstack.org/#/c/164327/ need reviews pretty soon - they are important for the v2 API stuffs :) | 15:58 |
| timsim | Yep, on my list for today. vinod is out :/ otherwise I'm sure he would. | 15:58 |
| mugsie | cool :) | 15:59 |
| timsim | Kiall: O.o http://paste.openstack.org/show/196655/ | 16:00 |
| timsim | That was way faster though. | 16:01 |
| Kiall | faster? really? | 16:01 |
| timsim | Yeah for some reason that took ~2 minutes intead of six or seven | 16:01 |
| timsim | Granted this is a fairly small vm on my my machine. | 16:02 |
| Kiall | Ah, well, with a LARGE zone I suppose building the string versus a filehandle-like object would account for some perf difference.. But that seems a tad much. | 16:02 |
| *** richm has quit IRC | 16:03 | |
| timsim | Maybe it abandoned ship when things went wrong, which only about 10% of the zone. | 16:03 |
| Kiall | Anyway - Issue at hand. I mentioned to graham earlier we're doing silly things building a dns.message.Message instance as the response, only to iterate everything AGAIN and build a dns.renderer.Renderer | 16:03 |
| *** richm has joined #openstack-dns | 16:04 | |
| timsim | Yeah that makes sense. Any idea why it build a bad packet? Looks like it did after the 65535 limit | 16:04 |
| timsim | s/build/built | 16:04 |
| Kiall | Yep - dns compression is per packet, I expected we'd see it.. | 16:05 |
| Kiall | I think we need to do 2 things.. 1) Get a patch in that get's us up to 65k, and probably adds the cStringIO stuff since it made such a difference.. 2) rework how we render to-wire so we can control it better | 16:08 |
| *** jschwarz has joined #openstack-dns | 16:08 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add _to_string() methods on RRData Objects. https://review.openstack.org/167672 | 16:11 |
| Kiall | Annoyingly, no option to disable the compression | 16:11 |
| *** rmoe has joined #openstack-dns | 16:11 | |
| timsim | That's unfortunate | 16:12 |
| Kiall | Ohh.. Looking again.. I'm not so sure. | 16:12 |
| Kiall | I just noticed L38 - ;; WARNING: ID mismatch: expected ID 51399, got 1 | 16:12 |
| Kiall | I don't suppose you have this zone, or a similarity large one, loaded into a BIND server? a pcap of the AXFR would be interesting | 16:13 |
| timsim | The version of that zone I have on the bind9 server is ;; XFR size: 3145 records (messages 2, bytes 65065) | 16:15 |
| timsim | + | 16:15 |
| timsim | I could look for a larger zone on one of our other nameservers, if you'd like? | 16:19 |
| Kiall | bytes 65065 - so under the limit yet split into 2 messages.. Can you fire up wireshark or similar to grab a pcap? My memory of the protcol is clearly wrong given it ain't working, a pcap showing the right thing to do should make it easier... | 16:19 |
| Kiall | Scratch that, found it.. Yea, we have some re-work to do. https://tools.ietf.org/html/draft-andrews-dns-more-01 | 16:21 |
| *** chlong has quit IRC | 16:22 | |
| Kiall | Humm - Reading, that doesn't look like what I remember .. | 16:22 |
| openstackgerrit | Kiall Mac Innes proposed openstack/designate: WIP: Implement Pool Targets https://review.openstack.org/167678 | 16:27 |
| *** MentalRay has quit IRC | 16:28 | |
| timsim | Kiall: I don't really know what I'm doing, but here's what I hope is a pcap of the dig: http://964700e4a3d9dbf5b5ba-7a27b8c5d9fcdc26d383a194ab4f0ebe.r14.cf2.rackcdn.com/capture.pcap | 16:28 |
| openstackgerrit | Kiall Mac Innes proposed openstack/designate: WIP: Implement Pool Targets https://review.openstack.org/167678 | 16:28 |
| Kiall | timsim: I don't see any DNS traffic in there, lots of SSH though ;) | 16:31 |
| *** jordanP has quit IRC | 16:31 | |
| *** jordanP has joined #openstack-dns | 16:31 | |
| *** pk__ has joined #openstack-dns | 16:32 | |
| *** jschwarz has quit IRC | 16:32 | |
| *** chlong has joined #openstack-dns | 16:33 | |
| timsim | Kiall: how about this one: http://964700e4a3d9dbf5b5ba-7a27b8c5d9fcdc26d383a194ab4f0ebe.r14.cf2.rackcdn.com/capture2.pcap | 16:34 |
| *** pk__ has quit IRC | 16:36 | |
| *** chlong has quit IRC | 16:39 | |
| Kiall | timsim: looks more like it :) | 16:42 |
| timsim | First time I've done that :P Thanks google | 16:43 |
| timsim | bbiab lunch :) | 16:46 |
| Kiall | lol - spotted this ;) | 16:52 |
| Kiall | # TODO(vinod) check if we dnspython has an upper limit on the number | 16:52 |
| Kiall | # of rrsets. | 16:52 |
| Kiall | response.answer = r_rrsets | 16:52 |
| Kiall | either way, I can see from the pcap the patch is doing it wrong ;) | 16:53 |
| * timsim forgot about the irc meet :P lunch will wait | 16:54 | |
| Kiall | :D | 16:56 |
| *** andrewbogott is now known as andrewbogott_afk | 16:58 | |
| *** untriaged-bot has joined #openstack-dns | 17:02 | |
| untriaged-bot | Untriaged bugs so far: | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432842 | 17:02 |
| openstack | Launchpad bug 1432842 in Designate "Property setters dont work." [Undecided,New] | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432850 | 17:02 |
| openstack | Launchpad bug 1432850 in Designate "Add periodic refresh task" [High,New] - Assigned to Endre Karlson (endre-karlson) | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1432839 | 17:02 |
| openstack | Launchpad bug 1432839 in Designate "Calling find_domain in any v1 operation is sub-optimal" [Medium,New] - Assigned to Endre Karlson (endre-karlson) | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1433645 | 17:02 |
| openstack | Launchpad bug 1433645 in Designate "Floating IP Reverse DNS API does not respect async operations" [Undecided,New] | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1435583 | 17:02 |
| openstack | Launchpad bug 1435583 in Designate "designatedashboard packaging is not very simple" [Undecided,In progress] - Assigned to Eric Peterson (ericpeterson-l) | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1425117 | 17:02 |
| openstack | Launchpad bug 1425117 in Designate "Designate does not work with postgres" [High,New] | 17:02 |
| untriaged-bot | https://bugs.launchpad.net/designate/+bug/1435568 | 17:02 |
| openstack | Launchpad bug 1435568 in Designate "pep8 gate taking longer than 30 mins on occasion" [Undecided,New] | 17:02 |
| *** untriaged-bot has quit IRC | 17:02 | |
| *** jmcbride has joined #openstack-dns | 17:06 | |
