| *** sjmc7__ has joined #heat | 00:04 | |
| *** tellesnobrega_ has joined #heat | 00:04 | |
| *** zhenzanz has joined #heat | 00:05 | |
| *** zhenzanz_ has joined #heat | 00:06 | |
| *** julienvey has quit IRC | 00:06 | |
| *** sjmc7_ has quit IRC | 00:07 | |
| *** rdo has quit IRC | 00:08 | |
| *** achanda has quit IRC | 00:08 | |
| *** zhenzanz has quit IRC | 00:09 | |
| *** zhenzanz_ is now known as zhenzanz | 00:09 | |
| *** Qiming has joined #heat | 00:09 | |
| *** rdo has joined #heat | 00:10 | |
| *** JayJ has quit IRC | 00:14 | |
| *** Qiming has quit IRC | 00:18 | |
| *** pmallya has quit IRC | 00:19 | |
| *** achanda has joined #heat | 00:26 | |
| *** dims_ has quit IRC | 00:30 | |
| *** achanda has quit IRC | 00:37 | |
| *** alexpilotti has quit IRC | 00:51 | |
| *** sjmc7__ has quit IRC | 00:54 | |
| *** achanda has joined #heat | 00:57 | |
| *** rdo has quit IRC | 01:02 | |
| *** zaneb has quit IRC | 01:03 | |
| *** rdo has joined #heat | 01:03 | |
| *** Qiming has joined #heat | 01:05 | |
| *** gokrokve has quit IRC | 01:06 | |
| *** julienvey has joined #heat | 01:17 | |
| *** sgordon has quit IRC | 01:20 | |
| *** julienvey has quit IRC | 01:22 | |
| *** achanda has quit IRC | 01:22 | |
| *** tellesnobrega_ has quit IRC | 01:23 | |
| *** tellesnobrega_ has joined #heat | 01:29 | |
| *** dimsum__ has joined #heat | 01:30 | |
| *** achanda has joined #heat | 01:31 | |
| *** achanda has quit IRC | 01:35 | |
| *** dimsum__ has quit IRC | 01:36 | |
| *** dimsum__ has joined #heat | 01:37 | |
| *** hdd has joined #heat | 01:40 | |
| *** dimsum__ has quit IRC | 01:41 | |
| openstackgerrit | Merged openstack/heat: Use correct IDs when collecting multipart config parts https://review.openstack.org/136906 | 01:42 |
|---|---|---|
| *** dimsum__ has joined #heat | 01:53 | |
| *** Yanyanhu has joined #heat | 02:03 | |
| *** mc__ has joined #heat | 02:04 | |
| *** serg_melikyan has quit IRC | 02:07 | |
| *** mc__ has quit IRC | 02:10 | |
| *** serg_melikyan has joined #heat | 02:10 | |
| openstackgerrit | Merged openstack/heat: Fix permission bits for source file. https://review.openstack.org/137074 | 02:12 |
| *** apporc has joined #heat | 02:14 | |
| *** nosnos has joined #heat | 02:14 | |
| *** mc__ has joined #heat | 02:14 | |
| *** erkules_ has joined #heat | 02:28 | |
| *** serg_melikyan has quit IRC | 02:30 | |
| *** erkules has quit IRC | 02:30 | |
| *** sdake_ has joined #heat | 02:31 | |
| *** sdake_ has quit IRC | 02:31 | |
| *** sdake_ has joined #heat | 02:31 | |
| *** sgordon_ has quit IRC | 02:32 | |
| *** serg_melikyan has joined #heat | 02:34 | |
| openstackgerrit | Ethan Lynn proposed openstack/heat: Fix package name when using heat-db-setup in rhel7 https://review.openstack.org/136763 | 02:36 |
| *** erkules_ is now known as erkules | 02:36 | |
| *** alexheneveld has quit IRC | 02:46 | |
| *** gokrokve has joined #heat | 02:52 | |
| *** gokrokve has quit IRC | 02:52 | |
| *** dimsum__ has quit IRC | 02:52 | |
| *** gokrokve has joined #heat | 02:52 | |
| *** gokrokve has quit IRC | 02:53 | |
| *** gokrokve has joined #heat | 02:53 | |
| *** gokrokve has quit IRC | 02:54 | |
| *** gokrokve has joined #heat | 02:55 | |
| *** gokrokve has quit IRC | 02:59 | |
| *** ramishra has joined #heat | 02:59 | |
| *** gokrokve has joined #heat | 02:59 | |
| *** gokrokve has quit IRC | 02:59 | |
| *** gokrokve has joined #heat | 02:59 | |
| *** gokrokve has quit IRC | 03:00 | |
| *** gokrokve has joined #heat | 03:00 | |
| huangtianhua | shardy: would you please to review https://review.openstack.org/#/c/130107/ and https://review.openstack.org/#/c/132625/ thx:) | 03:00 |
| *** nosnos has quit IRC | 03:01 | |
| *** LiJiansheng has joined #heat | 03:01 | |
| asalkeld | huangtianhua, he said he was travelling | 03:03 |
| *** Marga_ has quit IRC | 03:03 | |
| asalkeld | not sure where too | 03:03 |
| asalkeld | and for how long | 03:03 |
| huangtianhua | ah :) | 03:03 |
| asalkeld | he should have filled us in more :-O | 03:04 |
| *** hdd has quit IRC | 03:04 | |
| *** Marga_ has joined #heat | 03:04 | |
| *** gokrokve has quit IRC | 03:05 | |
| *** gokrokve has joined #heat | 03:05 | |
| *** gokrokve has quit IRC | 03:06 | |
| huangtianhua | asalkeld: about the neutron network constraint, http://paste.openstack.org/show/138570/ how do you think? it seems a litter odd:) | 03:07 |
| *** gokrokve has joined #heat | 03:07 | |
| asalkeld | looking | 03:07 |
| *** gokrokve has quit IRC | 03:07 | |
| asalkeld | Qiming, come chat where the cool kids are;) | 03:07 |
| Qiming | okay | 03:08 |
| *** gokrokve has joined #heat | 03:08 | |
| *** gokrokve has quit IRC | 03:08 | |
| Qiming | we were talking about the design/roadmap for autoscaling separation work | 03:08 |
| *** ananta has joined #heat | 03:09 | |
| asalkeld | huangtianhua, isn't there some similar discovery code somewhere | 03:09 |
| ananta | o/ | 03:09 |
| ananta | good morning fellas | 03:09 |
| asalkeld | hey ananta | 03:09 |
| ananta | Hi asalkeld | 03:09 |
| asalkeld | good morning apac | 03:10 |
| Qiming | the suggestion is to make AS RPC calls work first | 03:10 |
| ananta | asalkeld: when is the poc evaluation planned? | 03:11 |
| huangtianhua | asalkeld: not found, but for sahara resources, I found there is a validation of checking if running on neutron | 03:11 |
| asalkeld | ananta, we are probably going to have to push to next week | 03:11 |
| asalkeld | ananta, because of the US holidays | 03:11 |
| huangtianhua | asalkeld: but I think it's incorrect in using nova network situation | 03:12 |
| *** gokrokve has joined #heat | 03:12 | |
| asalkeld | i wish for the concept that were the same we had generic resources | 03:12 |
| asalkeld | so the user didn't have to put neurton in their templates | 03:13 |
| asalkeld | ananta, are you going to do the small poc (based on zane's repo)? | 03:13 |
| asalkeld | it would be nice to compare on equal ground | 03:14 |
| *** serg_mel_ has joined #heat | 03:14 | |
| ananta | asalkeld: no, our poc is based on current heat engine impl | 03:15 |
| asalkeld | ananta, and no plans? | 03:15 |
| asalkeld | :-( | 03:15 |
| asalkeld | maybe i need to try | 03:15 |
| ananta | asalkeld: I think as long as we canvey the idea it should be fine | 03:15 |
| asalkeld | ok | 03:16 |
| ananta | asalkeld: we as a team wanted to understand Heat implementation, without which we couldn't have done it.... | 03:16 |
| asalkeld | huangtianhua, we can use a nova network with nova server, right? | 03:16 |
| asalkeld | ananta, i understand | 03:16 |
| huangtianhua | asalkeld: yes | 03:17 |
| *** gokrokve has quit IRC | 03:17 | |
| ananta | I mean understanding where Heat is now and how convergence would be brought in was the challenge we were looking for | 03:17 |
| asalkeld | ananta, we are coming at it from the other point of view | 03:17 |
| *** serg_melikyan has quit IRC | 03:17 | |
| asalkeld | ananta, understanding heat - but trying to figure out convergence | 03:17 |
| ananta | asalkeld: true...we couldn't have done better with our limited understanding | 03:18 |
| stevebaker | ananta: The problem is that your solution can't be evaluated against the alternatives unless it is implemented in zaneb's simple POC. | 03:18 |
| asalkeld | stevebaker, it can - it | 03:19 |
| asalkeld | stevebaker, it can - it's just more difficult | 03:19 |
| stevebaker | asalkeld: true, but not directly with the POC test suite | 03:19 |
| ananta | nonetheless, it is important to convey the ideas | 03:19 |
| *** serg_mel_ has quit IRC | 03:19 | |
| asalkeld | stevebaker, ananta i'll try find some time to try that (convert what you have to zanes harness) | 03:19 |
| asalkeld | no promises tho' | 03:20 |
| asalkeld | stevebaker, who is on pto (zane/shardy) | 03:20 |
| stevebaker | ananta: the risk is that if it is more difficult to evaluate then your solution might be overlooked, that would be a real shame | 03:20 |
| ananta | stevebaker: its just not only the code with our PoC...we do have docs for it | 03:20 |
| ananta | I understand it would be bit difficult, but as folks who know Heat engine, I do believe that its not that difficult | 03:21 |
| stevebaker | ananta: that is all good, but my point remains | 03:21 |
| ananta | stevebaker: I think if we can work on getting the idea out of PoC we shouldn' have any problems | 03:22 |
| stevebaker | ananta: do you have a particular objection to reimplementing in the POC? I realise the rework would be annoying | 03:22 |
| *** killer_prince has quit IRC | 03:23 | |
| ananta | stevebaker: yes, the rework , for which we will have to again spend efforts in understanding that PoC | 03:24 |
| *** Marga_ has quit IRC | 03:24 | |
| elynn | morning guys :) | 03:24 |
| ananta | stevebaker: and I do feel that its much bettter to compare against the current impl | 03:25 |
| ananta | where convergence takes us from where we are | 03:25 |
| ananta | stevebaker: we have our own reason for choosing the way we went, so the focus for us should be on how we can convey the idea and converge on the idea | 03:26 |
| ananta | stevebaker: I will definitely give it a try | 03:26 |
| ananta | thanks for bringing that up | 03:27 |
| *** harlowja is now known as harlowja_away | 03:29 | |
