23:59:18 #startmeeting Ansible Azure Working Group 23:59:18 Meeting started Wed Feb 20 23:59:18 2019 UTC. 23:59:18 This meeting is logged and archived in a public location. 23:59:18 The chair is nitzmahone. Information about MeetBot at http://wiki.debian.org/MeetBot. 23:59:18 Useful Commands: #action #agreed #halp #info #idea #link #topic. 23:59:18 The meeting name has been set to 'ansible_azure_working_group' 23:59:37 #chair yuwei yungezz zikalino8265 23:59:37 Current chairs: nitzmahone yungezz yuwei zikalino8265 00:00:30 So what have we for today? I just merged the VMSS instance PR, there's a merge conflict on the other one Zim asked about 00:00:40 Great 00:00:52 I think PRs 00:00:54 yes, it's resolved now and tests passing 00:01:00 +1 00:01:02 hey 00:01:04 * nitzmahone looks 00:01:08 #chair jborean93 00:01:08 pls press merge button :-) https://github.com/ansible/ansible/pull/51462 00:01:08 Current chairs: jborean93 nitzmahone yungezz yuwei zikalino8265 00:01:12 Hi all 00:01:21 Hi 00:01:39 #chair Kylie_ 00:01:39 Current chairs: Kylie_ jborean93 nitzmahone yungezz yuwei zikalino8265 00:01:49 and i have 2 supersimple modules: 00:01:51 https://github.com/ansible/ansible/pull/45076 00:02:07 https://github.com/ansible/ansible/pull/45072 00:02:15 I have Pr for role definition, but test cannot run because of test service principals need owners permission 00:02:16 these are basically mysql / postgresql setting modules 00:02:53 https://github.com/ansible/ansible/pull/52468 00:03:03 yungezz i think we will have to disable these tests. may be quite difficult / unsafe to have these permissions in ci 00:03:21 I have tested it in my private azure account 00:03:53 in my opinion that should be sufficient 00:04:39 if those credentials can't be given then it's good to have the tests but just set it to disabled in the aliases 00:04:47 allows people to manually run the tests if they wish 00:04:59 Ok 00:05:59 Same story for role assignments, I will disable test also later 00:06:26 https://github.com/ansible/ansible/pull/51320 00:07:26 please also add hd insight cluster to the list: https://github.com/ansible/ansible/pull/52612 00:07:56 woof, some hefty ones in there 00:08:24 hehe, start with mysql/postgresql config, it's supersimple ;-) 00:08:31 #action nitzmahone to review 45076, 45072, 52468, 51320, 52612 00:08:40 btw, we encountered one issue: https://github.com/ansible/ansible/pull/52615 00:08:42 to discuss 00:09:20 basically i have added missing "tags", but as "tags" is in a generic doc fragment, we can't have "version_added" 00:09:49 so we have to merge with error and without "version_added" 00:10:26 ah, so that module didn't reference the tags doc fragment before (or support the tag args)? 00:10:26 if we would like to have "version_added" anyway, we would have to add "tags" manually, which is a kind of weird. 00:10:34 no 00:11:21 Yeah, that's a weird corner case... Lemme talk to the sanity test owner about that, but I suspect the answer will be "just merge with the failure" 00:11:48 `version_added` is less important now that we have version-specific module docs anyway 00:11:55 well, mattclay suggested it can be merged with failure, but later error won't be visible 00:12:32 Right; that test only runs on CI- looking for new args that weren't documented as such (by comparing to existing args in target branch). 00:12:45 So once it's merged, no problem 00:12:45 there are appropriate comments / discussion in the pr 00:12:45 nitzmahone: I mentioned it to sivel and he was going to look into it, but I don't think we need to hold up that PR due to the lack of ability to specify version_included on a docs fragment inclusion. 00:13:09 ok, thanks! 00:13:10 That's kinda where I'm at too- if we didn't have 2.8-specific docs, I'd be more worried about it, but since we do, meh 00:13:36 so yeah, I'll go ahead and merge if it's otherwise ready 00:13:55 Ideally we'd be able to set version_added when including a docs fragment for the case when the fragment and module are pre-existing and we're just adding a new feature to a module. 00:13:58 i have another question here. i am just improving documentation. and i was wondering if it would make sense to have some azure specific sanity tests introduced? 