| *** andrewbogott_afk is now known as andrewbogott | 17:08 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Decorator to log RPCAPI Calls https://review.openstack.org/167352 | 17:10 |
| openstackgerrit | Graham Hayes proposed openstack/designate: WIP: Logging Cleanup https://review.openstack.org/167353 | 17:10 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add __str__() methods to DesignateObjects https://review.openstack.org/142218 | 17:10 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Enhanced Logging to APIs https://review.openstack.org/167351 | 17:10 |
| *** andrewbogott is now known as andrewbogott_afk | 17:11 | |
| *** andrewbogott_afk is now known as andrewbogott | 17:14 | |
| *** jmcbride has quit IRC | 17:18 | |
| *** jmcbride has joined #openstack-dns | 17:18 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add _to_string() methods on RRData Objects https://review.openstack.org/167672 | 17:25 |
| *** fouxm_ is now known as fouxm | 17:44 | |
| *** jordanP has quit IRC | 17:48 | |
| *** ericpeterson has quit IRC | 17:54 | |
| *** ericpeterson has joined #openstack-dns | 17:58 | |
| *** crc32 has joined #openstack-dns | 18:03 | |
| *** rmoe has quit IRC | 18:04 | |
| *** kfox1111 has joined #openstack-dns | 18:06 | |
| kfox1111 | ah. there we go. | 18:06 |
| kfox1111 | trying out designate. havent put a service record for it in keystone yet. shouldn't need to for testing. tried setting both OS_DNS_ENDPOINT=http://localhost:9001 or --os-endpoint http://localhost:9001 but the client still is insisting on contacting keystone to find the endpoint. | 18:07 |
| kfox1111 | is this a known bug? | 18:07 |
| mugsie | kfox1111: yes | 18:11 |
| mugsie | let me dig it out | 18:11 |
| mugsie | https://bugs.launchpad.net/python-designateclient/+bug/1415560 | 18:12 |
| openstack | Launchpad bug 1415560 in python-designateclient "python-designateclient does not work against an non Keystoned designate" [High,Triaged] - Assigned to Graham Hayes (grahamhayes) | 18:12 |
| kfox1111 | sure enough. | 18:13 |
| kfox1111 | odd. | 18:15 |
| *** johnbelamaric has quit IRC | 18:15 | |
| mugsie | i think it was introduced as part of ks sessions | 18:15 |
| mugsie | its on my list to look at, but a bit down it atm | 18:16 |
| kfox1111 | yeah digging through it now. not obvious why its having a problem. | 18:16 |
| kfox1111 | ok. really not obvious. Guess I'll have to just add a dummy record for now. | 18:19 |
| *** johnbelamaric has joined #openstack-dns | 18:19 | |
| andrewbogott | Do post-icehouse versions of sink support inserting multiple dns entries for a given host? | 18:20 |
| kfox1111 | So, the server pool support seems to be manually seting up pools and then telling designate about them, rather then having designate launch vm's for the user, correct? | 18:22 |
| *** pk__ has joined #openstack-dns | 18:26 | |
| *** ericpeterson has quit IRC | 18:29 | |
| *** johnbelamaric has quit IRC | 18:29 | |
| *** ericpete_ has joined #openstack-dns | 18:29 | |
| mugsie | kfox1111: yup | 18:31 |
| kfox1111 | ok. can a user setup a pool, or is it only admins at this point? | 18:32 |
| mugsie | only admins right now | 18:32 |
| kfox1111 | IE, a user wants to setup a private dns server in their tenant and manage it with designate. | 18:32 |
| kfox1111 | ok. | 18:32 |
| mugsie | it is based on policy - but currently the info for the pool is loaded from config files on disk | 18:33 |
| kfox1111 | hmm. ok. | 18:33 |
| mugsie | so, when that moves to the DB, you could add a role that is allowed create pools | 18:33 |
| mugsie | and it should work | 18:33 |
| kfox1111 | ah. | 18:33 |
| kfox1111 | so maybe L time frame for that? | 18:33 |
| mugsie | yeah | 18:33 |
| kfox1111 | k. | 18:34 |
| kfox1111 | I'm just glad it supports it in any form. :) | 18:34 |
| kfox1111 | manual's ok so long as there is a path forward. :) | 18:34 |
| ekarlso | heya kfox1111, mind you that I ask what co you're with ? | 18:35 |
| kfox1111 | Pacific Northwest National Lab. | 18:35 |
| kfox1111 | http://www.pnl.gov/ | 18:36 |
| kfox1111 | hmm.. a lot of stuff in the v2 api. is it going to be considered stable for kilo? does the cli support it yet? horizon? | 18:36 |
| ekarlso | kfox1111: there's a "alpha" branch up on gerrit for v2 using openstack cli | 18:37 |
| kfox1111 | ah. cool. thanks. :) | 18:38 |
| kfox1111 | I can just curl it too, but would rather save some typing if I can help it. ;) | 18:38 |
| *** pk__ has quit IRC | 18:39 | |
| kfox1111 | ok. so a server is a dns server. the list of servers is the servers in the pool that will store the domains? | 18:39 |
| openstackgerrit | Endre Karlson proposed openstack/designate: Allow for duplicate domains in different pools https://review.openstack.org/167582 | 18:40 |
| timsim | kfox1111: Yep | 18:40 |
| kfox1111 | ok. I think I'm getting it. | 18:41 |
| ekarlso | kfox1111: https://review.openstack.org/133675 bindings for v2 https://review.openstack.org/133676 cli | 18:41 |
| openstackgerrit | Endre Karlson proposed openstack/designate: Add code to allow triggering of AXFR from API https://review.openstack.org/165331 | 18:42 |
| kfox1111 | so, whats the difference between server-list and domain-server-list? | 18:44 |
| kfox1111 | oh.... so all the server's in a pool might not be serving a domain, and thats what domain-server-list tells you? | 18:47 |
| kfox1111 | if so, that makes sense. | 18:47 |
| timsim | It's a bit outdated, but domain-server-list would essentially get the NS records of the zone for you. So that would tell you which DNS servers are serving for you. All of the servers in a pool in which the domain resides will be serving the zone. | 18:50 |
| kfox1111 | oh. I think I get it. you would use that if, as a user, you needed to register the name servers with a dns registrar, and you'd use that to find out what they were? | 18:52 |
| *** johnbelamaric has joined #openstack-dns | 18:52 | |
| *** ericpete_ has quit IRC | 18:52 | |
| timsim | Yep. That's correct. | 18:53 |
| kfox1111 | cool. | 18:53 |
| kfox1111 | can you blacklist domains too, or just zones? | 18:55 |
| mugsie | kfox1111: blacklists are a regex | 18:57 |
| mugsie | so if you can match it, it will be blocked :) | 18:57 |
