| *** russellb has joined #openstack-dns | 00:04 | |
| *** EricGonczer_ has quit IRC | 00:18 | |
| *** ryanpetrello has quit IRC | 00:26 | |
| *** eandersson has joined #openstack-dns | 00:29 | |
| eandersson | Is there a configuration to change the default queue name for Designate Sink? :D | 00:32 |
|---|---|---|
| *** ryanpetrello has joined #openstack-dns | 00:46 | |
| *** CaptTofu has quit IRC | 00:49 | |
| *** pfreund has quit IRC | 00:49 | |
| *** ToMiles has quit IRC | 00:49 | |
| *** pfreund has joined #openstack-dns | 00:51 | |
| *** CaptTofu has joined #openstack-dns | 00:51 | |
| *** ryanpetrello has quit IRC | 00:53 | |
| *** ryanpetrello has joined #openstack-dns | 00:56 | |
| *** rmoe has quit IRC | 00:56 | |
| *** amcrn_ has quit IRC | 00:57 | |
| *** ryanpetrello has quit IRC | 01:00 | |
| *** ToMiles has joined #openstack-dns | 01:13 | |
| *** rmoe has joined #openstack-dns | 01:14 | |
| *** rjrjr has quit IRC | 01:22 | |
| *** EricGonczer_ has joined #openstack-dns | 01:35 | |
| *** rjrjr has joined #openstack-dns | 01:44 | |
| *** EricGonczer_ has quit IRC | 01:44 | |
| rjrjr | kiall: i am getting an error whenever i attempt to patch my patch (create a patch with my original patch as a dependency) - http://paste.openstack.org/show/143088/ | 01:45 |
| rjrjr | no matter. i'll wait for the pool manager changes to go in, then i'll submit the bug fixes. | 01:46 |
| *** rjrjr_ has quit IRC | 01:58 | |
| *** nosnos has joined #openstack-dns | 02:02 | |
| *** jmcbride has joined #openstack-dns | 02:04 | |
| *** jmcbride has quit IRC | 02:07 | |
| *** richm has quit IRC | 02:25 | |
| *** richm has joined #openstack-dns | 02:27 | |
| *** EricGonczer_ has joined #openstack-dns | 02:27 | |
| *** richm has quit IRC | 02:28 | |
| *** ToMiles has quit IRC | 02:37 | |
| *** ryanpetrello has joined #openstack-dns | 02:51 | |
| *** ToMiles has joined #openstack-dns | 02:56 | |
| *** ToMiles_ has joined #openstack-dns | 03:01 | |
| *** ToMiles has quit IRC | 03:01 | |
| *** puck has quit IRC | 03:07 | |
| *** nosnos has quit IRC | 03:29 | |
| *** LinStatSDR has joined #openstack-dns | 03:31 | |
| *** puck has joined #openstack-dns | 03:34 | |
| *** EricGonczer_ has quit IRC | 03:38 | |
| *** EricGonczer_ has joined #openstack-dns | 03:39 | |
| *** puck has quit IRC | 03:40 | |
| *** GonZo2K has quit IRC | 03:43 | |
| *** EricGonczer_ has quit IRC | 03:43 | |
| *** puck has joined #openstack-dns | 03:46 | |
| *** EricGonczer_ has joined #openstack-dns | 03:52 | |
| *** EricGonczer_ has quit IRC | 04:00 | |
| *** nosnos has joined #openstack-dns | 04:17 | |
| *** ryanpetrello has quit IRC | 04:28 | |
| *** LinStatSDR has quit IRC | 04:56 | |
| *** veena has joined #openstack-dns | 05:11 | |
| veena | Hi, am facing an issue in Icehouse setup on 2 nodes. The VMs that are spawned on the second compute node don't get the IP. But the controller shows the IP is assigned. Please let me know if anybody faced the same and if you could help me. I'll share the config files. | 05:11 |
| veena | Should dnsmasq be running on the compute node also? | 05:12 |
| *** mwagner_lap has quit IRC | 06:19 | |
| *** mwagner_lap has joined #openstack-dns | 06:32 | |
| *** nihilifer_ has quit IRC | 06:43 | |
| *** jmcbride has joined #openstack-dns | 06:46 | |
| jmcbride | mugsie kiall_ can one of you review this change again? https://review.openstack.org/#/c/133712/ (made some minor improvements/fixes). | 07:05 |
| ekarlso- | jmcbride: isn't it the middle of the night over there ? :) | 07:08 |
| jmcbride | yep, just can't sleep tonight. | 07:10 |
| jmcbride | :) | 07:10 |
| jmcbride | ekarlso: What time is it in your region? | 07:11 |
| ekarlso- | 0800 | 07:11 |
| ekarlso- | 0811 really | 07:11 |
| *** k4n0 has joined #openstack-dns | 07:21 | |
| *** ToMiles_ has quit IRC | 07:23 | |
| *** ToMiles has joined #openstack-dns | 07:23 | |
| openstackgerrit | Endre Karlson proposed openstack/designate: Support secondary zones https://review.openstack.org/133682 | 07:27 |