| asalkeld | ananta, def _get_ready_resources(self): is really get resources that need to be converged? | 03:29 |
| ananta | asalkeld: yes | 03:29 |
| asalkeld | cool, thx | 03:29 |
| stevebaker | ananta: the POC was created so that ideas could be tried with minimal effort, so in theory it wouldn't take much coding. I think its fair to say that zaneb is the heat-core leading the convergence effort, so if he suggests doing something it would be very much to your benefit to give it a try. inc0 has contributed his own approach to the POC, and has grown the test cases, so it is becoming a useful focus point | 03:30 |
| stevebaker | for evaluating approaches | 03:30 |
| stevebaker | ananta: especially for building a test suite which exercises all those corner cases | 03:31 |
| ananta | stevebaker: zaneb's poc started much after we started doing it...so there was no way for us to work on top of it | 03:32 |
| ananta | as I said I do inderstand your concern | 03:33 |
| ananta | asalkeld: there are few minor work remaining in our poc so I wanted to know when the evaluation was planned | 03:34 |
| ananta | asalkeld: by next week we should be ready | 03:34 |
| asalkeld | ananta, it was going to be this week, but i forgot about all the US holidays | 03:34 |
| stevebaker | ananta: yes, you're right, the POC happened after | 03:34 |
| ananta | asalkeld: np | 03:34 |
| asalkeld | so maybe we should plan for next | 03:34 |
| *** hdd has joined #heat | 03:35 | |
| ananta | lets plan for next week | 03:35 |
| asalkeld | ananta, if/when you do any work on a zane-based poc, let me know please | 03:36 |
| asalkeld | (and like wise, I'll do the same) | 03:37 |
| ananta | asalkeld: sure | 03:37 |
| *** sarob has joined #heat | 03:45 | |
| asalkeld | stevebaker, send some of that cool nz air here please, phew hot-n-sweaty here | 03:46 |
| stevebaker | asalkeld: its hot here, like 19 degrees! | 03:49 |
| asalkeld | that's cool | 03:49 |
| asalkeld | it's not that hot here (only 29) but super humid | 03:49 |
| *** sarob has quit IRC | 03:49 | |
| ananta | stevebaker: cooler than in here 23 degree C :) | 03:51 |
| *** dimsum__ has joined #heat | 03:52 | |
| *** ananta has quit IRC | 03:54 | |
| stevebaker | hmm, actually it got up to 24 today | 03:55 |
| *** dimsum__ has quit IRC | 03:57 | |
| cmyster | morning | 04:03 |
| cmyster | had an issue yesterday with a template that was missing "resources:" but had "outputs:" stack-show gave me this: "output_error": "The specified reference \"private_network\" (in unknown) is incorrect.", | 04:08 |
| cmyster | is this what I am expected in this case ? | 04:08 |
| cmyster | expecting even | 04:09 |
| cmyster | bluh, brb + coffee | 04:09 |
| *** achanda has joined #heat | 04:12 | |
| *** nosnos has joined #heat | 04:16 | |
| *** killer_prince has joined #heat | 04:17 | |
| *** killer_prince is now known as lazy_prince | 04:18 | |
| stevebaker | cmyster: it seems like a nicer error would be good | 04:21 |
| cmyster | I'll open a bug here then, | 04:22 |
| *** asalkeld has quit IRC | 04:22 | |
| cmyster | BTW stevebaker this sort of thing just explodes on older versions | 04:22 |
| cmyster | puking ERROR:... blah blah when I do this, so this might need a back port as well | 04:23 |
| *** asalkeld has joined #heat | 04:24 | |
| stevebaker | cmyster: so we're already better than we were ;) | 04:25 |
| cmyster | stevebaker: moving forward means progress | 04:25 |
| cmyster | progress means that new introduced bugs are better looking ;) | 04:26 |
| cmyster | stevebaker: while at it, if I did not give a description in the outputs (2), one will show me only the value and the other will show me "description": "No description given", | 04:27 |
| cmyster | so the behavior is not the same, I'll add a different minor bug report for that as there is no reason to try both versions of the output in the same stack | 04:29 |
| cmyster | oh wait, sorry that was me not seeing clear, disregard... | 04:30 |
| *** achanda has quit IRC | 04:30 | |
| stevebaker | ok, bbl dinner | 04:32 |
| *** Marga_ has joined #heat | 04:35 | |
| *** mc__ has quit IRC | 04:37 | |
| *** Marga_ has quit IRC | 04:39 | |
| *** sabeen1 has joined #heat | 04:46 | |
| *** kebray has quit IRC | 04:56 | |
| *** LiJiansheng has quit IRC | 05:02 | |
| *** hdd has quit IRC | 05:09 | |
| *** rushiagr_away is now known as rushiagr | 05:10 | |
| *** kopparam has joined #heat | 05:11 | |
| *** ishant has joined #heat | 05:12 | |
| *** apporc has quit IRC | 05:15 | |
| *** kopparam has quit IRC | 05:17 | |
| *** kopparam has joined #heat | 05:17 | |
| elynn | Hi stevebaker , do you have time to review https://review.openstack.org/#/c/135197/ again? | 05:20 |
| *** mc__ has joined #heat | 05:21 | |
| *** nkhare has joined #heat | 05:21 | |
| *** rakesh_hs has joined #heat | 05:27 | |
| *** apporc has joined #heat | 05:28 | |
| *** LiJiansheng has joined #heat | 05:49 | |
| *** vijayagurug has quit IRC | 05:50 | |
| *** gokrokve has joined #heat | 05:50 | |
| *** gokrokve has quit IRC | 05:52 | |
| Qiming | asalkeld, there? | 05:52 |
| asalkeld | Qiming, yip | 05:53 |
| Qiming | again, for autoscaling | 05:53 |
| Qiming | once there is a dedicated engine doing autoscaling | 05:53 |
| Qiming | Heat won't see auto-scaling-group as a nested stack, right? | 05:53 |
| asalkeld | Qiming, it depends on whether you assign the owner stack id | 05:54 |
| asalkeld | (which you probably should) | 05:54 |
| Qiming | in the backgroud, as-engine is creating a stack for the ASG? | 05:55 |
| asalkeld | yip as normal | 05:55 |
| Qiming | so ... for 'normal' use cases, Heat will just ask AS-engine to create an asg, without caring how it is implemented | 05:58 |
| Qiming | Heat may also ask AS-engine to scale up/down an ASG, also without caring the details | 05:58 |
| asalkeld | Qiming, that would be ideal | 05:58 |
| *** rushiagr is now known as rushiagr_away | 05:59 | |
| Qiming | for 'advanced' users, they want to know the ASG details as they can do today: resource-list --nest-depth | 05:59 |
| Qiming | is it okay to break that assumption? | 05:59 |
| asalkeld | Qiming, it's good to keep that if possible | 06:00 |
| *** sanjayu has joined #heat | 06:00 | |
| Qiming | I see, then we will keep that requirement in mind while focusing on getting the simpler cases solved | 06:01 |
| *** k4n0 has joined #heat | 06:03 | |
| *** edmund has quit IRC | 06:06 | |
| *** tomek_adamczewsk has joined #heat | 06:22 | |
| *** apporc has quit IRC | 06:24 | |
| *** ananta has joined #heat | 06:28 | |
| *** ygoto has joined #heat | 06:38 | |
| openstackgerrit | Miguel Grinberg proposed openstack/heat: Consider invalid keywords a template validation error https://review.openstack.org/137290 | 06:57 |
| *** apporc has joined #heat | 07:00 | |
| *** jprovazn has joined #heat | 07:02 | |
| *** rushiagr_away is now known as rushiagr | 07:07 | |
| *** kopparam has quit IRC | 07:19 | |
| *** achanda has joined #heat | 07:31 | |
| *** achanda has quit IRC | 07:36 | |
| *** alexheneveld has joined #heat | 07:37 | |
| *** alexheneveld has quit IRC | 07:41 | |
| *** chris_ has joined #heat | 07:45 | |
| *** pas-ha has joined #heat | 07:49 | |
| *** jstrachan has joined #heat | 07:53 | |
| *** inc0 has joined #heat | 07:55 | |
| *** rdo_ has joined #heat | 07:56 | |
| *** kopparam has joined #heat | 07:57 | |
| *** rdo has quit IRC | 07:59 | |
| pas-ha | morning all | 08:00 |
| openstackgerrit | Qi Zhang proposed openstack/heat: Pickup the region name passed in from heatclient https://review.openstack.org/135521 | 08:03 |
| *** jtomasek has joined #heat | 08:04 | |
| *** tomek_adamczewsk has quit IRC | 08:08 | |
| *** GonZo2K has joined #heat | 08:12 | |
| inc0 | gmornign | 08:13 |
| cmyster | morning | 08:16 |
| *** ifarkas has joined #heat | 08:16 | |
| *** kopparam has quit IRC | 08:25 | |
| *** sorantis has joined #heat | 08:25 | |
| *** kopparam has joined #heat | 08:28 | |
| *** ifarkas has quit IRC | 08:28 | |
| *** ifarkas has joined #heat | 08:28 | |
| *** jcoufal has joined #heat | 08:29 | |
| *** achanda has joined #heat | 08:33 | |
| *** rdo_ has quit IRC | 08:42 | |
| *** rdo has joined #heat | 08:43 | |
| cmyster | should there be a limit to nested template depth when creating a stack ? | 08:45 |
| cmyster | I can call a template from its parent, but it won't call lower template for some reason | 08:46 |
| *** apporc has quit IRC | 08:46 | |
| *** apporc has joined #heat | 08:46 | |
| *** achanda has quit IRC | 08:48 | |
| cmyster | /should/is/ | 08:50 |
| *** tomek_adamczewsk has joined #heat | 08:54 | |
| openstackgerrit | Pavlo Shchelokovskyy proposed openstack/heat: Support volume type in Sahara Node Group Template https://review.openstack.org/127205 | 08:54 |
| openstackgerrit | Pavlo Shchelokovskyy proposed openstack/heat: Support secgroups in Sahara Node group templates https://review.openstack.org/136022 | 08:54 |
| openstackgerrit | Pavlo Shchelokovskyy proposed openstack/heat: Support availability zones in Node Group Templates https://review.openstack.org/136023 | 08:54 |
| *** serg_melikyan has joined #heat | 08:55 | |
| *** rdo has quit IRC | 08:56 | |
| *** rdo has joined #heat | 08:58 | |
| *** jistr has joined #heat | 08:59 | |
| *** zhenzanz has quit IRC | 08:59 | |
| *** jprovazn has quit IRC | 09:01 | |