| mugsie | also, in designate zone==domain (i know it is VERY annoying) | 18:57 |
| openstackgerrit | Endre Karlson proposed openstack/designate: Add code to allow triggering of AXFR from API https://review.openstack.org/165331 | 19:00 |
| kfox1111 | so what does the v2 zone api do? | 19:04 |
| mugsie | its the new name for domain | 19:04 |
| mugsie | if it is in v1 its a domain, in v2 its a zone | 19:05 |
| kfox1111 | ah. ok. | 19:05 |
| kfox1111 | is there a flag to invert the blacklist making it a white list? | 19:06 |
| mugsie | afaik - no | 19:06 |
| kfox1111 | say, I wanted to allow only *.cloud.pnnl.gov subdomains? | 19:06 |
| mugsie | oh | 19:06 |
| mugsie | that can be done | 19:06 |
| mugsie | or... | 19:06 |
| mugsie | let me take a look quickly | 19:06 |
| mugsie | i *think* you could do it with the tld api | 19:07 |
| kfox1111 | ah. | 19:07 |
| mugsie | if there is no tld's created we allow all | 19:08 |
| mugsie | if there is tlds, we limit domains to that | 19:08 |
| timsim | hah. | 19:08 |
| mugsie | so, creating a tld of cloud.pnnl.gov. might work | 19:08 |
| kfox1111 | ah. so you do a tld of 'cloud.pnnl.gov' | 19:08 |
| timsim | That's pretty cool. | 19:08 |
| kfox1111 | yeah. | 19:08 |
| kfox1111 | nice. | 19:08 |
| mugsie | I would say - test it :) | 19:08 |
| mugsie | i running on my gut here ;) | 19:08 |
| kfox1111 | sure. :) | 19:09 |
| mugsie | let me know if it works | 19:09 |
| mugsie | :) | 19:09 |
| kfox1111 | "The domain/zone name must not be present as a Tld entry in the database. e.g. If there is a Tld entry co.uk in the database, then a domain/zone with the name co.uk. cannot be created." | 19:09 |
| mugsie | yes | 19:11 |
| kfox1111 | ok. so, out of the box, is it easy to configure designate to give every tenant its own subdomain, or is that a lot of work? | 19:11 |
| mugsie | we generally allow each tenant to create their own | 19:11 |
| kfox1111 | how would they hook it up to the providers space without getting an admin to do that? or is that the manual part? | 19:12 |
| mugsie | but, it could be done relitily easily with a script | 19:12 |
| mugsie | our back ends create the zone on the DNS servers automatically | 19:12 |
| *** pk__ has joined #openstack-dns | 19:13 | |
| kfox1111 | maybe I just modify our "create project" scripts to create the domain for the user, and add the subdomain to the cloud.pnnl.gov domain. | 19:13 |
| mugsie | you could point the NS Records for cloud.pnnl.gov to the designate server, and then avoid the last step | 19:14 |
| kfox1111 | With that arangement, can I use designate to manage cloud.pnnl.gov when it has subdomains that designate is also managing? | 19:14 |
| kfox1111 | ok. yeah. I think that should work. | 19:14 |
| *** ericpeterson has joined #openstack-dns | 19:14 | |
| mugsie | that does present a problem - (the sub domain) | 19:14 |
| *** ericpeterson has quit IRC | 19:15 | |
| mugsie | we do generally block users from creating sub domains of other peoples tenants | 19:15 |
| *** ericpeterson has joined #openstack-dns | 19:15 | |
| mugsie | s/tenants/zones | 19:15 |
| kfox1111 | oh... | 19:15 |
| kfox1111 | can an admin do it? | 19:15 |
| mugsie | but. There is a transfer of ownership, or admins can do it as well | 19:15 |
| kfox1111 | then they can manage the entries in their own subdomain? | 19:16 |
| mugsie | yes | 19:16 |
| kfox1111 | well, the "tld" domain, cloud.pnnl.gov would be alwasy owned by the admin project, and the <tenantX>.cloud.pnnl.gov by the tenant. | 19:16 |
| mugsie | yes, that should work | 19:16 |
| kfox1111 | that would be ok if admin was the one creating the tenantX domain? | 19:17 |
| kfox1111 | ok. cool. | 19:17 |
| kfox1111 | man. I have a million questions today. :) | 19:17 |
| mugsie | yup - or the policy.jsonm file was changed to allow it | 19:17 |
| kfox1111 | what about about dnssec. any support yet? | 19:17 |
| mugsie | not yet | 19:17 |
| kfox1111 | I know next to nothing about it, but know pnnl.gov is using it. | 19:17 |
| kfox1111 | ok. | 19:18 |
| kfox1111 | one more crazy one... can you have one zone managed by multiple projects at a first come first served record level? | 19:19 |
| kfox1111 | /tmp sticky bit like. :) | 19:19 |
| *** johnbelamaric has quit IRC | 19:20 | |
| mugsie | kfox1111: no. that has been talked about, but no-one is sure how to do record level ACLs | 19:20 |
| kfox1111 | yeah. I could see it being messy. figured It wouldn't hurt to ask. | 19:21 |
| mugsie | we are planning to allow multiple project manage one zone, but it would be a free for all for each project (they could delete each others records) | 19:21 |
| mugsie | never hurts to ask :) | 19:21 |
| mugsie | you are not the first either | 19:21 |
| kfox1111 | subdomains are my preference anyway, but it seems our organization struggles with ssl certs/wild card certs. so subdomains have that one drawback. :/ | 19:21 |
| mugsie | yeah, when I did some IT work in 3rd level education, get one of those was nearly impossible | 19:22 |
| kfox1111 | I think the relatively simple, put a project on each record on a shared zone. only that project can delete/update it. | 19:22 |
| kfox1111 | fairly simple, but should be rather sane semantics? | 19:23 |
| kfox1111 | so, assume record level project owner was accepted as the way to do it, how hard do you think it would be to implement? | 19:24 |
| mugsie | yeah, but the increase in policy checks etc would have a detrimental effect on performance | 19:24 |
| kfox1111 | (Trying to decide if it would be harder to setup a CA with all the automation, or adding support to Designate :) | 19:24 |
| mugsie | to implement - not too hard. to implement so it scales - a bit tricky | 19:24 |
| kfox1111 | yeah. You'd have to have a state flag saying if it was a shared domain or not. If not, skip the extra checks. | 19:25 |
| *** andrewbogott is now known as andrewbogott_afk | 19:25 | |
| kfox1111 | on the other hand... | 19:25 |
| kfox1111 | its probably something you don't want normal users to be able to do, set the +t bit. so maybe the admin just puts the list of allowed sticky's in the config file, then its always in memory. | 19:26 |
| mugsie | but if someone did a zonefile import to that domain - pain ensues ;) | 19:27 |
| kfox1111 | would be disallowed if it was sticky? | 19:27 |
