| *** thomasem has joined #openstack-metering | 00:17 | |
| *** thomasem has quit IRC | 00:26 | |
| *** matsuhashi has joined #openstack-metering | 00:29 | |
| *** fnaval_ has quit IRC | 00:42 | |
| *** matsuhashi has quit IRC | 00:48 | |
| *** matsuhashi has joined #openstack-metering | 00:49 | |
| *** matsuhashi has quit IRC | 00:53 | |
| *** matsuhashi has joined #openstack-metering | 01:08 | |
| *** nosnos has joined #openstack-metering | 01:09 | |
| *** shakayumi has joined #openstack-metering | 01:20 | |
| *** matsuhashi has quit IRC | 01:34 | |
| *** matsuhashi has joined #openstack-metering | 01:35 | |
| *** fandikurnia01 has quit IRC | 01:38 | |
| *** fandikurnia01 has joined #openstack-metering | 01:39 | |
| *** matsuhashi has quit IRC | 01:40 | |
| *** matsuhashi has joined #openstack-metering | 01:43 | |
| *** bingbu has joined #openstack-metering | 01:55 | |
| *** nati_ueno has joined #openstack-metering | 01:59 | |
| *** shakayumi has quit IRC | 02:06 | |
| *** changbl has joined #openstack-metering | 02:09 | |
| *** dperaza has left #openstack-metering | 02:17 | |
| *** jsergent has quit IRC | 02:18 | |
| *** changbl has quit IRC | 02:19 | |
| *** mirrorbox has quit IRC | 02:25 | |
| *** dperaza has joined #openstack-metering | 02:30 | |
| *** lrsmith has joined #openstack-metering | 02:48 | |
| *** lrsmith has quit IRC | 02:55 | |
| *** d34dh0r53 has joined #openstack-metering | 02:59 | |
| *** herndon_ has quit IRC | 03:02 | |
| *** changbl has joined #openstack-metering | 03:02 | |
| *** fandikurnia01 has quit IRC | 03:07 | |
| *** fandikurnia01 has joined #openstack-metering | 03:08 | |
| *** nati_ueno has quit IRC | 03:14 | |
| *** nati_ueno has joined #openstack-metering | 03:16 | |
| *** sileht has quit IRC | 03:17 | |
| *** suo has joined #openstack-metering | 03:19 | |
| *** nati_ueno has quit IRC | 03:20 | |
| *** fandikurnia01 has quit IRC | 03:26 | |
| *** fandikurnia01 has joined #openstack-metering | 03:26 | |
| *** matsuhashi has quit IRC | 03:31 | |
| *** matsuhashi has joined #openstack-metering | 03:31 | |
| *** matsuhashi has quit IRC | 03:36 | |
| *** matsuhashi has joined #openstack-metering | 03:41 | |
| *** SergeyLukjanov has joined #openstack-metering | 04:01 | |
| *** boris-42 has joined #openstack-metering | 04:03 | |
| *** suo has quit IRC | 04:05 | |
| *** suo has joined #openstack-metering | 04:13 | |
| *** fnaval has joined #openstack-metering | 04:14 | |
| *** fandikurnia01 has quit IRC | 04:14 | |
| *** fandikurnia01 has joined #openstack-metering | 04:14 | |
| *** mgagne has joined #openstack-metering | 04:22 | |
| *** mgagne has joined #openstack-metering | 04:22 | |
| *** fandikurnia01 has quit IRC | 04:24 | |
| *** mirrorbox has joined #openstack-metering | 04:24 | |
| *** fandikurnia01 has joined #openstack-metering | 04:24 | |
| *** mgagne1 has joined #openstack-metering | 04:28 | |
| *** mgagne1 has joined #openstack-metering | 04:28 | |
| *** mgagne has quit IRC | 04:31 | |
| Anju | EmilienM: can I use this with swift | 04:36 |
|---|---|---|
| *** SergeyLukjanov has quit IRC | 04:37 | |
| *** fnaval_ has joined #openstack-metering | 04:39 | |
| *** fnaval has quit IRC | 04:39 | |
| *** sandywalsh has quit IRC | 04:49 | |
| *** boris-42 has quit IRC | 04:53 | |
| *** boris-42 has joined #openstack-metering | 04:55 | |
| *** boris-42 has joined #openstack-metering | 04:56 | |
| *** nati_ueno has joined #openstack-metering | 05:02 | |
| *** sdake_ has joined #openstack-metering | 05:11 | |
| *** nprivalova has joined #openstack-metering | 05:30 | |
| *** nprivalova has quit IRC | 05:41 | |
| *** mgagne1 is now known as mgagne | 05:44 | |
| *** SergeyLukjanov has joined #openstack-metering | 05:50 | |
| *** sdake_ has quit IRC | 05:59 | |
| *** sdake_ has joined #openstack-metering | 05:59 | |
| *** sdake_ has joined #openstack-metering | 05:59 | |
| *** d34dh0r53 has quit IRC | 05:59 | |
| *** nati_ueno has quit IRC | 06:03 | |
| *** nati_ueno has joined #openstack-metering | 06:04 | |
| openstackgerrit | Jenkins proposed a change to openstack/ceilometer: Imported Translations from Transifex https://review.openstack.org/54482 | 06:06 |
| *** matsuhashi has quit IRC | 06:08 | |
| *** matsuhashi has joined #openstack-metering | 06:08 | |
| *** matsuhashi has quit IRC | 06:13 | |
| *** nadya_ has joined #openstack-metering | 06:20 | |
| *** suo has quit IRC | 06:22 | |
| *** nosnos_ has joined #openstack-metering | 06:25 | |
| *** nosnos has quit IRC | 06:29 | |