00:14:08 such as? 00:14:32 mostly naming conventions in the documentation, etc. 00:15:30 mattclay could speak better to the sanity of making such a change 00:15:38 checking samples etc 00:16:03 zikalino8265: Can you give a specific example of something you'd like to check for? 00:16:43 IIUC there's always some extra overhead to running sanity tests, so we'd need to keep them pretty cheap and well-targeted (since they'll run against the whole codebase, would have to internally filter) 00:17:30 for instance resource id format in samples, and naming conventions 00:18:01 well, just an idea. i could propose something more detailed 00:18:13 now i was just wondering if it would make sense 00:18:25 zikalino8265: OK, so you want to enforce guidelines on the documentation within Azure modules? 00:18:37 yes 00:19:45 Matt and Jordan, you received some PRs for new Azure resources and the team will submit more to ensure we could enable customers use below resources. List them here and you could have a full picture. 00:19:45 A. DB - CosmosDB (https://docs.microsoft.com/en-us/azure/cosmos-db/introduction) 00:19:45 A.1 Double check we have full support for SQL, MySQL, PostgreSQL 00:19:45 B. Redis Cache (https://azure.microsoft.com/en-us/services/cache/) - Catherine, please note the service is renamed as Azure Cache for Redis. Please ensure use this name in the doc part. 00:19:45 C. Virtual networking peering (https://docs.microsoft.com/en-us/azure/virtual-network/virtual-network-peering-overview) 00:19:45 D. CDN (https://azure.microsoft.com/en-us/services/cdn/) 00:19:45 E. Azure Lab Services (https://docs.microsoft.com/en-us/azure/lab-services/lab-services-overview) 00:19:46 F. Service Bus (https://docs.microsoft.com/en-us/azure/service-bus/) 00:19:46 G. HDInsight (https://azure.microsoft.com/en-us/services/hdinsight/) 00:20:53 zikalino8265: That's probably something we can do, as long as the defined rules are consistent. If they're only suggestions and need to be ignored sometimes then they're probably not a good fit for sanity tests. 00:21:36 ... also probably can't require any new dependencies 00:21:44 ok, i will put azure guidelines proposal together 00:23:44 Oh yungezz- I saw the discussion on https://github.com/ansible/ansible/issues/51864 00:24:06 I *think* most of those should be trivial to fetch off the models we already have 00:25:04 Yes I will take care of the inventory fix 00:25:32 ok cool, thanks 00:25:59 * nitzmahone probably should have done a final apples-to-apples test between those two 00:27:13 Kylie_ was that just FYI, or something you wanted to discuss? 00:29:44 Anything else for today? 00:29:46 FYI. A heads-up for upcoming PRs to see whether we could catch up 3/20 feature freeze. 00:29:55 ah gotcha 00:30:03 No from me 00:30:14 Yes, the earlier the better, as we're all flying low to get our own features merged too ;) 00:30:48 That’s all. Thanks 00:30:57 I've just got 2 questions before we end sorry 00:31:03 fire away 00:31:43 I've had a report this morning that the latest `oathlib` release is incompatible with msrest.credentials 00:31:58 The user had to downgrade to the 2.1.0 library of oathlib to get it working again 00:32:16 I haven't had time to test it out but thought it best to mention in case you wanted to pass it along to the relevant people 00:32:35 We can talk that with Laurent 00:32:47 Our sdk owner 00:33:09 and the other? 00:33:16 Could you pls share the error? 00:33:33 All I got was an auth failure, downgraded had it working again 00:33:42 Got it 00:33:52 The other was a question around the paramiko requirement in `azure-cli-core` 00:34:02 Got it 00:34:17 hrm, that does seem an odd requirements 00:34:32 It's pinned at `2.4.0` and the Tower folks were wonder what this was used for. If it's just for generating ssh key pairs through the cli tool and if so can they just omit that 00:35:04 No idea, we will need ask cli team 00:35:15 And back to you 00:35:24 no worries 00:35:38 looks like ACS client uses it too 00:35:54 but yeah, I know why Tower wants to omit that ;) 00:36:14 Why 00:36:35 Red Hat security team doesn't want us to package paramiko 00:36:40 It's got ... some issues 00:37:05 Oh 00:37:49 OK, if nothing else for today, closing in 5.. 00:37:53 4.. 00:37:57 3.. 00:38:00 2.. 00:38:03 1.. 00:38:06 #endmeeting