| shardy | cmyster: there is, see max_nested_stack_depth in heat.conf | 09:02 |
| shardy | cmyster: it's defaulted to 3, which, arguably, is too low in many cases | 09:02 |
| *** rdo has quit IRC | 09:10 | |
| cmyster | oh hi there shardy | 09:12 |
| cmyster | sec lemme see | 09:12 |
| *** derekh has joined #heat | 09:12 | |
| *** rdo has joined #heat | 09:12 | |
| cmyster | shardy: if three then in my case I should see the grandchild created, but its not. also the logs are a bit useless in that regard becase there is no recollection to a limit that I could see | 09:15 |
| *** jprovazn has joined #heat | 09:15 | |
| shardy | cmyster: probably you're not hitting the limit then, it's most likely some other issue, or you'd get a "Recursion depth exceeds 3" error message | 09:19 |
| shardy | https://github.com/openstack/heat/blob/master/heat/engine/stack_resource.py#L133 | 09:19 |
| cmyster | hmmm interesting, I'll hit the logs again | 09:19 |
| *** serg_melikyan has quit IRC | 09:19 | |
| *** rdo has quit IRC | 09:20 | |
| cmyster | no thats not it. | 09:20 |
| cmyster | though shardy after yesterday I stopped trusting myself ;) | 09:20 |
| *** tomek_adamczews1 has joined #heat | 09:22 | |
| *** rdo has joined #heat | 09:22 | |
| *** tomek_adamczewsk has quit IRC | 09:23 | |
| huangtianhua | shardy: around:) would you please to review https://review.openstack.org/#/c/130107/ and https://review.openstack.org/#/c/132625/ thx | 09:25 |
| elynn | Hi shardy :) also this cherry-patch and related patches https://review.openstack.org/#/c/135528/ | 09:26 |
| elynn | heat doesn't support python2.6 in kilo? | 09:28 |
| *** mc__ has quit IRC | 09:31 | |
| shardy | elynn: all projects are dropping python 2.6 support during kilo | 09:33 |
| *** cmyster has quit IRC | 09:35 | |
| *** rdo has quit IRC | 09:35 | |
| inc0 | that will stir the waters... | 09:35 |
| *** cmyster has joined #heat | 09:36 | |
| elynn | wow so openstack kilo release can not be installed in rhel6.x? | 09:37 |
| *** rdo has joined #heat | 09:37 | |
| *** GonZo2K has quit IRC | 09:38 | |
| cmyster | no | 09:39 |
| shardy | that's not strictly true, python 2.7 can be installed via software collections | 09:41 |
| shardy | I can't comment on what vendors will actually support though, just saying it's probably technically possible | 09:41 |
| openstackgerrit | Thomas Herve proposed openstack/heat: Update Barbican resources to match library changes https://review.openstack.org/137315 | 09:48 |
| *** lazy_prince has quit IRC | 09:48 | |
| *** tochi has quit IRC | 09:49 | |
| *** LiJiansheng has quit IRC | 09:53 | |
| cmyster | shardy: if the default depth is 3, should I expect the same when gate is running ? | 09:55 |
| *** LiJiansheng has joined #heat | 09:55 | |
| cmyster | arrg, brb | 09:56 |
| *** jcoufal_ has joined #heat | 09:57 | |
| *** tomek_adamczewsk has joined #heat | 09:58 | |
| *** tomek_adamczews1 has quit IRC | 09:58 | |
| *** Qiming has quit IRC | 09:59 | |
| *** rdo has quit IRC | 09:59 | |
| *** jcoufal has quit IRC | 10:00 | |
| *** blues-man has joined #heat | 10:01 | |
| *** rdo has joined #heat | 10:01 | |
| *** julienvey has joined #heat | 10:09 | |
| openstackgerrit | Qi Zhang proposed openstack/heat: Pickup the region name passed in from heatclient https://review.openstack.org/135521 | 10:12 |
| *** kopparam has quit IRC | 10:14 | |
| openstackgerrit | Qi Zhang proposed openstack/heat: Pickup the region name passed in from heatclient https://review.openstack.org/135521 | 10:17 |
| *** elynn has quit IRC | 10:17 | |
| *** fayablazer has joined #heat | 10:18 | |
| *** mkollaro has joined #heat | 10:19 | |
| *** kopparam has joined #heat | 10:25 | |
| *** kopparam has quit IRC | 10:26 | |
| *** kopparam has joined #heat | 10:27 | |
| *** elynn has joined #heat | 10:30 | |
| *** ygoto has quit IRC | 10:33 | |
| asalkeld | pas-ha, are there no fixed values for volume type in sahara | 10:33 |
| *** KanagarajM has joined #heat | 10:34 | |
| *** stannie has joined #heat | 10:34 | |
| pas-ha | asalkeld, I think that's determined by what volume_types are available in Cinder | 10:35 |
| pas-ha | might worth a check probably | 10:35 |
| asalkeld | so there is no way to validate other that creating really | 10:36 |
| asalkeld | (that's ok, just asking) | 10:36 |
| pas-ha | not sure, if there is such a thing as cinder volume-type-list we may try to validate | 10:36 |
| pas-ha | will check | 10:36 |
| asalkeld | k | 10:37 |
| pas-ha | well, there is cinder type-list in shell. will implement validation then. | 10:38 |
| pas-ha | would it be worth to validate all the other stuff? secgroups are surely discoverable, and availablilty zones too | 10:39 |
| *** sorantis has quit IRC | 10:40 | |
| asalkeld | i don't know pas-ha, it is also expensive | 10:40 |
| asalkeld | lots of calls to services | 10:41 |
| asalkeld | there are pros and cons to it | 10:41 |
| inc0 | when we would validate that? in handle_create or in tpl publishing? | 10:42 |
| *** dkusidlo has quit IRC | 10:42 | |
| asalkeld | inc0, so what is there is validate by fire | 10:43 |
| asalkeld | which is one option | 10:43 |
| asalkeld | but i was hoping there was a static list of volume types | 10:44 |
| asalkeld | - guess not | 10:44 |
| inc0 | is there any other real option? We can validate before we even begin with stack create, but that doesn't mean that real thing won't change when we actually start doing that | 10:44 |
| inc0 | so potentially validation will return false positive, therefore it will be useless | 10:44 |
| asalkeld | brb | 10:44 |
| therve | volume types are defined in cinder configuration. Static but per deployment. | 10:45 |
| inc0 | you can add new volume types | 10:45 |
| inc0 | by conf ofc, but still | 10:45 |
| pas-ha | one surely can create volume-types on the fly - we even have e resource for that :) (may be still on review) | 10:47 |
| pas-ha | but probably only as admin | 10:48 |
| pas-ha | listing them is for everyone | 10:48 |
| pas-ha | yep, cinder type-create XXX is admin-only | 10:49 |
| *** tellesnobrega_ has quit IRC | 10:49 | |
| pas-ha | inc0, we usually try to check such stuff in validate() resource's method, but it is indeed puts some API calls burden | 10:50 |
| pas-ha | on the other hand, it is probably still worth it than failing (huge) stack in the middle of creation | 10:51 |
| inc0 | ehh, I don't like that we fail stack because of one failed resource | 10:51 |
| *** kopparam has quit IRC | 10:51 | |
| pas-ha | well, in this case (wrong input) even convergence would not help | 10:52 |
| *** tellesnobrega_ has joined #heat | 10:52 | |
| inc0 | yes, but convergence can fail part of stack | 10:52 |
| inc0 | branch of graph dependant on this one resource | 10:52 |
| pas-ha | and we must fail. or have some state OMG_I_ONLY_CREATED_HALF_OF_THE_STUFF | 10:52 |
| pas-ha | but having only part of the stack still means you have the full stack failed | 10:53 |
| inc0 | yeah, I know, I just emphasize my disapproval for this approach;) | 10:53 |
| *** Yanyanhu has quit IRC | 10:54 | |
| inc0 | we can cache some api calls | 10:54 |
| *** Qiming has joined #heat | 10:54 | |
| pas-ha | I mean other alternative is keep trying with wrong user input params, keep showing CREATE_IN_PROGRESS and fail only after timeout | 10:54 |
| inc0 | I mean, volume types are common for every user in stack... | 10:54 |
| pas-ha | but they might change on the fly | 10:55 |
| pas-ha | CRUDed by admin | 10:55 |
| inc0 | yeah, my point is, we don't even know if our validation is correct when we start doing actuall creating | 10:55 |
| inc0 | convergence could enable us "retry this part of stack" | 10:56 |
| pas-ha | not completely, sure, but with quite fair degree of confidence | 10:56 |
| *** killer_prince has joined #heat | 10:56 | |
| *** killer_prince is now known as lazy_prince | 10:56 | |
| inc0 | so we set up 99 vms, 1 is failed due to wrong input, we update this singular one | 10:56 |
| pas-ha | yes, but how/when to tell the user that he needs update? | 10:56 |
| *** kopparam has joined #heat | 10:56 | |
| inc0 | we'd need "stack partially complete" status.. | 10:57 |
| inc0 | and way of reporting how things really are right now | 10:57 |
| pas-ha | how convergence would understand that resource creation is failed due to some out-of-band problem or due to wrong params? | 10:57 |
| inc0 | but that would also make stack create not an atomic function | 10:57 |
| inc0 | we can just tell user - hey something is wrong, fix it | 10:58 |
| inc0 | maybe retry create this part of stack is enough | 10:58 |
| pas-ha | we actually had the idea of "created_enough_to_operate" partial state somehwere in summit etherpads | 10:58 |
| inc0 | can we actually make such decission? | 10:59 |
| pas-ha | which? | 10:59 |
| inc0 | if it is "enough to operate" | 10:59 |
| pas-ha | well, user sets a goal | 10:59 |
| inc0 | I'd support "partially complete" and let user decide | 10:59 |
| pas-ha | like a hadoop cluster, I spin up 1000 vms but would like to start diong stuff as soon as I have 100 | 11:00 |
| *** sorantis has joined #heat | 11:00 | |
| inc0 | but we'd need some nifty way of make changes to existing stack...not an atomic one | 11:00 |