| kfox1111 | export too for that matter. | 19:27 |
| kfox1111 | yeah. a few gotcha's in it... but doesn't seem too bad. | 19:28 |
| kfox1111 | still sounding like less pain then dealing with the ssl certs. :/ | 19:28 |
| mugsie | :) | 19:29 |
| *** pk__ has quit IRC | 19:29 | |
| kfox1111 | The irony is, it took us like 3 weeks to buy a cert to secure one of our machines recently. and had to go through three levels of management approval to do so. :/ | 19:30 |
| kfox1111 | I really don't get it sometimes... | 19:30 |
| kfox1111 | ok. I might propose a spec for that later. | 19:30 |
| kfox1111 | I'll ask my manager. | 19:31 |
| mugsie | cool. we use the -spec process that most of the projects use | 19:31 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Decorator to log RPCAPI Calls https://review.openstack.org/167352 | 19:31 |
| openstackgerrit | Graham Hayes proposed openstack/designate: WIP: Logging Cleanup https://review.openstack.org/167353 | 19:31 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add __str__() methods to DesignateObjects https://review.openstack.org/142218 | 19:31 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Enhanced Logging to APIs https://review.openstack.org/167351 | 19:31 |
| kfox1111 | ok. I've done a couple of those now. | 19:31 |
| kfox1111 | you going to the summit? | 19:32 |
| mugsie | hopefully | 19:32 |
| mugsie | seen as I am presenting | 19:32 |
| mugsie | you? | 19:32 |
| kfox1111 | hehe. yeah. | 19:32 |
| kfox1111 | will probably see you there then. :) | 19:32 |
| mugsie | its a long trip for us out, so got to get managers on board :) | 19:32 |
| kfox1111 | where are you located? | 19:33 |
| mugsie | ireland | 19:33 |
| kfox1111 | I'm in Washington state, so its a pretty short trip for me. | 19:33 |
| kfox1111 | ah. cool. :) | 19:33 |
| mugsie | ah, yeah it would be | 19:33 |
| timsim | All of this stuff always seems to be on the west coast, we should do something on the east coast for you guys :P | 19:33 |
| mugsie | timsim: its ok, I like seattle :) | 19:34 |
| Kiall | timsim: ... Or in EU ;) | 19:34 |
| mugsie | and, atalanta and paris were closer | 19:34 |
| timsim | Kiall: Nah. 'Mericuh ;) | 19:34 |
| mugsie | but midcycle in EU next time? ;) | 19:34 |
| timsim | I think it's our turn :P | 19:34 |
| Kiall | mugsie: ++ ;) | 19:34 |
| kfox1111 | :) | 19:34 |
| mugsie | although it would be a shame to have the Liberty midcycle not in the US :) | 19:35 |
| kfox1111 | reverse_name? | 19:36 |
| kfox1111 | in the database? | 19:36 |
| kfox1111 | that for an optomization? | 19:36 |
| mugsie | makes seraching by domain name much faster | 19:36 |
| Kiall | kfox1111: Yep, it is | 19:36 |
| timsim | Yep | 19:36 |
| kfox1111 | oh. yeah.... | 19:37 |
| kfox1111 | do you ever want to search forward then? | 19:37 |
| mugsie | sometimes. but the most searches are done the other way | 19:37 |
| kfox1111 | if not, you could drop the forward version and save a bit per row. | 19:37 |
| mugsie | i think timsim wrote it to be smart like that | 19:38 |
| mugsie | Kiall: didnt want that ;) | 19:38 |
| timsim | Yeah I think it searches the reverse name when beneficial, and the vice-versa. | 19:38 |
| mugsie | he was giving out about admins trying to read the DB | 19:38 |
| mugsie | he gets it wrong now and again :P | 19:39 |
| Kiall | dropping the name column in favor of only the reverse name column ain't happening ;) I don't do backwards reading! Tis hard enough reading forwards ;) | 19:39 |
| timsim | hahaha | 19:39 |
| * mugsie does ack that kiall barely does reading | 19:39 | |
| kfox1111 | select reverse(reverse_name) from domains? :) | 19:41 |
| *** mikeit has joined #openstack-dns | 19:42 | |
| mugsie | oh, this arguement has been had.... repeatedly | 19:42 |
| mugsie | :D | 19:42 |
| Kiall | can you imaging being in the middle of an incident, say partial DB corruption, you're panicing. Every second counts. 0 rows matched. Fek. double check the spelling, 0 rows matched. "Oh sh** - we've lost data".. 4 minutes later, AHHH YES, the name is in reverse! that's it! | 19:42 |
| * kfox1111 is amused by the double negative | 19:42 | |
| mugsie | Kiall: should have better backups | 19:42 |
| mugsie | :D | 19:42 |
| kfox1111 | no worries. I'll stop armchair quarterbacking. :) | 19:42 |
| mugsie | its OK i do it all the time | 19:43 |
| mugsie | its loads of fun | 19:43 |
| kfox1111 | yeah, but its harder to trip over it when its called reversed_name, and not name. | 19:43 |
| kfox1111 | select * from domains where name='foo.bar'; 'colum name not found' | 19:43 |
| kfox1111 | oh, yeah... reversed_name... :) | 19:44 |
| *** MentalRay has joined #openstack-dns | 19:44 | |
| kfox1111 | now calling it name when its reversed, that would be ugly. :/ | 19:44 |
| mugsie | tbh, its not a huge amount of data, and its not indexed (or it shouldnt be) - so its fine to denormalise it that way | 19:44 |
| kfox1111 | hmm... yeah. | 19:44 |
| * Kiall takes a screenshot | 19:45 | |
| mugsie | i said it was fine, not good ;P | 19:45 |
| * mugsie is being mis-represented | 19:45 | |
| * kfox1111 chuckles | 19:46 | |
| mugsie | ask timsim what midcycles / summit design sessions are like ;) | 19:46 |
| timsim | These two bickering like an old married couple, it's amazing :) | 19:50 |
| *** johnbelamaric has joined #openstack-dns | 20:00 | |
| *** johnbelamaric has quit IRC | 20:00 | |
| *** johnbelamaric has joined #openstack-dns | 20:00 | |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add __str__() methods to DesignateObjects https://review.openstack.org/142218 | 20:04 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Schema to RRData Objects https://review.openstack.org/167600 | 20:04 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add _to_string() methods on RRData Objects https://review.openstack.org/167672 | 20:11 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Decorator to log RPCAPI Calls https://review.openstack.org/167352 | 20:12 |
| openstackgerrit | Graham Hayes proposed openstack/designate: WIP: Logging Cleanup https://review.openstack.org/167353 | 20:12 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add __str__() methods to DesignateObjects https://review.openstack.org/142218 | 20:12 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add Enhanced Logging to APIs https://review.openstack.org/167351 | 20:12 |