| openstackgerrit | Endre Karlson proposed openstack/designate: MDNS part of Secondary zones https://review.openstack.org/133683 | 07:27 |
| openstackgerrit | Endre Karlson proposed openstack/designate: MDNS part of Secondary zones https://review.openstack.org/133683 | 07:36 |
| ekarlso- | wth | 07:51 |
| ekarlso- | oh well | 07:51 |
| ekarlso- | jmcbride: go to bed :p | 07:51 |
| *** veena has quit IRC | 07:52 | |
| jmcbride | wish I could :/ | 07:53 |
| *** LinStatSDR has joined #openstack-dns | 07:57 | |
| *** LinStatSDR has quit IRC | 08:06 | |
| *** nihilifer has joined #openstack-dns | 08:25 | |
| *** jmcbride has quit IRC | 08:42 | |
| *** jordanP has joined #openstack-dns | 08:45 | |
| *** veena has joined #openstack-dns | 08:55 | |
| *** ToMiles has quit IRC | 08:57 | |
| *** nosnos has quit IRC | 09:00 | |
| *** veena has quit IRC | 09:10 | |
| *** jordanP has quit IRC | 10:54 | |
| *** jordanP has joined #openstack-dns | 10:54 | |
| *** redixin has joined #openstack-dns | 11:01 | |
| *** rjrjr_ has joined #openstack-dns | 12:06 | |
| rjrjr_ | kiall/mugsie: will you get an opportunity to review https://review.openstack.org/#/c/131656/ today? | 12:10 |
| *** k4n0 has quit IRC | 12:28 | |
| *** ryanpetrello has joined #openstack-dns | 12:43 | |
| *** EricGonczer_ has joined #openstack-dns | 12:54 | |
| mugsie | rjrjr_: yup | 13:01 |
| *** mwagner_lap has quit IRC | 13:38 | |
| *** ryanpetrello has quit IRC | 13:38 | |
| *** ryanpetrello has joined #openstack-dns | 13:39 | |
| *** EricGonczer_ has quit IRC | 13:43 | |
| *** ryanpetrello_ has joined #openstack-dns | 14:06 | |
| *** ryanpetrello has quit IRC | 14:08 | |
| *** ryanpetrello_ is now known as ryanpetrello | 14:08 | |
| *** richm has joined #openstack-dns | 14:14 | |
| *** ryanpetrello has quit IRC | 14:18 | |
| *** ryanpetrello has joined #openstack-dns | 14:19 | |
| *** mwagner_lap has joined #openstack-dns | 14:29 | |
| openstackgerrit | Merged openstack/designate: Add docs for MX, SSHFP, SPF and SRV record set types https://review.openstack.org/133712 | 14:45 |
| *** alokj has joined #openstack-dns | 14:48 | |
| *** richm has quit IRC | 14:50 | |
| rjrjr | kiall: i see the merge of my changes failed. do you need me to fix that? | 14:54 |
| rjrjr | merge of https://review.openstack.org/#/c/131656/ failed. | 14:55 |
| kiall_ | rjrjr: yep, you'll need to rebase the patch | 14:55 |
| *** kiall_ is now known as Kiall | 14:55 | |
| *** Kiall has quit IRC | 14:55 | |
| *** Kiall has joined #openstack-dns | 14:55 | |
| rjrjr | so just run 'git rebase master' from the directory containing the patch? | 14:55 |
| Kiall | Simplest way is to `git fetch origin && git rebase origin/master` | 14:56 |
| rjrjr | from the directory containing the patch? | 14:56 |
| Kiall | yes | 14:56 |
| Kiall | It will "fail", and you'll need to fix the conflicts (`git status` will show you them), then `git add conflicting/file.py` and finally `git rebase --continue && git review` | 14:56 |
| Kiall | (don't commit, or amend the commit during the rebase_ | 14:56 |
| openstackgerrit | Ron Rickard proposed openstack/designate: Pool Manager https://review.openstack.org/131656 | 15:03 |
| Kiall | rjrjr: any chance you have a list of the conflicting files/sections? Makes a re-review of just those bits easier :) | 15:04 |
| rjrjr | kiall: the only file in conflict was contrib/devstack/lib/designate_plugins/backend-bind9_pool | 15:04 |
| Kiall | Ah - Fair enough :) | 15:05 |
| *** richm has joined #openstack-dns | 15:05 | |
| rjrjr | appreciate the procedures. adding it to my repertoire. | 15:07 |
| *** betsy has quit IRC | 15:08 | |
| Kiall | Pro-Tip: Never, Ever (*ever*) walk away from your computer or otherwise get distracted mid-rebase.. If you forget, then just start working again, you're in for a world of pain ;) | 15:08 |
| Kiall | (The amount of work I've lost due to ^ is embarrassingly large ;)) | 15:09 |
| rjrjr | all this time, i've been 'git clone'ing the master. diff the files that changed and adding them back one at a time. this was much, much easier. 8^) | 15:10 |