| *** matsuhashi has joined #openstack-metering | 06:29 | |
| *** suo has joined #openstack-metering | 06:39 | |
| *** fnaval has joined #openstack-metering | 06:43 | |
| *** sdake_ has quit IRC | 06:44 | |
| *** fnaval_ has quit IRC | 06:47 | |
| *** ildikov has joined #openstack-metering | 06:51 | |
| *** SergeyLukjanov has quit IRC | 06:52 | |
| *** ildikov has quit IRC | 06:52 | |
| *** ildikov has joined #openstack-metering | 06:52 | |
| *** nadya_ has quit IRC | 06:58 | |
| *** sdake_ has joined #openstack-metering | 07:00 | |
| *** nosnos_ has quit IRC | 07:05 | |
| *** nosnos has joined #openstack-metering | 07:05 | |
| *** mgagne1 has joined #openstack-metering | 07:08 | |
| *** mgagne1 has joined #openstack-metering | 07:08 | |
| *** mgagne has quit IRC | 07:09 | |
| *** fandikurnia01 has quit IRC | 07:10 | |
| *** fandikurnia01 has joined #openstack-metering | 07:11 | |
| *** matsuhashi has quit IRC | 07:22 | |
| *** matsuhashi has joined #openstack-metering | 07:23 | |
| *** nadya_ has joined #openstack-metering | 07:23 | |
| *** matsuhas_ has joined #openstack-metering | 07:30 | |
| *** taplax has quit IRC | 07:30 | |
| *** bingbu has quit IRC | 07:31 | |
| *** bingbu has joined #openstack-metering | 07:32 | |
| *** matsuhashi has quit IRC | 07:34 | |
| *** nsaje has joined #openstack-metering | 07:34 | |
| *** eglynn has joined #openstack-metering | 07:35 | |
| *** nsaje has quit IRC | 07:35 | |
| *** fandikurnia01 has quit IRC | 07:38 | |
| *** fandikurnia01 has joined #openstack-metering | 07:38 | |
| *** nadya_ has quit IRC | 07:43 | |
| *** taplax has joined #openstack-metering | 07:44 | |
| *** mgagne1 has quit IRC | 07:45 | |
| *** sileht has joined #openstack-metering | 07:55 | |
| *** sileht_ has joined #openstack-metering | 07:55 | |
| *** sileht_ is now known as sileht | 07:56 | |
| *** sdake_ has quit IRC | 07:56 | |
| *** SergeyLukjanov has joined #openstack-metering | 08:04 | |
| *** nsaje has joined #openstack-metering | 08:14 | |
| *** matsuhas_ has quit IRC | 08:18 | |
| *** yfujioka has joined #openstack-metering | 08:19 | |
| *** matsuhashi has joined #openstack-metering | 08:19 | |
| *** matsuhashi has quit IRC | 08:24 | |
| openstackgerrit | ChenZheng proposed a change to openstack/ceilometer: Add i18n warpping for all LOG messages https://review.openstack.org/54683 | 08:24 |
| *** boris-42 has quit IRC | 08:26 | |
| *** boris-42 has joined #openstack-metering | 08:28 | |
| *** matsuhashi has joined #openstack-metering | 08:31 | |
| *** shardy_afk is now known as shardy | 08:46 | |
| *** nprivalova has joined #openstack-metering | 08:48 | |
| *** boris-42 has quit IRC | 08:49 | |
| *** suo has quit IRC | 08:52 | |
| *** nati_ueno has quit IRC | 08:59 | |
| *** lsmola has joined #openstack-metering | 09:02 | |
| *** suo has joined #openstack-metering | 09:16 | |
| *** dperaza has quit IRC | 09:17 | |
| *** bingbu has quit IRC | 09:25 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Move dispatcher a level up https://review.openstack.org/56949 | 09:26 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Use map_method from stevedore 0.12 https://review.openstack.org/56897 | 09:26 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Simplify the dispatcher method prototype https://review.openstack.org/57221 | 09:26 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Remove the collector submodule https://review.openstack.org/57151 | 09:26 |
| *** nprivalova has quit IRC | 09:52 | |
| *** nprivalova has joined #openstack-metering | 10:00 | |
| openstackgerrit | ChenZheng proposed a change to openstack/ceilometer: Add i18n warpping for all LOG messages https://review.openstack.org/54683 | 10:02 |
| nprivalova | jd__, hi! | 10:06 |
| *** nprivalova is now known as nadya_ | 10:06 | |
| jd__ | hi nadya_ | 10:07 |
| nadya_ | jd__, did you see my bp https://blueprints.launchpad.net/python-ceilometerclient/+spec/proxy-support ? | 10:09 |
| nadya_ | jd__, I just wanted to be sure that I can start uploading patches | 10:10 |
| jd__ | let me check | 10:10 |
| jd__ | nadya_: sure approved, though I wonder if that's not somethign that could go into oslo | 10:10 |
| jd__ | I think oslo has now a few common code for client | 10:11 |
| nadya_ | jd__, I'll check oslo code. but didn't see examples in nova or swift client. anyway, we will upload a patch and will have a chance to -1 :) | 10:14 |
| *** matsuhashi has quit IRC | 10:14 | |
| jd__ | nadya_: there's apiclient in oslo-incubator, I think sending your patch on this would be useful too | 10:16 |