| skraynev | pas-ha: I thought, that we had some objections about partially complete stacks... | 11:00 |
| inc0 | so you get "partially create" status, user comes and sees "o, 100 out of 1000 is enough, lets keep these and retry other 900" | 11:01 |
| pas-ha | or pacemaker cluster, where voting for master needs min N/2+1 nodes | 11:01 |
| pas-ha | inc0, yeah, smth like that | 11:02 |
| inc0 | thats why I really don't like that we actually consider create as singular, atomic action :/ | 11:02 |
| inc0 | with all or nothing approach | 11:02 |
| asalkeld | inc0, there are lots of use cases for both approaches (every thing must succeed and partial) | 11:04 |
| asalkeld | no one solution | 11:04 |
| inc0 | asalkeld, thats why I say "let user decide" | 11:04 |
| asalkeld | tho' we brought this up at summit and most were against it | 11:04 |
| shardy | what's wrong with just having a stack status of create failed when it's partially created, then looking at the status of the individual resources to decide if doing an update to reach complete state is needed? | 11:05 |
| shardy | not all resources succeeded, so a top level status of failed is accurate IMO | 11:05 |
| shardy | it's not like we don't expose the per-resource status too | 11:06 |
| asalkeld | yo shardy | 11:07 |
| asalkeld | shardy my fingers are feeling worn out, do you mind running the meeting? | 11:07 |
| asalkeld | too much irc'ing today | 11:08 |
| shardy | asalkeld: Ah, hmm, I forgot about that as I'm travelling this week | 11:08 |
| shardy | asalkeld: Ok, will do | 11:08 |
| asalkeld | o, where are you | 11:08 |
| asalkeld | in a bus/train/plane | 11:08 |
| shardy | asalkeld: In Paris - so it's in ~50mins right? | 11:08 |
| asalkeld | yeah | 11:08 |
| asalkeld | oo, lucky you | 11:08 |
| asalkeld | finally got to paris | 11:09 |
| shardy | asalkeld: Ok, I'll grab some lunch then can run the mtg | 11:09 |
| asalkeld | i'll be there shardy | 11:09 |
| shardy | asalkeld: Heh, yeah made it eventually :) | 11:09 |
| inc0 | asalkeld, we wanted to discuss which convergence approach we finally take today, do you feel like you need to be here for that? | 11:10 |
| asalkeld | inc0, i'll be there | 11:10 |
| inc0 | ok | 11:10 |
| asalkeld | inc0, tho' zane won't | 11:10 |
| asalkeld | and lots of US people on holiday | 11:10 |
| shardy | zane needs to be here to make that decision IMO, so we should defer till next week | 11:10 |
| asalkeld | so I'd suggest we post pone til next week | 11:11 |
| * shardy -> lunch | 11:11 | |
| inc0 | sure, agreed | 11:11 |
| *** mkollaro has quit IRC | 11:11 | |
| inc0 | although Zane made an impression yesterday that he'll be here | 11:11 |
| inc0 | we'll see I guess | 11:12 |
| *** tellesnobrega_ has quit IRC | 11:14 | |
| *** julienvey has quit IRC | 11:16 | |
| asalkeld | any items to add: https://wiki.openstack.org/wiki/Meetings/HeatAgenda | 11:22 |
| *** kitch_ has joined #heat | 11:33 | |
| *** hdd has joined #heat | 11:35 | |
| *** sorantis has quit IRC | 11:36 | |
| *** kitch_ has quit IRC | 11:39 | |
| asalkeld | inc0, are you looking at this: https://github.com/zaneb/heat-convergence-prototype/commit/423087f899606273d7271390270acb2664b8cd46 | 11:39 |
| asalkeld | inc0, zane has pulled your code and applied his new tests to your branch | 11:40 |
| asalkeld | and tests are failing | 11:40 |
| *** kitch_ has joined #heat | 11:40 | |
| *** jamielennox is now known as jamielennox|away | 11:48 | |
| *** hdd has quit IRC | 11:49 | |
| *** viktors|afk is now known as viktors | 11:52 | |
| inc0 | asalkeld, we've spoke about that, there was problems with implementations | 11:53 |
| inc0 | were* | 11:53 |
| inc0 | I also added few tests on my own to my code | 11:53 |
| asalkeld | cool, are they solvable? | 11:54 |
| inc0 | haven't seen any case which isn't | 11:54 |
| asalkeld | ok | 11:54 |
| inc0 | I see that Zane added few more tests so I'll try to port these cases as well | 11:54 |
| inc0 | problem with common testing is that my architecture is stateless and async by design, so its hard to have same test code for sync and async | 11:55 |
| asalkeld | well i am happy you are both getting stuck into it, well done | 11:55 |
| shardy | +1, it's great to see the collaboration :) | 11:55 |
| inc0 | yeah, problem will be when we'll actually have to make decision;) | 11:56 |
| *** kitch_ has quit IRC | 11:56 | |
| inc0 | noone wants to drop his guard;) | 11:56 |
| inc0 | although I'm leaning towards trying to make both approaches at the same time;) | 11:56 |
| inc0 | only mine will be called convergence contionous observer | 11:57 |
| cmyster | shardy: that depth thingy, its not working . I changed to 5 and it created it all. | 11:57 |
| cmyster | nice name inc0 | 11:57 |
| inc0 | and someday will "accidently" start solving create/update/whatever as well as normal convergence engine | 11:57 |
| cmyster | deserves a comic super villain | 11:57 |
| *** sorantis has joined #heat | 11:58 | |
| asalkeld | meeting time in 2 mins | 11:58 |
| inc0 | http://img3.wikia.nocookie.net/__cb20100402161911/lotr/images/f/f5/Eye_of_sauron.jpg | 11:58 |
| shardy | cmyster: Ok, last time I tested it it worked fine, pls raise a bug and we can investigate | 11:58 |
| shardy | I actually thought we had tests which prove it works tbh | 11:59 |
| cmyster | shardy: I am working on it | 11:59 |
| cmyster | but it dod not work so I started investigating | 11:59 |
| cmyster | I built the templates for the test and am testing them manually to make sure I can later automate | 11:59 |
| *** kitch_ has joined #heat | 12:00 | |
| cmyster | time | 12:00 |
| *** dulek has joined #heat | 12:00 | |
| inc0 | thing is Zane wants to have something done asap, and for that his approach is better, its much closer to what heat does right now. I want to have it all in stateless, and thats kinda revolutionary | 12:00 |
| *** kitch_ has quit IRC | 12:00 | |
| asalkeld | meeting time folks | 12:00 |
| *** kopparam has quit IRC | 12:03 | |
| *** kitch_ has joined #heat | 12:06 | |
| *** dulek has quit IRC | 12:06 | |
| *** dulek has joined #heat | 12:06 | |
| *** kopparam has joined #heat | 12:09 | |
| *** rakesh_hs2 has joined #heat | 12:12 | |
| *** rakesh_hs has quit IRC | 12:12 | |
| *** rpothier has joined #heat | 12:20 | |
| *** serg_melikyan has joined #heat | 12:20 | |
| *** blues-man has quit IRC | 12:23 | |
| *** rpothier has quit IRC | 12:24 | |
| *** serg_melikyan has quit IRC | 12:24 | |
| *** julienvey has joined #heat | 12:27 | |
| *** KanagarajM has quit IRC | 12:28 | |
| *** julienvey has quit IRC | 12:31 | |
| *** kopparam has quit IRC | 12:33 | |
| *** julienvey has joined #heat | 12:34 | |
| asalkeld | g'night all | 12:34 |
| *** rpothier has joined #heat | 12:34 | |
| *** asalkeld has quit IRC | 12:34 | |
| cmyster | nn asa | 12:37 |
| cmyster | oops | 12:37 |
| cmyster | shardy: issue resolved | 12:38 |
| cmyster | it wasn't a real issue | 12:38 |
| *** sgordon_ has joined #heat | 12:39 | |
| *** rpothier has quit IRC | 12:39 | |
| inc0 | ananta, by the way, have you seen my approach yet? | 12:42 |
| *** rpothier has joined #heat | 12:42 | |
| inc0 | I'd be curious if you have any feedback on your own | 12:42 |
| *** kleini has joined #heat | 12:44 | |
| *** chris_ has quit IRC | 12:44 | |
| kleini | how can I get outputs of a stack when the stack creation failed? It seems to me that no outputs all are then available. | 12:44 |
| *** rushiagr is now known as rushiagr_away | 12:46 | |
| pas-ha | kleini, I think it depends on whether the resources that those outputs refer to have been created | 12:46 |
| pas-ha | i.e if yuo had get_attr: [server, networks] and server failed to create - you get nothing | 12:46 |
| *** jcoufal_ has quit IRC | 12:47 | |
| pas-ha | since most of the attributes are actually making API calls to the resource | 12:47 |
| kleini | I have outputs from three different resources. two of them are created completely without any failures. their outputs are not available, too. | 12:48 |
| *** hdd has joined #heat | 12:48 | |
| *** jcoufal has joined #heat | 12:48 | |
| pas-ha | than it seems I am wrong :( | 12:48 |
| kleini | here is what I see with the heat client: http://pastebin.com/icUq89yA | 12:51 |
| kleini | the outputs for failed stacks looks strange... | 12:53 |
| pas-ha | can you paste your outputs section from the template? | 12:53 |
| kleini | http://pastebin.com/sme1Cp9s | 12:55 |
| *** nkhare has quit IRC | 12:57 | |
| *** rpothier has quit IRC | 12:59 | |
| ananta | inc0, nope | 13:02 |
| ananta | great if you could tell in brief | 13:02 |
| pas-ha | so, what actually failed? if server, than none of the other two ref'd resources created either | 13:02 |
| pas-ha | kleini, ^ | 13:02 |
| *** zaneb has joined #heat | 13:08 | |
| zaneb | hey guys | 13:08 |
| zaneb | did I miss the meeting already? | 13:08 |
| dulek | zaneb: Yes. | 13:09 |
| zaneb | dammit | 13:09 |
| dulek | zaneb: But there weren't anything special I think. | 13:09 |
| zaneb | sorry, haven't adjusted my calendar for daylight savings ending | 13:09 |
| zaneb | these 7am ones are not working our for me :/ | 13:10 |
| zaneb | dulek: cool, thanks. I will read the log | 13:10 |
| inc0 | hi zaneb | 13:10 |
| dulek | zaneb: inc0 will be here in a moment, maybe you can confront both convergence ideas | 13:10 |