| rjrjr | i think this weekend i'm going to sit down and play with git a bit and learn how to use this tool better. 8^) | 15:11 |
| rjrjr | thanks for the tip! | 15:12 |
| rjrjr | so, does it take 2 core reviewers to get code merged? | 15:14 |
| *** EricGonczer_ has joined #openstack-dns | 15:15 | |
| Kiall | Yep, 2 core's need to +2 and the second one also +A's | 15:18 |
| Kiall | (Theres no technical restriction on that, just standard process for OpenStack) | 15:18 |
| rjrjr | kiall: what is the next step for pool manager? as soon as this code is merged, i have a patch that will fix the create domain and recordset issues. | 15:20 |
| rjrjr | i know the code in central is not streamlined for pool manager. we make way to many calls to the backend when we just need to get the database updated and make one call to the backend. | 15:21 |
| Kiall | Next steps.. Well, I think we get the second gate added for bind9 (I'm still gently encouraging the infra folks to +A it ;)), then start moving PowerDNS over to Pools, and begin stripping out the non-pools stuff.. | 15:21 |
| rjrjr | i can work on that, although it might be easier when we integrate pool manager properly and remove the direct backend/MDNS support. | 15:21 |
| Kiall | Yea, Agreed.. We should do it in 1 go I reckon.. | 15:22 |
| rjrjr | i also can work on unit tests for the pool manager objects. | 15:22 |
| Kiall | We should sync up with the others later today to figure out next steps | 15:22 |
| *** ToMiles has joined #openstack-dns | 15:23 | |
| rjrjr | okay. if it is okay, i'll start "refactoring" central for streamlining integration with pool manager. i can also start testing different use cases for pool manager (multiple pools, multiple backend servers, etc.) | 15:23 |
| *** GonZo2K has joined #openstack-dns | 15:24 | |
| rjrjr | there is plenty to do here, just want to make sure i'm working smart. | 15:24 |
| Kiall | Sure - Go for it.. Keep in mind that as part of that, we'll want to fix bug 1392762 once and for all.. i.e. we'll need to rejigger more than just if we call a backend vs poolmanager :) | 15:25 |
| uvirtbot | Launchpad bug 1392762 in designate "Concurrent requests can cause designate-central to lock up" [High,In progress] https://launchpad.net/bugs/1392762 | 15:25 |
| *** jmcbride has joined #openstack-dns | 15:25 | |
| *** vinod1 has joined #openstack-dns | 15:25 | |
| rjrjr | i'm not a big fan of our long methods in central. want to break them up more logically and i see lots of duplicate code in central that can be eliminated. | 15:26 |
| Kiall | Yea, moving stuff into distinct endpoints is something we've talked about a few times.. but the diff is going to be HUGE, probably too much churn for this cycle TBH | 15:27 |
| rjrjr | okay. i'll suffer through leaving it alone. 8^) | 15:27 |
| *** GonZo2K has quit IRC | 15:28 | |
| rjrjr | if someone verifies the change, i'll send up my fixes for some of the issues vinod saw yesterday. | 15:28 |
| rjrjr | verifies = merges | 15:30 |
| *** paul_glass has joined #openstack-dns | 15:31 | |
| *** GonZo2K has joined #openstack-dns | 15:35 | |
| vinod1 | rjrjr: Your change is on the way to being merged | 15:36 |
| rjrjr | vinod1: thanks! | 15:36 |
| *** timsim has joined #openstack-dns | 15:38 | |
| openstackgerrit | Merged openstack/designate: Pool Manager https://review.openstack.org/131656 | 15:41 |
| Kiall | \o/ | 15:42 |
| *** GonZo2K has quit IRC | 15:42 | |
| *** jmcbride has quit IRC | 15:44 | |
| *** EricGonczer_ has quit IRC | 15:48 | |
| openstackgerrit | Ron Rickard proposed openstack/designate: Pool Manager Bug Fixes https://review.openstack.org/138406 | 15:48 |
| Kiall | ^ .. +70, -51 .. That's a much more manageable review ;) | 15:49 |
| * Kiall would really prefer to get the gate job running before merging too many more changes though.. | 15:50 | |
| rjrjr | https://launchpad.net/bugs/1392762 is this one related to the gate job? | 15:52 |
| uvirtbot | Launchpad bug 1392762 in designate "Concurrent requests can cause designate-central to lock up" [High,In progress] | 15:52 |