| mugsie | oh rebases | 20:14 |
| mugsie | some idiot decided to get patches rebased into the wronjg chain | 20:15 |
| * mugsie whispers "it might have been me" | 20:15 | |
| *** jmcbride1 has joined #openstack-dns | 20:17 | |
| *** jmcbride has quit IRC | 20:18 | |
| *** pk__ has joined #openstack-dns | 20:23 | |
| *** pk__ has quit IRC | 20:28 | |
| *** jmcbride has joined #openstack-dns | 20:33 | |
| *** pk__ has joined #openstack-dns | 20:40 | |
| *** pk__ has quit IRC | 20:44 | |
| openstackgerrit | Merged openstack/designate: improving designatedashboard packaging https://review.openstack.org/167024 | 20:49 |
| *** pk__ has joined #openstack-dns | 20:50 | |
| *** pk__ has quit IRC | 20:51 | |
| *** jmcbride has quit IRC | 20:52 | |
| *** jmcbride has joined #openstack-dns | 21:00 | |
| *** MentalRay has quit IRC | 21:00 | |
| *** jmcbride has quit IRC | 21:07 | |
| *** MentalRay has joined #openstack-dns | 21:11 | |
| *** pk__ has joined #openstack-dns | 21:26 | |
| kfox1111 | heh. sweet. I was just looking into how to do designate dashboard. :) | 21:30 |
| *** richm has quit IRC | 21:31 | |
| kfox1111 | know if trunk will work with a juno dashboard or does it need javascript bits only in kilo? | 21:31 |
| mugsie | kfox1111: eh... | 21:32 |
| mugsie | not sure | 21:32 |
| mugsie | this is a code dump from an internal dashboard | 21:32 |
| mugsie | (the one used by HP Cloud) | 21:33 |
| mugsie | so, maybe? | 21:33 |
| mugsie | its pretty basic | 21:33 |
| kfox1111 | k. might just try it then. thanks. :) | 21:35 |
| *** penick has joined #openstack-dns | 21:36 | |
| *** pk__ has quit IRC | 21:41 | |
| openstackgerrit | Tim Simmons proposed openstack/designate: WIP: Reduce the # of SQL queries during AXFRs https://review.openstack.org/167809 | 21:42 |
| *** mikeit has quit IRC | 21:43 | |
| *** johnbelamaric has quit IRC | 21:44 | |
| Kiall | timsim: that was quick :) Did it work out any better in your large AXFR? | 21:44 |
| timsim | Got to put it on top of your patch to see. | 21:45 |
| Kiall | e.g. fail after 30 sec rather than 2 min? Since the 65k limit is still around;) | 21:45 |
| timsim | It's pretty gross code, but I wanted to see if it resembled what you had in mind | 21:45 |
| Kiall | It looks like the general idea all right :) | 21:46 |
| kfox1111 | is there a way to have designate manage reverse lookups on ip's too? | 21:46 |
| kfox1111 | I have one cloud that may benifit from designate, but their services are grid services, and those tend to be very very sensitive when it comes to reverse lookups. :/ | 21:47 |
| kfox1111 | if forward and reverse don't match up and ssl and everything, it bombs. :/ | 21:48 |
| mugsie | kfox1111: yup | 21:48 |
| mugsie | there is an extention that enables it | 21:49 |
| mugsie | one sec | 21:49 |
| kfox1111 | awesome. thanks. :) | 21:49 |
| Kiall | So, you can certainly manually line things up between the forward / reverse zones - and if your auto creating with sink, you can (it's pluggable) have it manage forward/reverse together.. | 21:49 |
| kfox1111 | nice. :) | 21:49 |
| mugsie | this is on neutron floating ips? | 21:50 |
| kfox1111 | yeah. | 21:50 |
| mugsie | http://docs.openstack.org/developer/designate/integrations.html | 21:51 |
| ericpeterson | you guys are going to want to punch me from this change: https://review.openstack.org/#/c/167024/ I forget to add the empty file designatedashboard/enabled/__init__.py | 21:51 |
| mugsie | it would be 2 calls though | 21:51 |
| * ericpeterson swallows his pride | 21:51 | |
| mugsie | ericpeterson: nice :) | 21:51 |
| mugsie | push up a fix, and we should get it through :) | 21:52 |
| Kiall | ericpeterson: lol.. fire another review up and we can +A right away since theres enough core online ;) | 21:52 |
| ericpeterson | thanks / sorry. I usually have a slightly higher level of incompetence than this | 21:52 |
| Kiall | No worries ;) | 21:52 |
| openstackgerrit | Eric Peterson proposed openstack/designate: forgot missing __init__.py file https://review.openstack.org/167812 | 21:54 |
| timsim | Kiall: It did something...different? http://paste.openstack.org/show/196750/ | 21:55 |
| Kiall | Wait, so, 4 sec before the explosion? | 21:55 |
| Kiall | BOO YA. | 21:55 |
| timsim | It went way farther too | 21:55 |
| Kiall | That's what we wanted :) | 21:55 |
| kfox1111 | cool. thanks. :) | 21:55 |
| timsim | I think earlier it stopped at like 3k | 21:55 |
| timsim | I thought it would blow up at the same place. | 21:56 |
| Kiall | Odd - It looks like the order is sorted here, while from memory it wasn't sorted | 21:56 |
| Kiall | Which in theory means better compression and more records in the 65k? | 21:57 |
| * timsim scrolls back | 21:57 | |
| Kiall | (at least with RRSets named like yours) | 21:57 |
| Kiall | Actually.. Not sorted.. Ignore me | 21:57 |
| Kiall | Either way - real 0m4.092s .. That's better than the 7-8 mins from earlier today :) | 21:58 |
| timsim | Yeah definitely | 21:58 |
| timsim | Here it is | 21:58 |
| timsim | http://paste.openstack.org/show/196655/ | 21:58 |
| timsim | Different error | 21:58 |
| Kiall | Yea, we're "randomly" cutting the response in 2 - a different sort of any kind would mean things line up different | 21:59 |
| *** penick has quit IRC | 21:59 | |
| Kiall | they both start with the same ;; WARNING: ID mismatch: expected ID 42417, got 12851 | 21:59 |
| timsim | Ok, it's doing about the same amount of records though. ~3k | 21:59 |
| timsim | Just ordered a bit different | 21:59 |
| timsim | Well 4 seconds is a lot better than 7 minutes :P | 22:00 |
| *** pk__ has joined #openstack-dns | 22:00 | |
| Kiall | ericpeterson: +A'd | 22:01 |
| Kiall | timsim: so, just doubling checking my reading of _prep_rrsets | 22:02 |
| Kiall | In the records come out out of order from the DB, it explodes right? | 22:03 |
| Kiall | RRSET1 Record1 | 22:03 |
| Kiall | RRSET1 Record2 | 22:03 |
| Kiall | RRSET2 Record1 | 22:03 |
| Kiall | RRSET1 Record3 | 22:03 |
| *** penick has joined #openstack-dns | 22:03 | |
| Kiall | the 4th row in that will create a duplicated RRSet? | 22:03 |
| timsim | Yes it would | 22:04 |
| timsim | I should be ordering by recordset id | 22:04 |