| jd__ | ceilometerclient should switch to apiclient at some point | 10:16 |
| jd__ | (hint hint) | 10:16 |
| *** SergeyLukjanov is now known as _SergeyLukjanov | 10:16 | |
| *** _SergeyLukjanov has quit IRC | 10:17 | |
| *** egallen has joined #openstack-metering | 10:20 | |
| *** SergeyLukjanov has joined #openstack-metering | 10:21 | |
| *** ruhe has joined #openstack-metering | 10:21 | |
| *** ruhe has quit IRC | 10:22 | |
| *** ityaptin has joined #openstack-metering | 10:24 | |
| *** SergeyLukjanov is now known as _SergeyLukjanov | 10:24 | |
| *** SergeyLukjanov has joined #openstack-metering | 10:24 | |
| *** nsaje has quit IRC | 10:26 | |
| *** nsaje has joined #openstack-metering | 10:27 | |
| *** SergeyLukjanov has quit IRC | 10:30 | |
| *** nsaje has quit IRC | 10:31 | |
| nadya_ | guys, I have one more question about resource and resource_metadata storing in dbs. I was trying to understand why list is thrown away from resource_metadata. And now I do. As it turned out all info about resources are stored in db. But when resource is fetched Resource instance is created and _flatten_metadata is called. During this method we just remove all lists and sets. Question: why do we store resource in normal way and "flat" it on ever | 10:31 |
| nadya_ | request? Can we store flatten metadata? | 10:31 |
| *** fandikurnia01 has quit IRC | 10:32 | |
| *** egallen has quit IRC | 10:33 | |
| *** taplax has quit IRC | 10:37 | |
| nadya_ | the problem in this approach is in HBase support. I cannot store lists or dicts directly, only unicode(list). So every time during flattening I need to guess is the unicode string from HBase dict or list | 10:37 |
| jd__ | nadya_: I think flatten move nested data structure into something like key1.key2 = value, that should work for HBase too, no? | 10:41 |
| nadya_ | jd__, yes, but HBase cannot keep dict or list. only string or unicode. So there is a code https://github.com/openstack/ceilometer/blob/master/ceilometer/api/controllers/v2.py#L502 . To make it work with HBase I need to add some additional check to "if type(v) not in set([list, set]))". It should be like "if type(v) not in set([list, set])) or (type(v) is string and (v.contains('{') or v.contains(']'))" and so on. Looks very awful | 10:46 |
| jd__ | I see, so 'value' can't be a list neither, so sad | 10:49 |
| jd__ | so it's key/value both strings? brrr | 10:49 |
| *** ruhe has joined #openstack-metering | 10:50 | |
| nadya_ | jd__, yes. And during flattening we just throw sets and lists away. We store them in db but do not use this info (at least during resource fetching) | 10:51 |
| *** taplax has joined #openstack-metering | 10:53 | |
| jd__ | I see, I'm not sure how to fix this though | 10:53 |
| nadya_ | jd__, I have only one question for now: why we keep original data and flatten it every time? can we keep flatten data in db? | 10:54 |
| nadya_ | as I see from git dhellmann has created "flatten" method. maybe he may help me here :)? | 10:57 |
| jd__ | nadya_: good question, not sure about that actually; I think it's mainly historical and related to the MongoDB driver | 10:58 |
| *** ruhe has quit IRC | 10:58 | |
| *** SergeyLukjanov has joined #openstack-metering | 10:59 | |
| nadya_ | jd__, I'll try to implementing storing flatten metadata directly. If successful, I'll just ask community to review it. I think this patch will not contain 100s of lines. | 11:08 |
| *** ruhe has joined #openstack-metering | 11:09 | |
| *** sandywalsh has joined #openstack-metering | 11:12 | |
| *** lsmola has quit IRC | 11:24 | |
| nadya_ | jd__, I decided to store flatten data to Mongo. InvalidDocument: key 'dict_properties.key' must not contain '.' . hehe. So looks like flattening was created for Mongo to avoid this 'dot' problem | 11:26 |
| *** taplax has quit IRC | 11:29 | |
| nadya_ | jd__, but if any field contains '.' the problem came back. E.g. https://bugs.launchpad.net/ceilometer/+bug/1246264 | 11:30 |
| *** lsmola has joined #openstack-metering | 11:36 | |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/python-ceilometerclient: Avoid reset of repeat_actions attribute on alarm update https://review.openstack.org/57423 | 11:36 |
| *** taplax has joined #openstack-metering | 11:46 | |
| *** yfujioka has quit IRC | 11:52 | |
| *** lsmola has quit IRC | 11:53 | |
| jd__ | nadya_: yeah the thing is that MongoDB has no problem storing nested data structure and we can build queries around that without much problem | 11:56 |
