| *** eglynn__ has joined #openstack-monasca | 00:01 | |
| *** eglynn_ has quit IRC | 00:13 | |
| *** craigbr has joined #openstack-monasca | 01:28 | |
| *** craigbr has quit IRC | 01:43 | |
| *** craigbr has joined #openstack-monasca | 02:37 | |
| *** craigbr has quit IRC | 02:57 | |
| *** craigbr has joined #openstack-monasca | 03:47 | |
| *** craigbr has quit IRC | 04:27 | |
| *** rbak has quit IRC | 13:29 | |
| *** rbak has joined #openstack-monasca | 14:08 | |
| openstackgerrit | litong01 proposed a change to stackforge/monasca-api: initial python implementation https://review.openstack.org/120788 | 14:38 |
|---|---|---|
| *** sandywalsh has quit IRC | 15:01 | |
| *** cindyo has joined #openstack-monasca | 15:20 | |
| *** sandywalsh has joined #openstack-monasca | 15:32 | |
| *** craigbr has joined #openstack-monasca | 15:34 | |
| *** craigbr has quit IRC | 15:48 | |
| openstackgerrit | Deklan Dieterly proposed a change to stackforge/monasca-api: Split methods for better modularity https://review.openstack.org/121611 | 15:54 |
| *** craigbr has joined #openstack-monasca | 15:56 | |
| *** dschroeder has joined #openstack-monasca | 15:58 | |
| *** tongli has joined #openstack-monasca | 16:04 | |
| tongli | @craigbr, craig, you added some comment to my patch. | 16:05 |
| craigbr | Yes | 16:05 |
| tongli | however, I can not see the comments. | 16:05 |
| tongli | which patchset did you comment on? | 16:05 |
| craigbr | The first one because I was commenting on a comment | 16:06 |
| tongli | is it possible that you add it to the latest patch set which is #4. | 16:06 |
| tongli | which file do you remember? | 16:06 |
| craigbr | Let me check and I will fix it if possible | 16:06 |
| tongli | found it. | 16:07 |
| tongli | so the message format should be this? | 16:08 |
| tongli | { | 16:08 |
| tongli | "metric": {The metric as a JSON object}, | 16:08 |
| tongli | "meta": { | 16:08 |
| tongli | "tenantId": "the tenant ID acquired via authentication with Keystone", | 16:08 |
| tongli | "region": "the region that the metric was submitted under", | 16:08 |
| tongli | }, | 16:08 |
| tongli | "creation_time": the time when the API received the metric as UNIX time | 16:08 |
| tongli | } | 16:08 |
| tongli | not seeing the value field? where the value come in then? | 16:09 |
| craigbr | Yes | 16:09 |
| craigbr | Value is in the metric | 16:09 |
| tongli | oh, i c. | 16:09 |
| tongli | so this will be different from the spec? | 16:09 |
| craigbr | There are two specs: external (consumed by the API) and internal (consumed by other monasca components) | 16:10 |
| tongli | hmm. ok, i c. | 16:10 |
| tongli | who uses internal format to send messages? | 16:11 |
| tongli | I mean to kafka? | 16:11 |
| craigbr | The API sends Metrics using the internal format, the Persister and the Threshold Engine consume that format. | 16:11 |
| tongli | seems to me that api should validate a message based on api spec (external one), | 16:11 |
| craigbr | I agree. But, it needs to augment the incoming data to match the internal spec | 16:12 |
| craigbr | before putting it on the kafka queue | 16:12 |
| tongli | hmm, ok. | 16:13 |
| tongli | the api server take external format, then do transformation, make it internal format. | 16:13 |
| tongli | is that what the java server does? | 16:14 |
| craigbr | Yes. | 16:14 |
| tongli | ok. | 16:14 |
| tongli | it can be done easily in a middleware. | 16:14 |
| tongli | if you look at the patchset 4, specially the metric_validator.py. | 16:15 |
| craigbr | As long as it is done before it gets puts on the kafka queue. Because only the API server knows which tenant the data comes from. | 16:15 |
| tongli | it does validation, then you can trasform it | 16:15 |
| craigbr | The API server gets the tenant from the login information | 16:16 |
| tongli | sure. | 16:16 |
| tongli | that is what keystone middleware provides actually. | 16:16 |
| craigbr | Of course | 16:16 |
| tongli | consider you can have all these logic in one middleware or a series of middleware. | 16:17 |
| tongli | but the API server will simply place the message it receives onto the kafka queue. | 16:18 |
| craigbr | Yes. I wasn't commenting on the actual implementation, but answering your question about formats | 16:18 |
| tongli | if the request pass all the middleware in the pipeline. | 16:18 |
| craigbr | I didn't even read the code, I was just commenting on your comment | 16:18 |
| tongli | ok. | 16:18 |
| tongli | i c. | 16:18 |
| tongli | thanks for clarifying this, I did not know there are two different formats. | 16:19 |
| craigbr | You're welcome | 16:19 |
| tongli | so the metric is the original external format message. | 16:19 |