| zaneb | but right now... breakfast | 13:10 |
| dulek | also ananta is there I think | 13:11 |
| kleini | pas-ha: only the last resource 'deployment' fails actually. the floating-ip and the server are created completely | 13:11 |
| zaneb | dulek, inc0, ananta: my plan is to start a mailing list discussion today | 13:12 |
| inc0 | thats probably good idea, but will make things longer.. | 13:13 |
| inc0 | I was thinking about other thing...I think we could proceed with both ideas zaneb - yours and mine | 13:14 |
| inc0 | I'm not sure how yours cope with ananta's but all I'm really doing is slightly different observeer | 13:14 |
| inc0 | which may one day be enough for creation and so on | 13:15 |
| ananta | zaneb: sure, we can start with the mailing list | 13:15 |
| ananta | and discuss in detail next week | 13:16 |
| inc0 | its thanksgiving week I think, that may make things a bit stressed about time | 13:16 |
| *** prazumovsky has joined #heat | 13:20 | |
| *** nosnos has quit IRC | 13:20 | |
| *** prazumovsky has left #heat | 13:20 | |
| *** prazumovsky has joined #heat | 13:20 | |
| *** alexpilotti has joined #heat | 13:21 | |
| *** mkollaro has joined #heat | 13:21 | |
| *** LiJiansheng has quit IRC | 13:22 | |
| openstackgerrit | Qiming Teng proposed openstack/heat: Extract group functions into a utility module. https://review.openstack.org/137356 | 13:24 |
| *** apporc has quit IRC | 13:24 | |
| *** swygue has joined #heat | 13:25 | |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat_integrationtests https://review.openstack.org/137126 | 13:30 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Remove ignoring [H302] in tox.ini https://review.openstack.org/137127 | 13:30 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/engine https://review.openstack.org/134187 | 13:30 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/tests https://review.openstack.org/135330 | 13:30 |
| *** alexpilotti has quit IRC | 13:32 | |
| *** serg_melikyan has joined #heat | 13:35 | |
| *** blues-man has joined #heat | 13:39 | |
| zaneb | inc0, ananta: back | 13:39 |
| zaneb | yeah, it is thanksgiving tomorrow, so I am out the rest of the week after today | 13:39 |
| *** serg_melikyan has quit IRC | 13:40 | |
| ananta | zaneb, inc0: sorry I have to go for a meeting, catch you later | 13:40 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat_integrationtests https://review.openstack.org/137126 | 13:40 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Remove ignoring [H302] in tox.ini https://review.openstack.org/137127 | 13:40 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/engine https://review.openstack.org/134187 | 13:40 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/tests https://review.openstack.org/135330 | 13:40 |
| *** ananta has quit IRC | 13:40 | |
| *** rushiagr_away is now known as rushiagr | 13:42 | |
| *** rakesh_hs2 has quit IRC | 13:43 | |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/engine https://review.openstack.org/134187 | 13:43 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat_integrationtests https://review.openstack.org/137126 | 13:45 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Remove ignoring [H302] in tox.ini https://review.openstack.org/137127 | 13:45 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/engine https://review.openstack.org/134187 | 13:45 |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Fix [H302] errors in heat/tests https://review.openstack.org/135330 | 13:45 |
| inc0 | I'm back as well | 13:50 |
| *** dimsum__ has joined #heat | 13:51 | |
| inc0 | zaneb, I was thinking if we could proceed with our ideas at same time | 13:51 |
| zaneb | interesting, how would that work do you think? | 13:52 |
| inc0 | I think if we'll keep in sync, we could model our work in a way which will solve problems of both approaches | 13:52 |
| inc0 | well we do want observer to happen eventuallyu | 13:53 |
| inc0 | and its pretty close to what I'm doing | 13:53 |
| inc0 | only difference will be its behaviour wont be just to schedule next graph traveral, but to schedule actual actions | 13:54 |
| *** che-arn8 has joined #heat | 13:54 | |
| *** che-arne has joined #heat | 13:54 | |
| *** che-arn8 has quit IRC | 13:54 | |
| inc0 | we both have notion of goal, just representation slightly differs | 13:55 |
| zaneb | so the idea behind the original convergence flow chart is that we do both | 13:55 |
| zaneb | we schedule graph traversals for the worker, and individual actions for the observer | 13:55 |
| inc0 | and difference would be that graph traversals won't really happen in worker | 13:56 |
| inc0 | it could happen in engine and worker will be just to do atomic actions | 13:56 |
| inc0 | like call nova.servers.create | 13:56 |
| inc0 | I mean handle_create | 13:56 |
| inc0 | so, client call stack create, your algorithm will schedule actions for creation based on graph and so on | 13:57 |
| zaneb | fwiw the plan is that the engines will be the workers | 13:57 |
| zaneb | but I think what you're suggesting is that each traversal will be owned by one engine? | 13:58 |
| inc0 | whatever we name it (personally I think workers would have to be more scallable than any other part) | 13:58 |
| inc0 | that wouldn't be much of an improvement;) | 13:58 |
| zaneb | no, exactly ;) | 13:59 |
| inc0 | what I mean is you keep your graph state in db and at some point you'll want to schedule next step right? | 13:59 |
| zaneb | actually in my approach we don't keep it in the DB | 14:00 |
| inc0 | distributed-graph? | 14:00 |
| zaneb | I guess we keep parts of it in the db, when we are joining two branches | 14:01 |
| inc0 | well, at least notion of dependency has to be kept somewhere | 14:01 |
| inc0 | it can be in resource table | 14:02 |
| inc0 | still, it will be some representation of an graph | 14:02 |
| *** hdd has quit IRC | 14:02 | |
| inc0 | and this representation will be goal my observer wants to achieve | 14:02 |
| zaneb | yes | 14:02 |
| inc0 | then there is slight difference of approaches, because you base on this graph and calculate next moves | 14:03 |
| inc0 | while I don't really calculate next moves with notion of graph as whole, just moves that brings us closer to what we want | 14:04 |
| inc0 | and then there is execution, and it will be 80+% of work in my opinion | 14:04 |
| inc0 | and it will be common for both approaches if we'll be careful to make it so | 14:05 |
| *** aweiteka has joined #heat | 14:06 | |
| *** sanjayu has quit IRC | 14:07 | |
| *** kitch_ has quit IRC | 14:15 | |
| *** kitch_ has joined #heat | 14:16 | |
| *** kitch_ has quit IRC | 14:17 | |
| *** julienvey has quit IRC | 14:18 | |
| *** kitch_ has joined #heat | 14:20 | |
| *** kitch_ has quit IRC | 14:20 | |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Change RouterGateway resource's name https://review.openstack.org/136339 | 14:21 |
| *** kitch_ has joined #heat | 14:23 | |
| openstackgerrit | Peter Razumovsky proposed openstack/heat: Change RouterGateway resource's name https://review.openstack.org/136339 | 14:23 |
| inc0 | zaneb, btw...if we consider complex actions like update-replace as atomic, I guess algorithm is almost the same.. | 14:23 |
| inc0 | I had problems because I didn't want to make this an atomic action... | 14:25 |
| inc0 | (but problem was solvable...nearly solved really) | 14:25 |
| zaneb | it can't be atomic, can it? what do you mean by atomic? | 14:25 |
| inc0 | update-replace as one synchronious function call | 14:26 |
| inc0 | a workflow so to say | 14:26 |
| *** Qiming has quit IRC | 14:26 | |
| zaneb | I don't see how it can be, given that other resources need to be processed between creating the replacement and deleting the original | 14:26 |
| inc0 | yeah, thats my point...and creation takes time | 14:27 |
| inc0 | quite frankly I'm trying to find some fundamental differences between our approaches | 14:31 |
| inc0 | could you help me with that? | 14:31 |
| *** ramishra has quit IRC | 14:33 | |
| zaneb | tbh I'm not entirely sure what your approach is, because there is so much stuff where you said 'oh, the real thing won't work like this' | 14:34 |
| inc0 | my approach is: "someone creates a stack, we create goal stack in db" | 14:34 |
| inc0 | then there is while True: | 14:34 |
| zaneb | I do know that it's so heavy on the db that even the simulator runs noticeably slow, which is something I wouldn't previously have believed possible ;) | 14:34 |
| *** julienvey has joined #heat | 14:34 | |
| inc0 | check goal stack -> check what is in reality -> put on queue action which we can do right now (all deps met) and will solve subset of diffs | 14:35 |
| inc0 | thats it | 14:36 |
| *** serg_melikyan has joined #heat | 14:36 | |
| inc0 | you do update, you change goal stack | 14:36 |
| inc0 | you do next update (even before previous was finished) you change goal stack | 14:36 |
| inc0 | and converger doesn't really care if you updated something or what | 14:37 |
| *** Qiming has joined #heat | 14:37 | |
| inc0 | it just looks, o, there is difference, all requirements are met, lets solve this one difference and check again | 14:37 |
| inc0 | or even schedule solving | 14:38 |