| jd__ | nadya_: I think we should let the driver handle that internally actually | 11:56 |
| openstackgerrit | Cyril Roelandt proposed a change to openstack/ceilometer: Add a get_resources() method to the pollsters https://review.openstack.org/53150 | 12:00 |
| *** lsmola has joined #openstack-metering | 12:06 | |
| *** Shaan7 has quit IRC | 12:08 | |
| *** Shaan7 has joined #openstack-metering | 12:08 | |
| *** nosnos_ has joined #openstack-metering | 12:09 | |
| *** nosnos has quit IRC | 12:12 | |
| *** nosnos_ has quit IRC | 12:14 | |
| *** nadya_ has quit IRC | 12:16 | |
| *** nsaje has joined #openstack-metering | 12:16 | |
| *** ruhe has quit IRC | 12:21 | |
| *** nsaje has quit IRC | 12:23 | |
| *** nsaje has joined #openstack-metering | 12:23 | |
| eglynn | jd__: I've restored the expired https://review.openstack.org/53889 ... | 12:24 |
| eglynn | jd__: could you take a look at my latest comment in gerrit? | 12:24 |
| eglynn | (the argument against fixing on the oslo side in the first instance, i.e. consumability by distros via stable/havana) | 12:25 |
| *** ruhe has joined #openstack-metering | 12:26 | |
| *** chuck__ is now known as zul | 12:26 | |
| *** ruhe has quit IRC | 12:27 | |
| *** nsaje has quit IRC | 12:28 | |
| *** boris-42 has joined #openstack-metering | 12:30 | |
| *** ruhe has joined #openstack-metering | 12:31 | |
| *** egallen has joined #openstack-metering | 12:35 | |
| openstackgerrit | ChenZheng proposed a change to openstack/ceilometer: Add i18n warpping for all LOG messages https://review.openstack.org/54683 | 12:42 |
| *** egallen has quit IRC | 12:48 | |
| *** egallen has joined #openstack-metering | 12:49 | |
| *** changbl has quit IRC | 12:57 | |
| *** nsaje has joined #openstack-metering | 13:05 | |
| *** ruhe has quit IRC | 13:08 | |
| *** suo has quit IRC | 13:10 | |
| *** thomasem has joined #openstack-metering | 13:12 | |
| *** taplax has quit IRC | 13:15 | |
| *** boris-42 has quit IRC | 13:17 | |
| *** nprivalova has joined #openstack-metering | 13:18 | |
| *** boris-42 has joined #openstack-metering | 13:18 | |
| *** nsaje has quit IRC | 13:18 | |
| *** nsaje has joined #openstack-metering | 13:19 | |
| *** gordc has joined #openstack-metering | 13:19 | |
| *** herndon_ has joined #openstack-metering | 13:24 | |
| *** nsaje has quit IRC | 13:24 | |
| herndon_ | silhet: are you around? | 13:24 |
| herndon_ | err, sileht | 13:24 |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/ceilometer: Selectively import RPC backend retry config https://review.openstack.org/53889 | 13:26 |
| *** zoresvit has quit IRC | 13:27 | |
| *** bogdando has quit IRC | 13:28 | |
| sileht | herndon_, o/ | 13:29 |
| *** bogdando has joined #openstack-metering | 13:30 | |
| *** zoresvit has joined #openstack-metering | 13:30 | |
| *** taplax has joined #openstack-metering | 13:30 | |
| herndon_ | hi! - I'm just curious what's going on with replacing oslo.rpc with oslo.messaging, and if there's any help you need? | 13:30 |
| *** egallen has quit IRC | 13:31 | |
| *** egallen has joined #openstack-metering | 13:32 | |
| sileht | herndon_, I have made a good progress, I have proposed a patch oslo.messaging for what we needs for consuming notifications | 13:33 |
| sileht | herndon_, I have a working ceilometer with oslo.messaging, I have some remaining tests to convert to submit the patch | 13:33 |
| herndon_ | cool! | 13:34 |
| *** ruhe has joined #openstack-metering | 13:34 | |
| *** ruhe has quit IRC | 13:34 | |
| herndon_ | sileht: so you've got an oslo patch right now that needs to land before the ceilo one can make it in, is that right? | 13:35 |
| *** ruhe has joined #openstack-metering | 13:35 | |
| sileht | herndon_, yes | 13:35 |
| herndon_ | sileht, great thanks for the info. | 13:36 |
| sileht | herndon_, but perhaps the oslo.messaging patch doesn't perfectly fit for now, I have some interrogation about how oslo.messaging works | 13:36 |
| sileht | herndon_, for example the current implentation I have made for notification consuming create many connnections insteads of one | 13:37 |
| sileht | herndon_, or for the rpc part, we are no more allowed to have topic/queue name different | 13:38 |
| sileht | herndon_, when I have finished the ceilometer patch I will start some discustions on those subjects | 13:39 |
| herndon_ | ok, not sure I completely understand the implications of a matching topic/queue names, is that going to be a big problem? | 13:40 |
| *** zul has quit IRC | 13:41 | |