| Kiall | Yep, it's why the rally job is still failing, and the fix would have been to change a pile of code that's going to change again anyway when pools becomes the 1 and only backend method.. | 15:53 |
| *** betsy has joined #openstack-dns | 15:53 | |
| Kiall | Basically, inside the @transaction wrapped methods, we can't do anything that would cause eventlet to context switch.. That includes anything containing the "yield" statement (e.g. wrap_backend_call()), and calling out to services over RPC | 15:54 |
| *** ToMiles has quit IRC | 15:55 | |
| *** paul_glass has quit IRC | 15:55 | |
| *** ToMiles has joined #openstack-dns | 15:55 | |
| Kiall | So, once the calls the old backends are gone, and we have far fewer calls to pool manager, we can move those outside of the @transaction wrapped methods (either by creating a new private method for the part needing @transaction, or a new decorator above @transaction that makes the call to pool manager.. I haven't looked at how doable that second route is though... | 15:57 |
| *** EricGonczer_ has joined #openstack-dns | 15:59 | |
| rjrjr | just so it's clear, i can move forward with eliminating other backend code from central and streamline on just pool manager? | 16:00 |
| Kiall | I *think* so.. | 16:00 |
| *** jmcbride has joined #openstack-dns | 16:00 | |
| rjrjr | someone else is working on creating another pool manager backend (pdns)? what about the other backends? | 16:00 |
| *** vinod1 has quit IRC | 16:01 | |
| Kiall | rpc can be removed, fake should be ported, nsd + dynect should be trivial ports, and IPA, well, I'm really not sure how we can handle IPA at all. | 16:02 |
| rjrjr | rich mentioned at the summit he would be interested in helping with that. | 16:02 |
| rjrjr | richm: you around? | 16:02 |
| *** nihilifer has quit IRC | 16:03 | |
| richm | rjrjr: yes | 16:03 |
| rjrjr | would you have time to help us port the IPA backend to a pool manager backend? | 16:03 |
| *** jmcbride has quit IRC | 16:04 | |
| *** jmcbride has joined #openstack-dns | 16:04 | |
| rjrjr | pool manager is in a state where it is workable and any port can be tested. | 16:04 |
| *** EricGonczer_ has quit IRC | 16:04 | |
| Kiall | I think the biggest difficulty with that is, FreeIPA has no support for receiving zone contents from another server via AXFR - The suggested route I've seen for that kinda thing is direct modification of the FreeIPA LDAP tree.. | 16:05 |
| *** EricGonczer_ has joined #openstack-dns | 16:05 | |
| *** EricGonczer_ has quit IRC | 16:05 | |
| Kiall | https://fedorahosted.org/bind-dyndb-ldap/ is the BIND plugin FreeIPA uses | 16:05 |
| *** vinod1 has joined #openstack-dns | 16:05 | |
| Kiall | Suggested migration path from a standard DNS server -> IPA DNS: https://fedorahosted.org/bind-dyndb-ldap/wiki/Migration | 16:08 |
| richm | IPA can only be a master, not a slave | 16:12 |
| *** redixin has quit IRC | 16:12 | |
| *** EricGonczer_ has joined #openstack-dns | 16:12 | |
| *** Kiall has quit IRC | 16:13 | |
| *** kiall_ has joined #openstack-dns | 16:13 | |
| *** kiall_ has quit IRC | 16:13 | |
| richm | It seems to be quite common in the private cloud case that the organization wants to keep their identity and security infrastructure as the "master" | 16:14 |
| *** Kiall has joined #openstack-dns | 16:14 | |
| *** vinod1 has quit IRC | 16:14 | |
| *** Kiall is now known as Guest6992 | 16:14 | |
| richm | I know in Keystone we run into the same problem - the customer wants Keystone to treat the corporate identity data as the "master" and read-only (through Keystone - that is, keystone user-create should error) | 16:15 |
| *** Guest6992 has quit IRC | 16:15 | |
| *** Guest6992 has joined #openstack-dns | 16:15 | |
| richm | We have IPA customers that, for example, want Windows Server to be the authoritative source of DNS - so we have to tell them how to manually create the SRV and other DNS records to make IPA work | 16:16 |
| *** Guest6992 is now known as Kiall | 16:16 | |
| *** LinStatSDR has joined #openstack-dns | 16:22 | |