| zaneb | it sounds similar, except for the while True part | 14:38 |
| inc0 | yeah, you just run it | 14:38 |
| inc0 | but problem is | 14:38 |
| inc0 | you don't really know how many steps are needed | 14:38 |
| inc0 | it will end...eventually | 14:38 |
| inc0 | so you can't really just say "do converge" | 14:38 |
| inc0 | because end will happen in finite number of steps, but you never know this number | 14:39 |
| zaneb | I think the difference is you're trying to combine the worker and the observer into a single process, effectively | 14:39 |
| inc0 | its like...stack is complete when there is no futher diff to process | 14:39 |
| inc0 | no, why? observer schedules n tasks which are possible to do now | 14:40 |
| inc0 | worker takes 1 task from queue, performs it and thats all | 14:40 |
| zaneb | which I don't think will be helpful when we have two different kinds of resources (phase 2-compatible and legacy) | 14:40 |
| inc0 | what wouldn't be compatible? | 14:40 |
| *** serg_melikyan has quit IRC | 14:40 | |
| dulek | inc0: observer schedules tasks? Or engine? | 14:40 |
| zaneb | see, you're using your own definition of observer again | 14:40 |
| inc0 | dulek, observer...whatever does converge really | 14:41 |
| zaneb | which makes this discussion really hard | 14:41 |
| zaneb | because we are not speaking the same language | 14:41 |
| *** mikeit has joined #heat | 14:42 | |
| inc0 | ok, so lets call worker==process which takes task from queue and performs it (like single handle_create) | 14:42 |
| inc0 | converger==event loop which checks whats real state (from db) and goal state(from db) and puts actions on queue which can be done right now | 14:43 |
| inc0 | observer==process which polls real resources (literally checks if nova instance is active) and puts that status to real_state_db | 14:43 |
| inc0 | engine==process API calls for template resolution and puts goal_db_state | 14:44 |
| zaneb | no, let's back up | 14:45 |
| inc0 | ok.. | 14:45 |
| zaneb | I'm not even talking about your proposal | 14:45 |
| zaneb | I'm talking about the original convergence flowchart | 14:45 |
| *** che-arne has quit IRC | 14:46 | |
| zaneb | in that flowchart you had a worker, which is responsible for handling the update (or not) of one resource | 14:46 |
| zaneb | which it can do one of two ways: | 14:46 |
| zaneb | phase 1: by calling handle_create()/handle_update() directly | 14:47 |
| *** mikeit has quit IRC | 14:47 | |
| zaneb | phase 2: by farming work out to an observer | 14:47 |
| zaneb | (provided that the resource is written to support that, otherwise the phase 1 method remains an option) | 14:48 |
| *** mikeit has joined #heat | 14:48 | |
| zaneb | your proposal effectively combines the two parts, so each step is a step that in the original flowchart would have been performed by the observer | 14:49 |
| inc0 | with a few changes, because observer would also have notion of dependency | 14:49 |
| inc0 | but basically yes | 14:49 |
| inc0 | and problem is, handle_create now is completely stateful and synchronous right? | 14:50 |
| zaneb | yes | 14:51 |
| inc0 | allright, so for start lets do something like that: | 14:51 |
| zaneb | and for some resources will remain that way for quite some time | 14:51 |
| inc0 | worker (I'm reffering to worker I described) | 14:51 |
| inc0 | will call rsc.handle_pre_create() | 14:51 |
| inc0 | rsc.handle_creat() | 14:51 |
| inc0 | rsc.hadnle_post_create() | 14:52 |
| inc0 | pre and post create will be methods implemented in base resource class | 14:52 |
| inc0 | so will be inherited | 14:52 |
| inc0 | and pre: will add resource with "new" status to reality, so effectively create a resource lock | 14:53 |
| inc0 | post: will release a lock | 14:53 |
| inc0 | and handle_create will be just synchronous handle_create | 14:54 |
| inc0 | later we can override these methods in resources to slowly migrate them to phase2-compatible | 14:54 |
| inc0 | you can put synchronous code to async framework | 14:55 |
| *** che-arne has joined #heat | 14:56 | |
| zaneb | I don't want to have to implement the algorithm in every new resource; implement it once in create() and choose on the basis of what handle_*() methods are available | 14:57 |
| *** Qiming has quit IRC | 14:58 | |
| *** andersonvom has joined #heat | 14:59 | |
| inc0 | agreed | 15:00 |
| zaneb | inc0: btw did you look at the changes I pushed to iterative-adapted yesterday? | 15:00 |
| zaneb | I fixed the test failures (my fault) | 15:01 |
| zaneb | but then I imported the basic_create test from my branch | 15:01 |
| inc0 | ah | 15:01 |
| zaneb | and it fails because it doesn't respect dependencies | 15:01 |
| zaneb | I had a go at fixing it but couldn't find anything obvious | 15:02 |
| inc0 | yeah, thats probably because I didnt resolve getrsc or getatt as new deps | 15:02 |
| zaneb | yep, I think so | 15:02 |
| inc0 | I simply put all of this explicitly to template | 15:02 |
| zaneb | but I couldn't find the code where you were checking for any dependencies | 15:02 |
| zaneb | so I didn't know how to fix it | 15:02 |
| dulek | I'll check if this is the issue. | 15:02 |
| inc0 | https://github.com/inc0/heat-convergence-prototype/blob/iterative/converge/resource.py#L79 | 15:03 |
| zaneb | but I think you could consider merging the -adapted branch now, so that we are both using the same test framework | 15:03 |
| inc0 | I will | 15:03 |
| zaneb | aha | 15:04 |
| inc0 | I also will write a detailed spec of my idea, because I feel there is lot of misunderstands between us;) | 15:04 |
| zaneb | btw Template.dependencies() calculates it all for you | 15:04 |
| inc0 | or even misunderstandings | 15:04 |
| inc0 | yeah, I figured that out afterwards;) | 15:04 |
| inc0 | also I could do something about this db heavy stuff...lookups are done in horrible way | 15:05 |
| zaneb | yeah | 15:05 |
| dulek | If you put the dependencies explicitely basic_create is working fine on iterative-adapted | 15:05 |
| zaneb | so some of my design goals were | 15:06 |
| dulek | So it's like inc0 said. | 15:06 |
| zaneb | - minimise db access | 15:06 |
| zaneb | - constrain db locking to the smallest scope possible, for minimal contention | 15:06 |
| *** elynn has quit IRC | 15:06 | |
| zaneb | - avoid reading attributes of resources (i.e. hitting the OpenStack APIs) multiple time | 15:07 |
| zaneb | dulek: ok cool, thanks | 15:07 |
| inc0 | yeah, I probably could rewrite this now in better way | 15:07 |
| *** sgordon has joined #heat | 15:08 | |
| inc0 | basically all dependency stuff are stored in check_*_readiness | 15:09 |
| inc0 | where each resource will check if things its dependant on are in state which it wants it to be | 15:09 |
| inc0 | or needs it to be | 15:09 |
| *** jcoufal has quit IRC | 15:10 | |
| *** mkollaro1 has joined #heat | 15:11 | |
| *** mkollaro has quit IRC | 15:12 | |
| *** randallburt has joined #heat | 15:19 | |
| *** randallburt has quit IRC | 15:19 | |
| *** randallburt has joined #heat | 15:19 | |
| *** lazy_prince is now known as killer_prince | 15:21 | |
| *** edmund has joined #heat | 15:28 | |
| *** dimsum__ has quit IRC | 15:28 | |
| *** dimsum__ has joined #heat | 15:29 | |
| *** gokrokve has joined #heat | 15:30 | |
| jpeeler | shardy: of all(any) of your stable updates going to make tomorrow's freeze? | 15:31 |
| *** radez is now known as radez_g0n3 | 15:32 | |
| *** andersonvom has quit IRC | 15:32 | |
| *** andersonvom has joined #heat | 15:33 | |
| shardy | jpeeler: I'd love them to, I just have no idea how to make them pass the **** gate tests | 15:34 |
| jpeeler | was assuming they'd make it in, is there a packaging deadline for RDO? | 15:35 |
| shardy | jpeeler: We may have to carry them downstream, atm I don't understand why the grenade job keeps failing, let me look at it again | 15:38 |
| shardy | I've nearly worn out the letters r e c h e and k on my keyboard :\ | 15:39 |
| jpeeler | heh | 15:39 |
| *** alexpilotti has joined #heat | 15:39 | |
| jpeeler | ok, i'm prepared to do that. i'll at least assume at this point it's okay to roll into the next stable release, rather than right this moment. | 15:39 |
| shardy | jpeeler: Ok, sounds good, I'll have another attempt at rechecking them | 15:40 |
| shardy | Hmm, grenade actually failed for a different reason this time I think, maybe it will work this time | 15:44 |
| *** jcoufal has joined #heat | 15:46 | |
| jpeeler | it's been added to this list just in case: https://etherpad.openstack.org/p/StableJuno | 15:47 |
| *** GonZo2K has joined #heat | 15:49 | |
| *** JayJ has joined #heat | 15:52 | |
| *** EricGonczer_ has joined #heat | 15:56 | |
| *** GonZo2K has quit IRC | 15:57 | |
| *** sarob has joined #heat | 16:00 | |
| inc0 | cyas guys, happy thanksgiving for US folks | 16:00 |
| *** EricGonczer_ has quit IRC | 16:01 | |
| *** dimsum__ is now known as dims | 16:03 | |
| *** mikeit has quit IRC | 16:04 | |
| *** inc0 has quit IRC | 16:05 | |
| *** serg_melikyan has joined #heat | 16:06 | |
| *** vijendar has joined #heat | 16:09 | |
| *** ishant has quit IRC | 16:09 | |
| *** serg_melikyan has quit IRC | 16:10 | |
| *** JayJ has quit IRC | 16:12 | |
| *** JayJ has joined #heat | 16:13 | |
| *** zigo has quit IRC | 16:16 | |
| *** JayJ has quit IRC | 16:17 | |