| thomasem | hey all | 13:41 |
| *** zul has joined #openstack-metering | 13:41 | |
| sileht | herndon_, not really, but this is a behavior change, so that needs discution | 13:41 |
| herndon_ | sileht, Ok. I'll keep my ears open | 13:42 |
| thomasem | Matching topic to queue names creates rigidity around how multiple consumers can use the same topic. | 13:42 |
| thomasem | So, for us we have downstream listening to one topic, StackTach listening to another, and soon we'll have to tell other openstack components to emit notifications on an additional topic in order to stand up Ceilometer for load testing and what-not. | 13:43 |
| sileht | thomasem, yes, oslo.messaging needs to allow this | 13:43 |
| thomasem | Current oslo.rpc allows it on the consumer side with the pool_name option. | 13:43 |
| thomasem | Ceilo doesn't allow a deployer to alter that config, though, which I'm thinking of patching. | 13:44 |
| thomasem | Currently it just assumes the topic/queue name in the collector. | 13:44 |
| *** Shaan7 has quit IRC | 13:45 | |
| thomasem | Sorry, I mean it just assume the queue name is the same as the topic and sets it that way in the collector without allowing us to configure whether we use a different queue. | 13:45 |
| thomasem | I think the reason for having topic = queue name on the producing side is so we never drop a notification on the floor. | 13:45 |
| thomasem | That way no, say notification.info, will ever be missed on the notification.info queue, because the producer stands it up along with binding it to the producer's exchange. | 13:46 |
| thomasem | Am I saying something you already know, though? :) | 13:47 |
| herndon_ | makes sense to me | 13:47 |
| sileht | sure | 13:47 |
| *** ityaptin has quit IRC | 13:48 | |
| thomasem | Yeah, I spent about 2 weeks digging into that with my optional samples branch that I abandoned for Dragon's changes. | 13:48 |
| *** ityaptin has joined #openstack-metering | 13:48 | |
| thomasem | So, anywho - I think there ought to be more configuration on the producing side to allow us to force other queues open, so we can cause no notifications to be dropped at restart for multiple consuming services. :) | 13:50 |
| thomasem | Of course, the dangerous side effect being an unused queue would overflow. | 13:50 |
| *** nsaje has joined #openstack-metering | 13:51 | |
| *** nsaje has quit IRC | 13:51 | |
| thomasem | But, I'd rather call that out in documentation and warn. | 13:51 |
| thomasem | Sometimes the best approach is to just say "Hey, if you do this, you're going to have a bad time." | 13:52 |
| *** nsaje has joined #openstack-metering | 13:52 | |
| *** nsaje has quit IRC | 13:52 | |
| *** nsaje has joined #openstack-metering | 13:52 | |
| herndon_ | doesn't that mean you'd have to configure the notification driver in nova to know all about the queues that ceilometer will be reading from? | 13:55 |
| *** julim has joined #openstack-metering | 14:04 | |
| *** ruhe has quit IRC | 14:10 | |
| *** ruhe has joined #openstack-metering | 14:12 | |
| thomasem | herndon_, if you wanted to guarantee no messages were dropped at restart. | 14:13 |
| thomasem | herndon_, If Ceilometer isn't already listening, that is. | 14:13 |
| thomasem | herndon_, otherwise, nah | 14:13 |
| thomasem | So, if you were to restart Nova and Ceilometer wasn't already listening for notifications.* on some arbitrary queue, the messages wouldn't start flowing in until Ceilometer establishes its queue. | 14:14 |
| thomasem | meaning all of the messages before Ceilometer setting up that queue would never make it to that queue. | 14:15 |
| *** ruhe has quit IRC | 14:16 | |
| thomasem | herndon_, Sorry, I was off getting an omelet. :) | 14:17 |
| *** jergerber has joined #openstack-metering | 14:18 | |
| *** thomasm has joined #openstack-metering | 14:22 | |
| *** thomasm is now known as Guest98408 | 14:23 | |
| *** thomasem has quit IRC | 14:23 | |
| *** Guest98408 is now known as thomasem | 14:23 | |
| thomasem | and my cable popped out of my laptop. :( | 14:24 |
| *** r4nd_ has joined #openstack-metering | 14:27 | |
| *** dperaza has joined #openstack-metering | 14:34 | |
| *** Shaan7 has joined #openstack-metering | 14:37 | |
| *** boris-42_ has joined #openstack-metering | 14:40 | |
| *** boris-42 has quit IRC | 14:41 | |
| openstackgerrit | Nadya Privalova proposed a change to openstack/ceilometer: Insertion in HBase should be fixed https://review.openstack.org/52670 | 14:42 |
| *** thomasem has quit IRC | 14:44 | |