| *** Kiall has quit IRC | 16:24 | |
| rjrjr | does timsim's agent proposal help with this? | 16:25 |
| *** Kiall has joined #openstack-dns | 16:25 | |
| rjrjr | does timsim's agent proposal help with this? | 16:26 |
| *** Kiall is now known as Guest45286 | 16:26 | |
| rjrjr | (repeated because kiall seems to be having issues.) | 16:26 |
| timsim | link: http://docs-draft.openstack.org/30/131530/3/check/gate-designate-specs-docs/e0ad530/doc/build/html/specs/kilo/new-agent.html | 16:27 |
| timsim | :) | 16:27 |
| richm | timsim: What is "An agent deployed on the Master DNS server"? For example, if the Master DNS server is a Windows machine, does that mean a Windows service would need to be installed? | 16:34 |
| Guest45286 | I suppose it wouldn't be too dissimilar to https://github.com/stackforge/designate-msdnsagent .. (An agent that runs on windows to hook up MS DNS ;)) | 16:35 |
| *** Guest45286 has quit IRC | 16:35 | |
| *** Guest45286 has joined #openstack-dns | 16:35 | |
| *** Guest45286 is now known as Kiall | 16:35 | |
| *** alokj has quit IRC | 16:36 | |
| timsim | richm: more than likely, yes. If there's an issue getting it running directly on the Windows server, you could have a Linux server that ran the agent, with a backend that called into the Windows one. | 16:36 |
| richm | interesting - do you guys have private cloud customers who use that? | 16:36 |
| richm | because we invariably get a lot of pushback from Windows admins who want _nothing_ to be installed on their precious Windows boxes | 16:36 |
| richm | for example, 389 has a password sync feature that is a "plugin" to Active Directory - it is always a struggle to get that installed | 16:37 |
| *** alokj has joined #openstack-dns | 16:37 | |
| Kiall | I'd want nothing on my windows servers if I had any ;) | 16:37 |
| timsim | As long as there is some sort of API that would be accessible from the Agent (python) you should be fine. | 16:37 |
| *** alokj is now known as Guest46927 | 16:37 | |
| Kiall | There is WMI - But no decent libs for Windows | 16:37 |
| Kiall | Sorry | 16:37 |
| Kiall | no decent libs to Linux | 16:37 |
| * timsim shrugs | 16:39 | |
| *** jmcbride has quit IRC | 16:40 | |
| timsim | I don't really know how it works, so I'm not sure what the interface would be, if there is one. | 16:40 |
| richm | What is the purpose of the "Master DNS server agent"? To translate updates/zone transfers into a format that is understandable by the DNS server? I guess in the case of Windows, if you could treat Windows DNS as read-write by Designate, that would be fine. | 16:43 |
| richm | In the case of IPA, you would end up writing something that was like a wsgi frontend to the code that's already in the designate ipa backend | 16:44 |
| timsim | Essentially, yeah. PowerDNS for example, doesn't do inbound IXFR, so the agent could do that, and then talk into PowerDNS. For bind9, you could maintain a typical master-slave setup, with the agent getting zone data, writing files and doing rndc addzone/delzone. | 16:44 |
| *** EricGonczer_ has quit IRC | 16:45 | |
| Kiall | richm: " that is a "plugin" to Active Directory" <-- I can see AD being considered the Holy Grail which shall not be touched ;) | 16:45 |
| Kiall | I think the core goal behind timsim's spec is about providing a tree like structure to spread out the load of managing more zones than anyone should manage ;) | 16:45 |
| timsim | The agent would get the information from designate via DNS protocol, and then chat with backends plugin-like any way they need to. If the DNS server can dance with MiniDNS, no problem. But as the deployment gets larger, you might not want a massive cluster of MiniDNS nodes :P | 16:45 |
| timsim | Yeah, that. | 16:46 |
| *** jmcbride has joined #openstack-dns | 16:46 | |
| richm | or if the master DNS server can't do [AI]XFR | 16:47 |
| timsim | Yep. As long as there is *some* interface that the agent can talk to the DNS server. | 16:50 |
| richm | and this is because minidns is to be the one source of sending updates, rather than going through the old backend code? | 16:50 |