| Kiall | Yea, I think that makes sense | 22:04 |
| ericpeterson | thanks Kiall | 22:05 |
| *** andrewbogott_afk is now known as andrewbogott | 22:07 | |
| kfox1111 | wow... the cli doesn't work with the endpoint workaround thing broke the cloud. :/ | 22:07 |
| kfox1111 | keystone endpoint-create --region Pilot --service-id e7f8c7213e4144f9b0e38334ed3aeb52 --publicurl http://127.0.0.1:9001 --adminurl http://127.0.0.1:9001 --internalurl http://127.0.0.1:9001 | 22:07 |
| kfox1111 | then horizon won't let users login. | 22:07 |
| kfox1111 | keystone endpoint-delete de4a10efd6e94df28f7d1df6b566a5c4 | 22:08 |
| kfox1111 | and horizon lets users in again. :/ | 22:08 |
| timsim | mugsie: I promise I'll review those changes first thing tomorrow morning. | 22:08 |
| kfox1111 | is there any designate related code in juno's dashboard to begin with? | 22:10 |
| *** crc32 has quit IRC | 22:11 | |
| openstackgerrit | Tim Simmons proposed openstack/designate: WIP: Reduce the # of SQL queries during AXFRs https://review.openstack.org/167809 | 22:11 |
| ericpeterson | kfox1111 - are you asking "does horizon have native support for designate?" | 22:13 |
| *** pk__ has quit IRC | 22:14 | |
| Kiall | timsim: 1 tiny comment on that BTW ;) | 22:14 |
| kfox1111 | no, I'm grasping for straws at why a stock juno dashboard seems to break when I put that endpoint into keystone. :/ | 22:14 |
| kfox1111 | no plugins loaded. | 22:15 |
| ericpeterson | I have an idea.... | 22:15 |
| *** andrewbogott is now known as andrewbogott_afk | 22:16 | |
| timsim | Kiall: Good call. I didn't think about names at all. _prep_rrsets isn't my favorite thing either. I was also wondering if find_recordsets_axfr needs to be private, that seems to be the pattern for those. | 22:16 |
| ericpeterson | horizon's session store keeps your session info in a cookie, that is limited to 4k. if your catalog gets too big, the session information is halfway saved.... and bad things happen | 22:16 |
| kfox1111 | ah. | 22:17 |
| ericpeterson | kfox1111 - you can switch horizon to use memcached for it's session store, as a test to my theory | 22:17 |
| kfox1111 | k. | 22:17 |
| ericpeterson | (assuming memcached is installed on your horizon node - most likely is) | 22:17 |
| kfox1111 | its not, but should be easy enough to do. | 22:17 |
| ericpeterson | https://docs.djangoproject.com/en/1.7/ref/settings/#std:setting-SESSION_ENGINE - set it to cache | 22:19 |
| ericpeterson | in real / production setups, I'd take the questions into horizon room, also recommend db session backend | 22:19 |
| kfox1111 | this is in the local config: CACHES={'BACKEND': 'django.core.cache.backends.memcached.MemcachedCache', LOCATION': '127.0.0.1:11211'} | 22:20 |
| kfox1111 | where do I set the django options? | 22:20 |
| ericpeterson | https://github.com/openstack/horizon/blob/master/openstack_dashboard/settings.py#L177 kfox1111 | 22:21 |
| kfox1111 | so, I set that to django.contrib.sessions.backends.cache then. | 22:22 |
| ericpeterson | kfox1111 - let's take further questions into openstack-horizon room if you like | 22:22 |
| *** pk__ has joined #openstack-dns | 22:22 | |
| kfox1111 | ok. thanks for the help. | 22:22 |
| ericpeterson | np :D | 22:22 |
| Kiall | timsim: re private, no - I don't think so.. it's akin to find_recordsets(), which happens to use some private stuff | 22:23 |
| kfox1111 | ok. that seemed to do the trick. thanks. :) | 22:24 |
| Kiall | I also wonder if it's find_recordsets_raw? | 22:24 |
| Kiall | (matching up with the _select_raw kinda-ish) | 22:24 |
| *** pk__ has quit IRC | 22:24 | |
| *** pk has joined #openstack-dns | 22:24 | |
| *** pk has quit IRC | 22:33 | |
| *** paul_glass has quit IRC | 22:35 | |
| mugsie | timsim: yeah, whatever :P | 22:40 |
| kfox1111 | hmm.... so I assume the designate horizon dashboard depends on the python client? | 22:40 |
| mugsie | i am looking at you code now - might have a few other speedups to try :D | 22:40 |
| mugsie | kfox1111: yes | 22:40 |
| kfox1111 | hmm... ok. this will get interesting... :) | 22:40 |
| ekarlso | mugsie: what speedups ? | 22:41 |
| mugsie | kfox1111: rm -rf / && go live in the cascades ;) | 22:41 |
| mugsie | ekarlso: for the patch he just put up | 22:41 |
| mugsie | kfox1111: i have been tempted many times with that horizon plugin | 22:42 |
| ekarlso | mugsie: btw, i hope you have your alcohol budget set for when you come over :P | 22:42 |
| mugsie | though, i will say tripleO was worse | 22:42 |
| mugsie | ekarlso: yeah - 0 | 22:43 |
| kfox1111 | is it painful to setup? | 22:43 |
| ekarlso | mugsie: :o | 22:43 |
| mugsie | i can;t afford to drink in Norway ;) | 22:43 |
| openstackgerrit | Merged openstack/designate: forgot missing __init__.py file https://review.openstack.org/167812 | 22:43 |
| ekarlso | mugsie: there is "happy" hour though I think some days in the neighbour city | 22:43 |
| kfox1111 | I've got a vm with rdo juno dashboard in it. just going to try installing it and see what springs come out. :) | 22:43 |
| mugsie | kfox1111: not really... I just had never done a horizon plugin before, and kept twisting mydself into knots | 22:43 |
| kfox1111 | ah. | 22:43 |
| ekarlso | like 15 mins away :P | 22:43 |
| mugsie | am I right in saying stravagner is an oil town? | 22:44 |
| *** jmcbride has joined #openstack-dns | 22:44 | |
| mugsie | so, what - 15e a pint? :P | 22:44 |
| *** jmcbride has quit IRC | 22:46 | |
| ekarlso | mugsie: more like 6-7 I think | 22:46 |
| mugsie | oh. I can live with that | 22:46 |
| ekarlso | :P | 22:46 |
| mugsie | when are we going again? | 22:46 |
| mugsie | soon, isnt it? | 22:46 |
| *** jmcbride has joined #openstack-dns | 22:46 | |
| ekarlso | mugsie: :P april | 22:47 |
| kfox1111 | hmm... so the docs that talk about updating settings to enable the dashboard don't seem to be refering to /etc/openstack-dashboard/local_settings... | 22:49 |
| *** MentalRay has quit IRC | 22:50 | |
| ericpeterson | kfox1111 - correct, all though I think it is possible to make changes there too | 22:51 |
| kfox1111 | parse errors if I do it there. | 22:51 |
| kfox1111 | /usr/share/openstack-dashboard/openstack_dashboard/settings.py? | 22:51 |
| ericpeterson | yes, there should be an existing line which you can change. Are you working with the changes that just merged? | 22:52 |