| *** thomasem has joined #openstack-metering | 14:44 | |
| *** ekarlso has quit IRC | 14:51 | |
| *** ekarlso has joined #openstack-metering | 14:52 | |
| *** ekarlso has quit IRC | 14:52 | |
| *** ekarlso has joined #openstack-metering | 14:53 | |
| *** gordc has quit IRC | 14:55 | |
| *** changbl has joined #openstack-metering | 14:55 | |
| *** sandywalsh has quit IRC | 15:00 | |
| *** egallen has quit IRC | 15:01 | |
| *** sandywalsh has joined #openstack-metering | 15:03 | |
| *** nati_ueno has joined #openstack-metering | 15:03 | |
| lsmola | eglynn, hello | 15:11 |
| eglynn | lsmola: hey | 15:11 |
| lsmola | eglynn, would you have time to look at https://bugs.launchpad.net/horizon/+bug/1249279 ? | 15:11 |
| eglynn | lsmola: one sec | 15:11 |
| lsmola | eglynn, it would be maybe also good to add to documentation how to use each meter | 15:12 |
| *** herndon_ has quit IRC | 15:12 | |
| lsmola | eglynn, I guess not it can be pretty hard for users to get stats from the samples | 15:12 |
| eglynn | lsmola: add documentation where? http://docs.openstack.org/developer/ceilometer or? | 15:13 |
| lsmola | eglynn, yeah probably, it would be for whole statistics section | 15:13 |
| *** egallen has joined #openstack-metering | 15:14 | |
| eglynn | lsmola: not sure what you mean by "for whole statistics section" | 15:14 |
| lsmola | eglynn, e.g. how to get average value of some time span from cumulative meter | 15:14 |
| eglynn | lsmola: a-ha, got it | 15:14 |
| lsmola | eglynn, or how to get duration of instance from the instance meter | 15:14 |
| lsmola | eglynn, this is also kind of stuff somebody was asking me | 15:14 |
| eglynn | lsmola: actually it would be good also to guide the user around which aggregates make no sense for a a cumulative meter | 15:16 |
| lsmola | eglynn, yes | 15:16 |
| *** ruhe has joined #openstack-metering | 15:17 | |
| eglynn | lsmola: e.g. sum ... that's what I was trying to get at with this bug https://bugs.launchpad.net/horizon/+bug/1243694 | 15:18 |
| eglynn | lsmola: (... suggesting such meaningless aggregates are excluded from the dashboard, though the concept could also be explained in docco) | 15:18 |
| eglynn | lsmola: anyway, gimme a couple mins to finish up what I'm doing and I'll look at the bug | 15:18 |
| lsmola | eglynn, cool | 15:19 |
| *** ruhe has quit IRC | 15:25 | |
| *** nprivalova has quit IRC | 15:25 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: [WIP] replace oslo.rpc by oslo.messaging https://review.openstack.org/57457 | 15:26 |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: [WIP] replace oslo.rpc by oslo.messaging https://review.openstack.org/57457 | 15:32 |
| *** egallen_ has joined #openstack-metering | 15:35 | |
| *** egallen has quit IRC | 15:36 | |
| *** egallen_ is now known as egallen | 15:36 | |
| *** nsaje has quit IRC | 15:38 | |
| *** nsaje has joined #openstack-metering | 15:39 | |
| *** fnaval has quit IRC | 15:39 | |
| *** fnaval has joined #openstack-metering | 15:40 | |
| eglynn | lsmola: can you define what you mean by "Global" in the context of https://bugs.launchpad.net/horizon/+bug/1249279 ? | 15:41 |
| *** gordc has joined #openstack-metering | 15:42 | |
| lsmola | eglynn, oh, it's naming from the guy who had that patch before | 15:43 |
| *** nsaje has quit IRC | 15:43 | |
| lsmola | eglynn, I guess it doesn't mean anything :-) | 15:43 |
| lsmola | eglynn, just maybe that it shows statistics over all tenants, each tenant for one table line | 15:44 |
| *** fnaval has quit IRC | 15:44 | |
| eglynn | lsmola: I don't have the dashboard panels in front on me now, but isn't there a dropdown box that allows selection of 'Group by' as either project_id or resource_id? | 15:46 |
| *** nprivalova has joined #openstack-metering | 15:46 | |
| eglynn | lsmola: (or is that just for the line-charting panel as opposed to the table view?) | 15:46 |
| lsmola | eglynn, nope for tables, there is group by Tenant, AVG in 30 days | 15:46 |
| lsmola | eglynn, that approach doesn;t fit well | 15:47 |
| *** boris-42_ is now known as boris-42 | 15:47 | |
| eglynn | lsmola: so the choice of per-tenant and average is *fixed* for that view? | 15:47 |
| lsmola | eglynn, we can probably yes | 15:48 |
| lsmola | eglynn, sorry, I haven't finished the question | 15:48 |
| lsmola | eglynn, yes | 15:48 |
| eglynn | lsmola: hmmm, making it fixed in one view and selectable in other seems very odd to me | 15:49 |
| lsmola | eglynn, and we can probably add some selectbox to switch date range, and group_by | 15:49 |