| *** EricGonczer_ has joined #openstack-dns | 16:51 | |
| mugsie | richm: yup | 16:52 |
| richm | Is the old backend code going to be supported at all? | 16:54 |
| mugsie | not in the near future. however, the old backend code *should* be loadable into the new agent | 16:55 |
| *** GonZo2K has joined #openstack-dns | 16:55 | |
| *** rjrjr has quit IRC | 16:59 | |
| *** rjrjr_ has quit IRC | 17:00 | |
| Kiall | rjrjr: new gate job for bind + poolmgr merging ;) | 17:01 |
| Kiall | Oh.. He's gone | 17:01 |
| Kiall | If it's broken, all changes will be blocked ;) | 17:02 |
| *** vinod1 has joined #openstack-dns | 17:08 | |
| *** nkinder has joined #openstack-dns | 17:09 | |
| *** GonZo2K has quit IRC | 17:10 | |
| *** LinStatSDR has quit IRC | 17:12 | |
| *** Guest46927 is now known as alokj_ | 17:15 | |
| *** rmoe has quit IRC | 17:15 | |
| *** alokj_ has quit IRC | 17:22 | |
| Kiall | Nearly.. gate-designate-dsvm-powerdns: NOT_REGISTERED | 17:24 |
| Kiall | gate-designate-dsvm-bind9: NOT_REGISTERED | 17:24 |
| *** rmoe has joined #openstack-dns | 17:32 | |
| *** vinod2 has joined #openstack-dns | 17:34 | |
| *** vinod1 has quit IRC | 17:35 | |
| *** betsy has quit IRC | 17:35 | |
| *** betsy has joined #openstack-dns | 17:43 | |
| *** jmcbride has quit IRC | 17:44 | |
| *** EricGonczer_ has quit IRC | 17:49 | |
| *** rjrjr has joined #openstack-dns | 17:51 | |
| *** boris-42 has joined #openstack-dns | 17:52 | |
| *** jordanP has quit IRC | 17:52 | |
| *** betsy has quit IRC | 17:53 | |
| rjrjr | ummm.... what happened? | 17:56 |
| rjrjr | https://review.openstack.org/#/c/138406/ | 17:58 |
| mugsie | looks like the pypi mirror in rax died | 18:01 |
| mugsie | but the devstack test passed :) | 18:02 |
| rjrjr | rechecking... | 18:02 |
| *** vinod2 has quit IRC | 18:03 | |
| richm | I guess the agent-per-master will work - but (in the ipa case at least, if not pdns, bind, etc.) it should use the existing backend code as much as possible | 18:07 |
| *** jmcbride has joined #openstack-dns | 18:08 | |
| richm | note that this is going to be an additional burden on packagers and on end users - to have another package to manage, install, configure, and monitor on DNS server systems | 18:08 |
| rjrjr | the agent solution might be geared to solving another issue entirely and it if doesn't fit for this, we should look at other solutions. especially if adopting it for IPA add extra overhead that isn't needed. | 18:11 |
| richm | If minidns is going to be the one and only way to push designate updates to other DNS servers, then you have to have an agent | 18:13 |
| richm | for DNS servers that can't/won't allow XFR | 18:14 |
| Kiall | rjrjr: all the OpenStack pypi mirrors died it seems | 18:14 |
| rjrjr | kiall: mugsie mentioned that. i'm rerunning the check now. just powerdns needs to pass. 8^) | 18:15 |
| Kiall | Yep, All good now :) | 18:15 |
| eandersson | Morning. | 18:15 |
| rjrjr | is the bind check using pool manager? the powerdns check is not? | 18:16 |
| eandersson | Is there a way to change the queue that Designate-Sink consumes from? We are seeing Ceilometer trying to consume from the same queue. | 18:16 |
| Kiall | Correct | 18:16 |
| rjrjr | i guess i can look at the logs and check this myself. | 18:16 |
| Kiall | eandersson: yes, there's a notification_topics config option | 18:16 |
| eandersson | I changed notification_topic to designate.notifications and manually added the correct bidning for now. | 18:16 |
| eandersson | but won't that add the wrong binding? | 18:16 |
| eandersson | or am I missing something obvious? | 18:17 |
| Kiall | The binding is done from the other side actually, you'll want to update nova+neutron to "notification_topics = notifications, notifications-designate" so it sends the notification to 2 queues | 18:18 |
| eandersson | ohh | 18:18 |
| eandersson | I didnt even think about that :D | 18:18 |
| Kiall | Yea, It's backwards from normal AMQP ;) | 18:18 |
| eandersson | And I am supposed to be an AMQP expert lol | 18:18 |