| ericpeterson | I can speak to those, at least | 22:52 |
| kfox1111 | arg. no module named pbr.version. :/ | 22:53 |
| kfox1111 | yup. just git pulled right before trying. | 22:53 |
| ericpeterson | https://github.com/openstack/designate/blob/master/contrib/designate-dashboard/README.rst has the changes I made | 22:54 |
| kfox1111 | yeah. thats what I'm following. it would be good to put in that its /usr/share/openstack-dashboard/openstack_dashboard/settings.py, not /etc/openstack-dashboard/local_settings like all the other settings go. | 22:55 |
| kfox1111 | also it still doesn't seem to work because of the pbr thing. | 22:55 |
| kfox1111 | not sure why, but I think redhat disables that stuff in the rpm's. | 22:55 |
| ericpeterson | I'm not sure where the pbr thing is comming from, that is surprising me | 22:56 |
| ericpeterson | I know pbr, but surprised that is occuring | 22:56 |
| kfox1111 | hmm... maybe I missed the requirements install step. let me retry that. | 22:57 |
| kfox1111 | yeah. that installed pbr. | 22:57 |
| kfox1111 | hmm... ImportError: Could not import settings 'openstack_dashboard.settings' (Is it on sys.path? Is there an import error in the settings file?): No module named enabled | 22:58 |
| ericpeterson | you can also run horizon in more of a developer mode too, which sometimes help with this. it uses a venv to isolate add/removing/upgrading python libs | 22:58 |
| kfox1111 | ok, thats cause there isn't an enabled anywhere in the code... | 22:59 |
| ericpeterson | my change, where I forget the init.py file.... did you grab that too? | 22:59 |
| * ericpeterson is ashamed of that change / ommission | 22:59 | |
| kfox1111 | not sure. can you point me at it again? | 23:00 |
| ericpeterson | https://review.openstack.org/#/c/167812/ | 23:00 |
| ericpeterson | you can also just touch that file, wherever the designatedashboard was installed too | 23:01 |
| kfox1111 | so, enabled is an empty module? | 23:01 |
| *** pk has joined #openstack-dns | 23:01 | |
| *** pk has quit IRC | 23:01 | |
| ericpeterson | no, there are other files, but I forgot to include the init.py file - which makes python loader barf | 23:02 |
| *** penick has quit IRC | 23:02 | |
| ericpeterson | import designatedashboard.enabled -> this assumes there will be an __init__.py file in that folder, for the import to succeed | 23:02 |
| ericpeterson | and I am stupid, and had that file in my sandbox during the first commit / forgot that file | 23:03 |
| kfox1111 | Ah, there we go. | 23:03 |
| ericpeterson | does it work? | 23:03 |
| kfox1111 | a dashboard! :) | 23:03 |
| kfox1111 | testing... | 23:04 |
| * ericpeterson has a small moment of pride | 23:04 | |
| *** penick has joined #openstack-dns | 23:04 | |
| kfox1111 | awesome. :) | 23:04 |
| ericpeterson | sorry about this mixup kfox1111, I think you might have pulled from git at a window of time where stuff was messed up | 23:05 |
| kfox1111 | with those couple of tweaks it slid right into rdo juno's dashboard. :) | 23:05 |
| kfox1111 | no, its all cool. better to smooth it all out now, then after release. :) | 23:05 |
| kfox1111 | very very cool. thanks for the help. :) | 23:06 |
| ericpeterson | sure, no prob. We are working on designate deployment right now, I expect some more changes / bug fixes as we find issues. :D | 23:06 |
| kfox1111 | you hooked up with the RDO folks? it would be awesome to be able to just yum install this stuff. :) | 23:07 |
| ekarlso | ericpeterson: what version you deploying ? | 23:07 |
| ericpeterson | most of our services are j-2 I think. I am in charge of horizon, and I ship it from master ;) | 23:08 |
| ekarlso | ah ok | 23:08 |
| ericpeterson | but I'll dig into the service stuff too, as needed / etc | 23:08 |
| kfox1111 | would be very nice to get it to the point where a stock openstack dashboard will detect if designate's there and load it up on demand. | 23:10 |
| kfox1111 | man. this is awesome. :) | 23:10 |
| *** chlong has joined #openstack-dns | 23:10 | |
| kfox1111 | now I just gota talk with the dns folks here and get a subdomain pointing at designate managed servers. | 23:11 |
| mugsie | well, it kind of does that now | 23:11 |
| ericpeterson | kfox1111 - I agree, but there is contention between the horizon team owning tons of code, and understanding it all. | 23:11 |
| kfox1111 | awesome job all. :) | 23:11 |
| mugsie | ericpeterson: with the new "big tent" stuff - whats happening with horizon? | 23:11 |
| ericpeterson | good question ;) | 23:12 |
| mugsie | (are you doing a neutron and moving everything to out of tree, or taking all and sundry? ) | 23:12 |
| mugsie | ah | 23:12 |
| kfox1111 | yeah. they should be able to manage some code in their stuff though that says, if 'dns' in servicecatalog: import designatemodule, enableit() | 23:12 |
| mugsie | there is the if in service catalog, enable_it() | 23:12 |
| mugsie | so if you remove the service entry it *should* dissappear | 23:12 |
| kfox1111 | ah. | 23:13 |
| mugsie | its just the installing it bit thats not done (for us - as we are out of tree) | 23:13 |
| ericpeterson | the problem is horizon developers / cores don't have deep service domain logic. so if you want to add / change new designate horizon code - is the horizon team the best folks to review that? | 23:13 |
| mugsie | I would love to move it in tree though | 23:13 |
| kfox1111 | yeah. | 23:14 |
| mugsie | ericpeterson: oh, I understand that entirely | 23:14 |
| mugsie | but there has to be a better way than what we are doing now :( | 23:14 |
| mugsie | it puts people off stright away | 23:15 |
| ericpeterson | that might be an ok model, but there would be like 80 horizon core devs, and some of them shouldn't be able to approve certain parts of code. like a horizon-designate core should not +2 code from the swift-horizon feature set | 23:15 |
| ericpeterson | there certainly should be a better way, agreed | 23:15 |
| mugsie | i am thinking it may be a topic for vancouver | 23:16 |
| mugsie | in the cross project sessions | 23:16 |
| ericpeterson | I attend most of the horizon stuff, will be in vancouver | 23:16 |
| mugsie | cool, so will I (hopefully) | 23:16 |
| ericpeterson | also, familiar with most of the horizon cores | 23:16 |