| tongli | then it wraps it into this internal format. | 16:19 |
| tongli | right? | 16:19 |
| craigbr | Yes | 16:20 |
| tongli | k. i c. | 16:20 |
| tongli | let me make a bit of changes. so that this logic will all be in the middleware specifc for HP deployment. | 16:20 |
| craigbr | It really isn't HP specific because that format is used by other components of monasca. Monasca is NOT just the API server | 16:21 |
| craigbr | Without the correct message format, the measurements won't get written the database and alarms won't get triggered | 16:22 |
| tongli | most likely IBM won't have this kind of wrapping. | 16:22 |
| tongli | we may pick completely different logic. | 16:23 |
| craigbr | IBM will write its own persister and threshold engine? | 16:23 |
| tongli | totally possible. | 16:23 |
| craigbr | Seems pointless | 16:23 |
| tongli | it is too specific. | 16:24 |
| tongli | many things won't work | 16:24 |
| tongli | thinking about events. | 16:24 |
| tongli | thinking about cadf. | 16:24 |
| craigbr | IBM is not interested in multi tenancy? | 16:24 |
| tongli | or any other messages. | 16:24 |
| tongli | has nothing to do with multi tenancy. | 16:24 |
| tongli | the original message may very well contain the project_id from OpenStack already | 16:25 |
| tongli | this is only required since HP did not send the project_id (tenant id) in the original msg | 16:25 |
| tongli | imaging you may already have messages contained the information. | 16:26 |
| tongli | it will be quite stupid to wrap it again. | 16:26 |
| tongli | there could be many many different kind of things coming in. | 16:26 |
| tongli | not only metrics, | 16:26 |
| craigbr | We effectively do, because the user has authenticated. You are asking the user to doubly specify the data and keep it in sync | 16:26 |
| tongli | this won't work if authentication is not using keystone. | 16:27 |
| craigbr | keystone is used for openstack. Monasca is used for openstack. | 16:28 |
| tongli | right, if I want to use it monitor other cloud? | 16:28 |
| tongli | not even all OS cloud uses keystone, you know that right? | 16:29 |
| *** jodah has joined #openstack-monasca | 16:29 | |
| *** garyh has joined #openstack-monasca | 16:29 | |
| craigbr | I am in a meeting at the moment. I will have to get back to you | 16:32 |
| *** dshurtle has joined #openstack-monasca | 16:33 | |
| *** dshurtle has quit IRC | 16:33 | |
| *** dshurtle has joined #openstack-monasca | 16:35 | |
| *** dshurtle has quit IRC | 16:37 | |
| openstackgerrit | litong01 proposed a change to stackforge/monasca-api: initial python implementation https://review.openstack.org/120788 | 17:18 |
| openstackgerrit | Gary Hessler proposed a change to stackforge/monasca-agent: Adding RabbitMQ detection and setup https://review.openstack.org/121630 | 17:18 |
| *** openstackgerrit has quit IRC | 18:32 | |
| *** openstackgerrit has joined #openstack-monasca | 18:32 | |
| *** jodah has quit IRC | 18:33 | |
| *** cindyo has quit IRC | 18:39 | |
| openstackgerrit | Jonathan Halterman proposed a change to stackforge/monasca-common: WIP - Event changes for alarmed-metrics support https://review.openstack.org/121652 | 18:53 |
| *** dshurtle has joined #openstack-monasca | 19:03 | |
| openstackgerrit | Deklan Dieterly proposed a change to stackforge/monasca-api: Split methods for better modularity https://review.openstack.org/121611 | 19:22 |
| sandywalsh | *mental note ... don't ask about stackforge any more :) | 19:25 |
| *** craigbr has quit IRC | 19:26 | |
| *** craigbr has joined #openstack-monasca | 19:36 | |
| *** cindyo has joined #openstack-monasca | 20:06 | |
| openstackgerrit | A change was merged to stackforge/monasca-api: Split methods for better modularity https://review.openstack.org/121611 | 20:21 |
| *** tongli has quit IRC | 20:26 | |
| cindyo | tongli: just so you know our other packages have unique names. This is important so that installs don’t overwirte each other. We have separate modules so they each have their own package | 20:43 |
| *** dshurtle has quit IRC | 21:46 | |
| *** rbak has quit IRC | 22:46 | |
| openstackgerrit | Jonathan Halterman proposed a change to stackforge/monasca-common: WIP - Event changes for alarmed-metrics support https://review.openstack.org/121652 | 22:58 |
| *** dschroeder has quit IRC | 23:04 | |
| openstackgerrit | Craig Bryant proposed a change to stackforge/monasca-persister: Remove compiler warning https://review.openstack.org/121705 | 23:12 |
| openstackgerrit | Jonathan Halterman proposed a change to stackforge/monasca-common: WIP - Event changes for alarmed-metrics support https://review.openstack.org/121652 | 23:15 |
| *** craigbr has quit IRC | 23:20 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!