| rjrjr | kiall: i just read the description for https://review.openstack.org/#/c/138406/ | 18:18 |
| rjrjr | is the proposed solution going to eliminate the problem or just make it less likely to happen? | 18:19 |
| Kiall | Normal AMQP, the consumer sets up the queue and binds it to an exchange, and with oslo.messaging, it's the other way around | 18:19 |
| Kiall | rjrjr: was that the right link? | 18:19 |
| rjrjr | https://bugs.launchpad.net/designate/+bug/1392762 | 18:19 |
| uvirtbot | Launchpad bug 1392762 in designate "Concurrent requests can cause designate-central to lock up" [High,In progress] | 18:19 |
| rjrjr | sorry, wrong link. | 18:19 |
| Kiall | Well, the fix is to tighten the TX window to only code that can't cause a context switch | 18:20 |
| Kiall | anything that can cause a eventlet context switch can trigger the lockip :'( | 18:20 |
| Kiall | lockup* | 18:20 |
| rjrjr | so, nothing in storage can trigger this? | 18:21 |
| Kiall | None of code in storage will trigger this issue today | 18:21 |
| Kiall | (That I know of) | 18:21 |
| rjrjr | basically, we are going to only have transactions for storage calls. | 18:21 |
| Kiall | Well, more than just storage calls.. let me grab a central method and "fix it" (while breaking everything of course ;)) | 18:22 |
| rjrjr | i understand actually. | 18:23 |
| rjrjr | we just don't want to decorate the calls to pool manager. | 18:23 |
| rjrjr | and possibly other code. | 18:24 |
| Kiall | http://paste.openstack.org/show/XsRia0TfOaRMOvkrQ49I/ | 18:25 |
| *** harmw_ is now known as harmw | 18:25 | |
| Kiall | note the "** XXX ONLY IN THE DB. XXX ***" bit, and the move of the TX to the nested/private method | 18:25 |
| rjrjr | that makes sense. | 18:26 |
| Kiall | (some of that code should probably live outside the TX, like the policy check) | 18:26 |
| Kiall | But other methods have stuff which all needs to be in 1 TX.. That was just a simple method to "fix" while breaking everything ;) | 18:26 |
| Kiall | Anyway, you can see why pools only makes this easier? Since it's basically always just a call to | 18:27 |
| Kiall | self.pool_manager_api.ive_updated_this_domain(context, domain) | 18:27 |
| rjrjr | yes. definitely easier. | 18:27 |
| Kiall | (which could possibly be done as a decorator, like @notification, so as to avoid code bloat) | 18:28 |
| rjrjr | final question: will https://review.openstack.org/#/c/138406/ get reviewed or should i include it in my solution? | 18:28 |
| Kiall | I'd recommend keeping commits small + focused at this point, that way we can review them much faster and know there done when they fix 1 issue :) | 18:29 |
| Kiall | (But you'll want to get familiar with chaining git patchsets etc to make your life easier for that ;)) | 18:29 |
| *** EricGonczer_ has joined #openstack-dns | 18:30 | |
| rjrjr | so, the patch i sent will be considered even though the rally test is still failing? | 18:30 |
| *** vinod1 has joined #openstack-dns | 18:31 | |
| rjrjr | (you mentioned earlier that ideally we wouldn't fix anything else until https://bugs.launchpad.net/designate/+bug/1392762 is fixed.) | 18:31 |
| uvirtbot | Launchpad bug 1392762 in designate "Concurrent requests can cause designate-central to lock up" [High,In progress] | 18:31 |
| rjrjr | lovely bot there. 8^) | 18:31 |
| Kiall | Yea, the rally job has never succeeded ;) | 18:32 |
| *** paul_glass has joined #openstack-dns | 18:32 | |
| Kiall | (It's new, and is how we discovered bug 1392762) | 18:32 |
| uvirtbot | Launchpad bug 1392762 in designate "Concurrent requests can cause designate-central to lock up" [High,In progress] https://launchpad.net/bugs/1392762 | 18:32 |
| rjrjr | okay. i'll assume 138406 will get reviewed and build on it. | 18:33 |
| Kiall | Yep :) | 18:34 |
| timsim | Kiall: The rally job succeeded on the latest patchset of this somehow: https://review.openstack.org/#/c/108859/ lol | 18:34 |
| rjrjr | timsim: luck! sometimes the bug is not exercised. | 18:35 |
| timsim | I thought it was pretty funny :P | 18:35 |