| lsmola | eglynn, yeah, it is plan to make it switchable | 15:49 |
| eglynn | lsmola: yep, I think that would lead to a much more consistent UX | 15:49 |
| lsmola | eglynn, though that page was waiting for UX review, so it was possible ot will be reworked totally | 15:50 |
| *** nati_ueno has quit IRC | 15:50 | |
| eglynn | lsmola: sure, but in our ignorance as non-UX-experts, we can still make incremental improvements to what's there | 15:50 |
| eglynn | lsmola: (pending any major rework that may not occur for a long time ...) | 15:51 |
| lsmola | eglynn, I guess, as it seems there is no UX coming soon | 15:51 |
| eglynn | lsmola: one sec | 15:51 |
| lsmola | eglynn, exactly | 15:51 |
| eglynn | lsmola: so on the issue of calculating "computing (min - max) for each resource" ... | 15:55 |
| eglynn | lsmola: there's a wrinkle there in that these cumulative values are in general not strictly monotonic | 15:55 |
| lsmola | eglynn, hmm | 15:56 |
| eglynn | lsmola: I was just explaining this to a colleague wrt to cpu time, so I'll be lazy and copy'n'paste that for background ... | 15:56 |
| lsmola | eglynn, I do see in my test data that it is not strictly rising | 15:56 |
| eglynn | lsmola: http://fpaste.org/55446/13849628/raw/ | 15:56 |
| eglynn | lsmola: so (max - min) over a long period may not be accurate if the instance restarts or pause/resume cycle occurred | 15:57 |
| lsmola | eglynn, hmm, that is not very good | 15:58 |
| lsmola | eglynn, is there some way how to get it? | 15:58 |
| eglynn | lsmola: it would be more accurate to take the sum of the local maxima minus the first value | 15:58 |
| eglynn | lsmola: that would require a fine-grained statistics query (e.g. period=60s) then some local calculation | 15:59 |
| lsmola | eglynn, how can you get this using API? given I want to do it via one aggregate query | 15:59 |
| eglynn | lsmola: you can't currently, IIUC | 15:59 |
| lsmola | eglynn, hmm, ok | 16:00 |
| eglynn | lsmola: (i.e. can't get that directly from the API as an built-in aggregate) | 16:00 |
| lsmola | eglynn, does it act the same for all cumulative values? | 16:00 |
| eglynn | lsmola: I think that it's an artifact of how libvirt does its accounting for cumulative values | 16:00 |
| eglynn | lsmola: it certainly occurs for CPU time | 16:01 |
| *** Ruetobas has quit IRC | 16:01 | |
| eglynn | lsmola: you'd need to check individually for the other meters ... spin up an instance, pause, resume, then check how the values progressed over that period | 16:01 |
| lsmola | eglynn, right, is it possible somebody already did that? if not I will do it and describe it to documentation | 16:02 |
| lsmola | eglynn, this is very needed documentation feature, given not all stats of the same type acts uniform | 16:03 |
| *** Ruetobas has joined #openstack-metering | 16:03 | |
| eglynn | lsmola: "somebody already did that" == "somebody already figured out the logic needed to accurately sum resetable cumulative meters over long durations"? | 16:04 |
| lsmola | eglynn, yes | 16:04 |
| eglynn | lsmola: IIRC yolanda was looking at this towards the end of the grizzly cycle / start of havana | 16:05 |
| lsmola | eglynn, if the behaviour is somewhere docuemnted, it would be great | 16:05 |
| lsmola | eglynn, where can I find him..her? | 16:05 |
| eglynn | lsmola: I'm not sure if yolanda's work on that saw the light of day | 16:05 |
| lsmola | hehe | 16:05 |
| lsmola | eglynn, well I am poking about deleting those tables, until I figure it out, now it's there only to confuse users :-) | 16:06 |
| eglynn | lsmola: https://launchpad.net/~yolanda.robla | 16:06 |
| *** nprivalova has quit IRC | 16:06 | |
| *** Ruetobas has quit IRC | 16:08 | |
| lsmola | eglynn, it is the last thing I had kept from the previous owner of the patch :-) so I wanted keep at least anything :-D | 16:08 |
| lsmola | eglynn, though the general chart might be enough for now | 16:08 |
| lsmola | eglynn, + integration of the sparklines to each tab | 16:09 |
| eglynn | lsmola: I agree, especially if the choice of aggregates is made made meter-specific | 16:09 |
| eglynn | (as per https://bugs.launchpad.net/horizon/+bug/1243694) | 16:09 |
| lsmola | eglynn, ok, i think it was even previously mentioned that it make more sense to have table stats in the sections of openstak where it belongs | 16:11 |
| lsmola | eglynn, rather than have it all together | 16:11 |
| eglynn | lsmola: yeah, in context with the rest of the instance info for example | 16:11 |