| *** rpothier has joined #heat | 16:17 | |
| *** EricGonczer_ has joined #heat | 16:18 | |
| *** JayJ has joined #heat | 16:18 | |
| *** zaneb has quit IRC | 16:18 | |
| *** tomek_adamczewsk has quit IRC | 16:19 | |
| *** JayJ has quit IRC | 16:22 | |
| *** JayJ has joined #heat | 16:23 | |
| *** EricGonczer_ has quit IRC | 16:23 | |
| *** swygue has quit IRC | 16:24 | |
| *** crose has joined #heat | 16:25 | |
| *** swygue has joined #heat | 16:26 | |
| openstackgerrit | Tetiana Lashchova proposed openstack/python-heatclient: Fix H302 errors https://review.openstack.org/136067 | 16:26 |
| *** swygue has quit IRC | 16:26 | |
| *** swygue has joined #heat | 16:26 | |
| *** dulek has quit IRC | 16:26 | |
| *** zigo has joined #heat | 16:27 | |
| *** sarob has quit IRC | 16:28 | |
| *** gokrokve_ has joined #heat | 16:30 | |
| *** k4n0 has quit IRC | 16:30 | |
| *** gokrokve has quit IRC | 16:31 | |
| *** zaneb has joined #heat | 16:31 | |
| *** jcoufal has quit IRC | 16:33 | |
| *** JayJ has quit IRC | 16:34 | |
| jpeeler | zaneb: i know you're probably trying to get things wrapped up today, but if you could let me know if https://review.openstack.org/137237 is even close to what you were wanting, that would be helpful | 16:34 |
| zaneb | sure, I'll take a look | 16:34 |
| *** sarob has joined #heat | 16:35 | |
| *** packet has joined #heat | 16:35 | |
| *** packet has quit IRC | 16:37 | |
| *** ananta has joined #heat | 16:40 | |
| zaneb | jpeeler: looks really close | 16:42 |
| zaneb | jpeeler: can you explain "# PROBLEM: get_class doesn't have access to the stack object?"? | 16:42 |
| *** rpothier has quit IRC | 16:45 | |
| *** gokrokve_ has quit IRC | 16:45 | |
| *** Marga_ has joined #heat | 16:46 | |
| *** serg_melikyan has joined #heat | 16:46 | |
| jpeeler | zaneb: i didn't see a way to always return an outputs object, such as in generate_class of template_resource | 16:48 |
| jpeeler | which sort of defeats everything that i was trying to do | 16:48 |
| zaneb | ah, ok | 16:48 |
| zaneb | I have an idea for that | 16:48 |
| zaneb | just writing a comment now | 16:48 |
| *** ananta has quit IRC | 16:49 | |
| jpeeler | super, I figured you had all the answers | 16:50 |
| *** fayablazer has quit IRC | 16:51 | |
| *** serg_melikyan has quit IRC | 16:51 | |
| *** kleini has left #heat | 16:53 | |
| *** viktors is now known as viktors|afk | 16:53 | |
| zaneb | jpeeler: rofl | 16:56 |
| *** andersonvom has quit IRC | 16:57 | |
| *** Marga_ has quit IRC | 16:57 | |
| *** Marga_ has joined #heat | 16:58 | |
| *** Marga_ has quit IRC | 16:59 | |
| *** Marga_ has joined #heat | 16:59 | |
| *** Marga_ has quit IRC | 16:59 | |
| *** Marga_ has joined #heat | 17:00 | |
| *** prazumovsky has quit IRC | 17:02 | |
| *** andersonvom has joined #heat | 17:13 | |
| zaneb | jpeeler: ah, it's trickier than I thought. I was pretty sure you can call parse() with stack=None though? obviously all bets are off for validating/getting outputs, but if you're only using it to grab the schema it should be fine... | 17:16 |
| openstackgerrit | Merged openstack/heat: Fix [H302] errors in heat/engine https://review.openstack.org/134187 | 17:17 |
| openstackgerrit | Merged openstack/heat: Fix [H302] errors in heat/tests https://review.openstack.org/135330 | 17:18 |
| openstackgerrit | Merged openstack/heat: Fix [H302] errors in heat_integrationtests https://review.openstack.org/137126 | 17:20 |
| openstackgerrit | Merged openstack/heat: Remove ignoring [H302] in tox.ini https://review.openstack.org/137127 | 17:21 |
| openstackgerrit | Merged openstack/heat: Update Barbican resources to match library changes https://review.openstack.org/137315 | 17:21 |
| shardy | jpeeler: FYI I raised bug #1396696 for the most persistent juno grenade check failure, as it's not clear to me if it's an existing bug or not | 17:23 |
| uvirtbot | Launchpad bug 1396696 in glance "grenade tests failing with glance 400 error" [Undecided,New] https://launchpad.net/bugs/1396696 | 17:23 |
| shardy | I'll keep rechecking, but if anyone else wants to recheck while I'm AFK this evening, feel free | 17:23 |
| shardy | jpeeler: I linked the affected patches in that bug report | 17:24 |
| jpeeler | shardy: thanks - Alan is at least aware of them | 17:24 |
| *** Marga_ has quit IRC | 17:24 | |
| jpeeler | zaneb: hrm, I don't think I tried that - will look into that | 17:25 |
| *** Marga_ has joined #heat | 17:26 | |
| *** sdw_ has joined #heat | 17:26 | |
| *** Marga_ has quit IRC | 17:30 | |
| *** Marga_ has joined #heat | 17:30 | |
| openstackgerrit | Anderson Mesquita proposed openstack/heat: Add Digest intrinsic function https://review.openstack.org/137188 | 17:32 |
| *** gokrokve has joined #heat | 17:39 | |
| *** blues-man has quit IRC | 17:40 | |
| *** mkollaro1 has quit IRC | 17:44 | |
| *** sorantis has quit IRC | 17:51 | |
| *** Marga_ has quit IRC | 17:52 | |
| *** sdw_ has quit IRC | 17:52 | |
| *** Marga_ has joined #heat | 17:53 | |
| *** Marga_ has quit IRC | 17:54 | |
| *** Marga_ has joined #heat | 17:55 | |
| *** Marga_ has quit IRC | 17:55 | |
| *** jistr has quit IRC | 17:56 | |
| sdake | anyone ever seen this ERROR: create() argument after ** must be a mapping, not Namespace | 17:56 |
| *** Marga_ has joined #heat | 17:56 | |
| sdake | (unrleated to heat) | 17:56 |
| *** derekh has quit IRC | 17:58 | |
| *** julienvey has quit IRC | 17:58 | |
| *** steel13 has joined #heat | 17:59 | |
| *** harlowja_away is now known as harlowja | 18:04 | |
| *** Marga_ has quit IRC | 18:06 | |
| *** Marga_ has joined #heat | 18:06 | |
| *** Marga_ has quit IRC | 18:07 | |
| steel13 | when creating heat stack, through the command line, an error returned, "ERROR: Could not fetch remote template 'software_config_server.yaml': Invalid URL scheme", in the template i'm using i have a property "type" referencing another template and here is the problem, anyone have any ideas to resolve this? | 18:07 |
| *** Marga_ has joined #heat | 18:07 | |
| *** Marga_ has quit IRC | 18:12 | |
| *** Marga_ has joined #heat | 18:12 | |
| *** Marga_ has quit IRC | 18:13 | |
| *** crose has quit IRC | 18:14 | |
| *** Marga_ has joined #heat | 18:14 | |
| steel13 | exist some specific structure to use in property type resource? | 18:15 |
| larsks | steel13: can you post your template? | 18:17 |
| steel13 | http://paste.openstack.org/show/101278/ | 18:17 |
| larsks | steel13: I'm doing pretty much exactly that here, and it works: https://github.com/larsks/heat-kubernetes/blob/master/kubecluster.yaml#L214 | 18:17 |
| *** Marga_ has quit IRC | 18:18 | |
| *** Marga_ has joined #heat | 18:18 | |
| *** harlowja has quit IRC | 18:18 | |
| *** andersonvom has quit IRC | 18:18 | |
| larsks | And there exists a file named "software_config_server.yaml" in the same directory as that template? | 18:18 |
| *** andersonvom_ has joined #heat | 18:18 | |
| *** Marga_ has quit IRC | 18:19 | |
| *** harlowja has joined #heat | 18:19 | |
| *** Marga_ has joined #heat | 18:19 | |
| steel13 | larsks: yes, i have the two files in the same directory | 18:20 |
| larsks | What version of the heat client are you using? | 18:21 |
| *** Marga_ has quit IRC | 18:21 | |
| *** Marga_ has joined #heat | 18:22 | |
| steel13 | larsks: heat --version 0.2.8 | 18:24 |
| *** Marga_ has quit IRC | 18:25 | |
| larsks | Hmmm. I've got 0.2.12 locally; I don't know if that is significant or not. | 18:25 |
| *** Marga_ has joined #heat | 18:25 | |
| larsks | For what it's worth, the template itself looks fine to me. | 18:25 |
| steel13 | hmm, i will update version to test. Thanks | 18:26 |
| *** ifarkas has quit IRC | 18:27 | |
| *** Marga_ has quit IRC | 18:32 | |
| *** Marga_ has joined #heat | 18:32 | |
| *** Marga_ has quit IRC | 18:35 | |
| *** Marga_ has joined #heat | 18:36 | |
| *** Marga_ has quit IRC | 18:39 | |
| *** Marga_ has joined #heat | 18:39 | |
| *** harlowja_ has joined #heat | 18:41 | |
| *** jstrachan has quit IRC | 18:44 | |
| *** harlowja has quit IRC | 18:45 | |
| *** rushiagr is now known as rushiagr_away | 18:45 | |
| openstackgerrit | Tetiana Lashchova proposed openstack/heat: Remove resource_by_refid() from OS::Heat::HARestarter https://review.openstack.org/137446 | 19:00 |
| *** reed has quit IRC | 19:02 | |
| *** steel13 has quit IRC | 19:18 | |
| *** mkollaro has joined #heat | 19:18 | |
| *** jprovazn has quit IRC | 19:20 | |
| *** reed has joined #heat | 19:23 | |
| *** bitblt has joined #heat | 19:28 | |
| *** bitblt has left #heat | 19:30 | |
| *** tomek_adamczewsk has joined #heat | 19:37 | |
| *** rpothier has joined #heat | 19:43 | |
| *** pas-ha has quit IRC | 19:47 | |
| *** GonZo2K has joined #heat | 19:56 | |
| *** dims has quit IRC | 19:56 | |
| *** radez_g0n3 is now known as radez | 20:01 | |
| *** rpothier has quit IRC | 20:15 | |
| *** sarob has quit IRC | 20:28 | |
| *** sarob has joined #heat | 20:29 | |
| *** sarob has quit IRC | 20:34 | |
| *** ananta has joined #heat | 20:37 | |
| ananta | Hi zaneb | 20:40 |
| zaneb | o/ | 20:40 |
| ananta | Good afternoon | 20:40 |
| zaneb | ananta: isn't it like 2am in India??? | 20:41 |
| ananta | zaneb: yes | 20:41 |
| ananta | zaneb: I thought to discuss about and heard that there will be thanksgiving and folks may not be available | 20:42 |