| rjrjr | i've seen some of the create and delete records succeed and figured eventually i would submit a patch that passed rally. cool to see it happen. | 18:36 |
| Kiall | WOOO .. Bug fixed, closing as "Can't Reproduce" ;) | 18:36 |
| Kiall | Yea, It's literally blind luck that it passed.. | 18:36 |
| Kiall | nondeterministic bugs are my personal favorite bugs - recheck enough times and you've fixed it :P | 18:37 |
| timsim | Just "wat" things. | 18:38 |
| *** EricGonczer_ has quit IRC | 18:39 | |
| *** EricGonczer_ has joined #openstack-dns | 18:39 | |
| rjrjr | so, this is where the patch grows, moving to pool manager only means changes to central and devstack. this will impact the gate job too, correct? not too bad if powerdns pool backend is available... | 18:42 |
| rjrjr | you know, enough talking about it. i'm going to start making the changes now. expect something tonight/early tomorrow morning (daytime for our EU friends.) | 18:43 |
| Kiall | Yea, I'm going to do the PDNS pools bit now.. | 18:46 |
| Kiall | I'm also thinking "bind9" should just go, and "bind9_pools" backend should be renamed "bind9" .. Having 2 seems wasteful ;) | 18:49 |
| *** amcrn has joined #openstack-dns | 18:49 | |
| *** EricGonczer_ has quit IRC | 18:55 | |
| *** betsy has joined #openstack-dns | 18:59 | |
| *** mikedillion has joined #openstack-dns | 19:07 | |
| *** Kiall has quit IRC | 19:18 | |
| *** Kiall has joined #openstack-dns | 19:18 | |
| *** Kiall is now known as Guest40425 | 19:18 | |
| *** Guest40425 has quit IRC | 19:36 | |
| *** Kiall_ has joined #openstack-dns | 19:37 | |
| rjrjr | i'm going to be breaking the tsigkey methods by only updating the storage. | 19:40 |
| *** Kiall_ has quit IRC | 19:42 | |
| rjrjr | (we don't have a pool backend solution for tsigkeys yet.) | 19:42 |
| *** Kiall_ has joined #openstack-dns | 19:42 | |
| *** jmcbride has quit IRC | 19:44 | |
| *** amcrn has quit IRC | 19:45 | |
| *** EricGonczer_ has joined #openstack-dns | 19:49 | |
| *** jmcbride has joined #openstack-dns | 19:55 | |
| *** jmcbride has quit IRC | 19:56 | |
| *** jmcbride has joined #openstack-dns | 19:56 | |
| *** EricGonczer_ has quit IRC | 20:04 | |
| *** EricGonc_ has joined #openstack-dns | 20:04 | |
| *** vinod1 has quit IRC | 20:05 | |
| *** jmcbride has quit IRC | 20:07 | |
| *** vinod1 has joined #openstack-dns | 20:17 | |
| *** jmcbride has joined #openstack-dns | 20:28 | |
| *** GonZo2K has joined #openstack-dns | 20:46 | |
| *** mikedillion has quit IRC | 21:32 | |
| *** openstackgerrit has quit IRC | 21:34 | |
| *** openstackgerrit has joined #openstack-dns | 21:34 | |
| *** ChanServ sets mode: +v openstackgerrit | 21:34 | |
| *** jmcbride has quit IRC | 21:49 | |
| *** mikedillion has joined #openstack-dns | 21:53 | |
| *** jmcbride has joined #openstack-dns | 22:01 | |
| *** paul_glass1 has joined #openstack-dns | 22:04 | |
| *** paul_glass has quit IRC | 22:07 | |
| *** ryanpetrello has quit IRC | 22:18 | |
| *** mikedillion has quit IRC | 22:23 | |
| *** ryanpetrello has joined #openstack-dns | 22:23 | |
| *** EricGonc_ has quit IRC | 22:33 | |
| *** boris-42 has quit IRC | 22:37 | |
| *** ryanpetrello has quit IRC | 22:40 | |
| *** paul_glass1 has quit IRC | 22:43 | |
| *** ryanpetrello has joined #openstack-dns | 22:44 | |
| *** timsim has quit IRC | 22:45 | |
| *** bitblt has joined #openstack-dns | 22:46 | |
| *** ryanpetrello has quit IRC | 22:48 | |
| *** vinod1 has quit IRC | 22:59 | |
| *** jmcbride has quit IRC | 23:00 | |
| *** ryanpetrello has joined #openstack-dns | 23:06 | |
| *** EricGonczer_ has joined #openstack-dns | 23:10 | |
| *** ryanpetrello has quit IRC | 23:20 | |
| *** jmcbride has joined #openstack-dns | 23:21 | |
| *** jmcbride has quit IRC | 23:27 | |
| *** mwagner_lap has quit IRC | 23:31 | |
| *** bitblt has quit IRC | 23:34 | |
| rjrjr | vinod1: your comments, while valid, didn't have anything to do with the bug fixes. | 23:36 |
| *** ryanpetrello has joined #openstack-dns | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!