| ericpeterson | (I am not a core - I am a operator who codes) | 23:17 |
| mugsie | thats even better :) - you see the impact of theoretical decissions we make :) | 23:17 |
| ericpeterson | sometimes it's best not to see some changes coming ;) | 23:18 |
| mugsie | yeah. I know exactly what is in store for us when we upgrade our designate service next | 23:18 |
| mugsie | it gives me the shivers | 23:18 |
| kfox1111 | feature request... while its perfectly understandable to specify the name like 'bar.kfox-test.cloud.pnnl.gov.', its kind of not obvious to non dns experts and very verbose. It would be nice if it supported entry of a name with no dots, and it fills out the rest. | 23:22 |
| ekarlso | kfox1111: soo, v2 client that does.. | 23:22 |
| kfox1111 | oh cool. :) | 23:22 |
| ekarlso | if your name doesn't contain a ending "." or so when creating a record | 23:22 |
| kfox1111 | one step ahead of me. :) | 23:22 |
| ekarlso | it will fetch the domain first before doing post to records | 23:23 |
| ekarlso | based on the domain id you give | 23:23 |
| kfox1111 | cool. | 23:23 |
| kfox1111 | so when the dashboard switches over to v2, then it will get that for free. | 23:23 |
| ekarlso | :p | 23:23 |
| ekarlso | kfox1111: whatever decade that happens :D | 23:24 |
| mugsie | we need to fix the client not to require the ids as well :/ | 23:24 |
| mugsie | it is a PITA most of the time | 23:24 |
| ekarlso | mugsie: ... uh that's a fun one with pools and all... | 23:24 |
| mugsie | ekarlso: sounds like a problem for whoever is doing the v2 client :D | 23:25 |
| ekarlso | or hmm | 23:25 |
| kfox1111 | I think once you get this stuff into rdo, there will be enough of us users kicking the tires, that one or two will come along to help polish up some of the rough edges. | 23:25 |
| ekarlso | shouldn't be too hard | 23:25 |
| ekarlso | *will get to it tmrw* | 23:25 |
| mugsie | kfox1111: yeah | 23:25 |
| kfox1111 | if its not me... ;) | 23:25 |
| ekarlso | mugsie: why isn't your changes in yet for the api ? :( | 23:25 |
| *** andrewbogott_afk is now known as andrewbogott | 23:25 | |
| mugsie | need to find people in RH to take it into RDO | 23:25 |
| kfox1111 | depends on just how much of our users ask. :) | 23:25 |
| mugsie | ekarlso: waiting on a +A | 23:26 |
| mugsie | it is a massive amount of code | 23:26 |
| kfox1111 | so, with all the big tent stuff going on, how is the incubation stuff working? | 23:35 |
| kfox1111 | is designate still incubated or what? | 23:36 |
| mugsie | kfox1111: incubation no longer exists from what I can tell | 23:36 |
| kfox1111 | hmm... yeah. http://git.openstack.org/cgit/openstack/governance/tree/reference/projects.yaml no longer references incubation at all. | 23:37 |
| *** chlong has quit IRC | 23:37 | |
| kfox1111 | which kind of sucks. it was usefull to some extent to tell what was mature and what wasnt. :/ | 23:37 |
| *** chlong has joined #openstack-dns | 23:37 | |
| kfox1111 | hmm.. though the file has tags now... | 23:38 |
| kfox1111 | so I'm not sure how thats really any different. what does it take for designate to get the 'integrated-release' tag? | 23:39 |
| mugsie | kfox1111: no one else is getting that tag | 23:44 |
| kfox1111 | hmm.... ok. | 23:44 |
| mugsie | it is in place for Kilo, then it is just a historic thing | 23:44 |
| mugsie | to be replaced with tags to be defined | 23:44 |
| mugsie | (is my understanding) | 23:44 |
| kfox1111 | ah. "we'll replace something thats kind of working, with nothing for now and maybe implement something later" cute. | 23:45 |
| mugsie | I am open to being corrected by anyone who has a better unstanding | 23:45 |
| mugsie | yeh | 23:45 |
| mugsie | my read is the didnt want to make the hard calls, and just delegated the responsiblities | 23:45 |
| mugsie | the TC* | 23:46 |
| mugsie | welcoming everyone is great | 23:46 |
| mugsie | but things like who is in horizon, devstack, etc wasnt worked out | 23:46 |
| mugsie | this is where people start shouting that I am wrong usually :D | 23:48 |
| * mugsie </end_rant> | 23:48 | |
| kfox1111 | yeah. I think the zaqar thing broke them. :/ | 23:49 |
| ericpeterson | the horizon thing is historical. if it was re-done today..... every project would be a horizon plugin / addon | 23:49 |
| kfox1111 | they were doing pretty well until then. | 23:49 |
| ericpeterson | have any of you tried to keep rabbit running to support openstack? the idea that openstack has a great messaging system is a little concerning | 23:50 |
| kfox1111 | I guess part of it is the swift vs radosgw thing too. its kind of a hard one to call too. we're using rados gw in all of our clouds now. | 23:51 |
| mugsie | ericpeterson: yeah, we run a rabbit cluster for our designate install | 23:51 |
| kfox1111 | havent had a problem with rabbit yet. | 23:51 |
| mugsie | we have had network partition issues before, but mostly its OK | 23:51 |
| mugsie | we had an older version a while back that was awfull | 23:52 |
| kfox1111 | yeah, havent ran a rabbit cluster yet. if you dig into their comments about how it works, it scares the crap out of me. | 23:52 |
| ericpeterson | we constantly have rabbit problems, especially with the ha setup and making sure everyone stays connected correctly | 23:52 |
| kfox1111 | "you should run a fast backend network with them close together so you don't run into issues". | 23:52 |
| kfox1111 | aka, the code's racy/buggy, and we don't care (tm) | 23:52 |
| mugsie | they assume the network never partitions | 23:53 |
| kfox1111 | yeah. and thats a really bad assumption. | 23:53 |
| mugsie | we have had situations where we had to nuke a whole node just to get it to rejoin | 23:53 |
| ericpeterson | I think the best advice at operators meetup was - run rabbit on single node, and don't try to make it resillent | 23:53 |
| kfox1111 | in our experience, often the networks one of the first things to go. :/ | 23:53 |
| mugsie | as someone who runs on cloud, on neutron - yes, x1000 yes | 23:54 |
| ericpeterson | yep kfox1111 | 23:54 |
| *** ericpeterson has quit IRC | 23:55 | |
| mugsie | right, home time | 23:55 |
| mugsie | later o/ | 23:55 |
| kfox1111 | later. thanks again for all the help. | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