| ananta | *about poc | 20:42 |
| zaneb | yeah, US-based folks will be away the next couple of days | 20:42 |
| ananta | zaneb: I couldn't get much time to see the poc you have been working on, but I have tried to read the log messages | 20:43 |
| ananta | we have the convergence poc at github https://github.com/anantpatil/heat-convergence-poc and would like to have some feedback for us | 20:45 |
| zaneb | yes, I am going to try to write a summary for the list today | 20:45 |
| zaneb | although I have struggled to be able to review what you have | 20:46 |
| zaneb | I think where I ended up was similar, except for the method of traversing the graph | 20:46 |
| *** tomek_adamczewsk has quit IRC | 20:46 | |
| *** tomek_adamczewsk has joined #heat | 20:46 | |
| zaneb | I have a DB record for each node that is waiting on input in a given transition, rather than one per node | 20:47 |
| zaneb | I think you will run into locking problems with your approach, but I haven't looked closely to see how you addressed that yet | 20:47 |
| ananta | zaneb: ok...how would it behave if an engine were to go down? | 20:48 |
| zaneb | that is the big undecided question | 20:49 |
| ananta | zaneb: we are seeing that we will have to lock frequently, but the lock is for minimum duration | 20:49 |
| zaneb | if we know something has died, we can start another traversal | 20:49 |
| zaneb | ananta: yeah, I'm more concerned that everything is in contention on the same lock | 20:50 |
| ananta | zaneb: yes, we hit the same wall...and we have few idea around removing the lock yet making it resilient | 20:51 |
| zaneb | I think your approach and mine have similar issues if an engine goes down, so I would be interested to hear your solution | 20:51 |
| *** aweiteka has quit IRC | 20:52 | |
| ananta | zaneb: in our approach, we keep the graph in db and send the nodes for convergence and exactly one engine listens for motifications on the stack topic | 20:53 |
| ananta | the engine which got the stack operation request creates a topic in AMQP and listens for the notifications there | 20:53 |
| ananta | when the notifs arrives, the engine queues them and processes them from there | 20:53 |
| zaneb | notifications to mark a node as traversed? | 20:54 |
| ananta | since there is only one engine listening, other engines will not contend for the lock | 20:54 |
| ananta | zaneb: yes...notifications mark as traversed, and new set of ready nodes a re computed | 20:54 |
| ananta | within one engine the notifications are queued so that they are processed sequentially | 20:55 |
| zaneb | heh, that sounds very similar to what inc0 is doing | 20:55 |
| ananta | the egine queue is simple queue datastructure | 20:55 |
| zaneb | and if that engine dies? | 20:55 |
| ananta | there is tack lock table in DB and if the engine dies another engine can pick up from there | 20:56 |
| zaneb | what about the lost notifications? | 20:56 |
| ananta | for that we need a periodic task running in each heat engine or an external api request | 20:56 |
| ananta | the notifications will be there in AMQP since they will not be ACKed untill they are processed and updated in DB | 20:57 |
| ananta | the idea is use the AMQP infrastructure effectivly and improve on relaibility | 20:57 |
| zaneb | hmmm.... ok. oslo.messaging does not support that. did you modify it somehow? | 20:57 |
| ananta | we are still in the process of doing it :) | 20:58 |
| ananta | doing this approach in poc... | 20:58 |
| zaneb | good, I think we will need that for whatever we implement :) | 20:58 |
| ananta | sure... | 20:59 |
| ananta | also we were thinking about concurrent updates and would like you and other folks to also think about it | 20:59 |
| ananta | in our case, if a stack is being provisioned, the concurrent request goes to the same engine which is processing the notification...so it just drops a;; the old notifs and starts fromm sratch again | 21:00 |
| ananta | *all | 21:01 |
| ananta | we are still brainstorming on concurrent update | 21:01 |
| zaneb | yeah, mine would have the same behaviour I think | 21:01 |
| ananta | sure | 21:01 |
| zaneb | I think that is probably the right thing | 21:02 |
| ananta | i think next week when you are back we can collaborate furthur and converge on a design | 21:02 |
| zaneb | you can't know what the rest of the template contains, so it doesn't make sense to continue the previous traversal when you have a new template | 21:02 |
| zaneb | +1 | 21:02 |
| zaneb | the tricky part is what to do with the resources that are already in progress | 21:03 |
| ananta | zaneb: sure, I have a feeling your approach doesn't drastically differ from what we are thinking :) | 21:03 |
| zaneb | yeah, I have a feeling you're right | 21:03 |
| ananta | zaneb: we wait for them to finish so that we move from one stable state to another for the resource | 21:03 |
| zaneb | yup, so the question is how long do you wait? | 21:04 |
| ananta | can't afford to just kill or abruptly stop on that resource | 21:04 |
| zaneb | in current Heat it's hard-coded at 4 minutes | 21:04 |
| *** andersonvom_ has quit IRC | 21:04 | |
| zaneb | that isn't a very good solution, but I don't really know what is | 21:04 |
| ananta | zaneb: I haven't thought about that...it will wait for default I guess... so we even wanted to bring the resource timeout | 21:05 |
| ananta | zaneb: but it was becoming a overkill... | 21:05 |
| ananta | nonetheless I feel someday we would need the resource timeout | 21:06 |
| zaneb | yeah, the discussion got sidetracked into nonsense about zookeeper :/ | 21:07 |
| ananta | zaneb: yeah... the only problem in current approach is that if the engine dies then another engine has no otherway to know that apart from an external request or periodic task to check engine status | 21:08 |
| ananta | so folks in my team were thinking we have some supporting infrastructure which can tell about the events in group membership | 21:09 |
| ananta | group of heat egines...if one goes down otherw would know via tooz or zookeeper etc. | 21:09 |
| ananta | so that they can take over from there | 21:09 |
| ananta | I do have a feeling for long term benifits of Heat engine we would need such external infra | 21:10 |
| ananta | otherwise we endup writing lot of code aroound this in HEat :( | 21:11 |
| ananta | zaneb: beileve me...I wish we were sitting face to face ...there is so much complexitity and feedaback needed and discussion needed | 21:12 |
| ananta | :) | 21:12 |
| *** randallburt has quit IRC | 21:12 | |
| zaneb | yeah, it's a struggle trying to figure out stuff like this collaboratively with no face-to-face meetings | 21:13 |
| *** achanda has joined #heat | 21:15 | |
| *** radez is now known as radez_g0n3 | 21:17 | |
| ananta | zaneb: please do draft an e-mail as you had planned...next week when you are back we can discuss more | 21:17 |
| zaneb | ok cool, will do | 21:20 |
| *** achanda has quit IRC | 21:20 | |
| ananta | zaneb: sure...thanks for you time and Happy Thanksgiving! | 21:21 |
| zaneb | no worries, thanks for getting up at 2am to discuss! | 21:21 |
| ananta | zaneb: no problem at all...feeling very sleepy...catch you later...bye | 21:23 |
| *** ananta has quit IRC | 21:23 | |
| *** mkollaro has quit IRC | 21:26 | |
| *** zaneb has quit IRC | 21:31 | |
| *** JayJ has joined #heat | 21:36 | |
| *** rpothier has joined #heat | 21:37 | |
| *** tomek_adamczewsk has quit IRC | 21:38 | |
| *** tomek_adamczewsk has joined #heat | 21:38 | |
| *** zaneb has joined #heat | 21:45 | |
| *** achanda has joined #heat | 21:48 | |
| *** achanda has quit IRC | 21:49 | |
| *** matt__ is now known as mattoliverau | 21:53 | |
| *** Drago has joined #heat | 21:55 | |
| *** ifarkas_ has joined #heat | 21:56 | |
| *** hdd has joined #heat | 21:56 | |
| *** ifarkas_ has quit IRC | 21:59 | |
| *** asalkeld has joined #heat | 22:03 | |
| *** tomek_adamczewsk has quit IRC | 22:06 | |
| *** tomek_adamczewsk has joined #heat | 22:06 | |
| *** alexpilotti has quit IRC | 22:15 | |
| *** Drago has quit IRC | 22:20 | |
| *** ccrouch has quit IRC | 22:21 | |
| *** hdd has quit IRC | 22:21 | |
| *** hdd has joined #heat | 22:24 | |
| *** kitch_ has quit IRC | 22:24 | |
| *** tomek_adamczewsk has quit IRC | 22:25 | |
| *** tomek_adamczewsk has joined #heat | 22:25 | |
| *** rpothier has quit IRC | 22:26 | |
| *** zaneb has quit IRC | 22:29 | |
| *** tellesnobrega_ has joined #heat | 22:40 | |
| *** zaneb has joined #heat | 22:41 | |
| *** alexpilotti has joined #heat | 22:45 | |
| *** EmilienM has quit IRC | 22:47 | |
| *** EmilienM has joined #heat | 22:47 | |
| *** Drago has joined #heat | 22:49 | |
| *** achanda has joined #heat | 22:50 | |
| *** jamielennox|away is now known as jamielennox | 22:51 | |
| *** Drago has quit IRC | 22:54 | |
| *** achanda has quit IRC | 22:55 | |
| *** tellesnobrega_ has quit IRC | 23:18 | |
| *** JayJ has quit IRC | 23:21 | |
| *** gokrokve has quit IRC | 23:27 | |
| *** hdd has quit IRC | 23:40 | |
| *** tochi has joined #heat | 23:44 | |
| *** ygoto has joined #heat | 23:45 | |
| *** tellesnobrega_ has joined #heat | 23:47 | |
| *** GonZoPT has joined #heat | 23:47 | |
| *** GonZo2K has quit IRC | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