| lsmola | eglynn, ok, I will tell Horizon guys I will delete the tables until i figure it out | 16:12 |
| lsmola | eglynn, yeah | 16:12 |
| eglynn | lsmola: fair enough | 16:12 |
| lsmola | eglynn, also having it as sparklines is much better then just one aggregate value | 16:12 |
| lsmola | eglynn, s/then/than | 16:12 |
| *** fnaval has joined #openstack-metering | 16:13 | |
| *** egallen has quit IRC | 16:13 | |
| *** Ruetobas has joined #openstack-metering | 16:13 | |
| lsmola | eglynn, ok maybe you can just reasure me with the approach on gauge | 16:15 |
| lsmola | eglynn, it's just so I know for a future | 16:15 |
| lsmola | eglynn, I will be doing enhancement of overview pages soon | 16:15 |
| eglynn | lsmola: not sure what you mean by reassure you with the approach on gauge | 16:16 |
| eglynn | lsmola: what particular aspect? | 16:16 |
| lsmola | eglynn, "network" - Gauge - Duration , in the bug | 16:16 |
| lsmola | eglynn, those are the stat that just shows 1 or 0, even in the chart, I suspect the timestamp is most important here | 16:17 |
| eglynn | lsmola: one sec | 16:18 |
| lsmola | eglynn, it's basically whole NetworkUsage in the bug, or e.g. instance meter | 16:18 |
| eglynn | lsmola: it's emitted as a constant 1 ... https://github.com/openstack/ceilometer/blob/master/ceilometer/network/notifications.py#L87 | 16:22 |
| eglynn | lsmola: ... presumably to indicate existence | 16:22 |
| eglynn | lsmola: the duration can then be inferred from the first and last timestamps | 16:23 |
| lsmola | eglynn, yes | 16:25 |
| *** nsaje has joined #openstack-metering | 16:25 | |
| lsmola | eglynn, but first and last of row of 1 | 16:25 |
| lsmola | eglynn, so counting the real duration would require to go through the samples | 16:26 |
| lsmola | eglynn, or use (number of 1 - and number of 0) * some sort of period of sample | 16:27 |
| eglynn | lsmola: the duration is returned in the statistics response | 16:27 |
| lsmola | eglynn, so it is like duration of that particular state? | 16:27 |
| lsmola | eglynn, or duration of that sample | 16:28 |
| eglynn | lsmola: duration_start and duration_end of all the samples that were rolled into the aggregates reported for that statistics query | 16:28 |
| eglynn | lsmola: i.e. first timestampt, last timestamp | 16:29 |
| lsmola | eglynn, hm | 16:29 |
| lsmola | eglynn, so for 111111110001111111000000001111 | 16:29 |
| lsmola | eglynn, does it returns duration of 1? | 16:30 |
| lsmola | eglynn, or of all together? | 16:30 |
| eglynn | lsmola: each 1 is an aggregate statistic right? | 16:30 |
| *** nprivalova has joined #openstack-metering | 16:31 | |
| eglynn | lsmola: each statistic has a duration_start & duration_end | 16:31 |
| eglynn | lsmola: (rolling up all the samples that fed into that period) | 16:31 |
| lsmola | eglynn, well that is the thing, I cant get it by aggregate | 16:31 |
| lsmola | eglynn, I can get soemthin like sum 8, count 10 > means there was 8 ones and 2 zeros | 16:32 |
| lsmola | eglynn, with aggregate of some time | 16:32 |
| eglynn | lsmola: what period are you using? | 16:33 |
| eglynn | lsmola: and what kind of group-by? | 16:33 |
| eglynn | lsmola: per tenant, per resource, what? | 16:33 |
| lsmola | eglynn, for the chart, the period is computed | 16:33 |
| lsmola | eglynn, yeah for chart, it is everything | 16:34 |
| eglynn | lsmola: this isn't the chart, right? | 16:34 |
| eglynn | lsmola: this is the table view, no? | 16:34 |
| lsmola | eglynn, yeah, both I guess | 16:34 |
| lsmola | eglynn, for table it's avg value, so it doesnt make sense | 16:34 |
| lsmola | eglynn, it will return you like 0.5 | 16:35 |
| eglynn | lsmola: for the chart, say you group by project | 16:35 |
| lsmola | eglynn, which mean it was off half of the month | 16:35 |
| eglynn | lsmola: then you can all the networks for that tenant right? | 16:35 |
| lsmola | eglynn, right | 16:35 |
| lsmola | eglynn, the it doesnt make sense to show any of the sum, avg, min ,max | 16:36 |
| eglynn | lsmola: right, in plain English what do you want to display ... "the total number of network-hours for that tenant"? | 16:36 |
| lsmola | eglynn, it has to be more complex computation considering count and sum probably | 16:36 |
| lsmola | eglynn, yes, it should be all duration | 16:37 |
| lsmola | eglynn, that is waht is in the description of the meters | 16:37 |
| eglynn | lsmola: I think "duration" was overloaded there | 16:38 |
| eglynn | lsmola: it would be more accurate to say it represents existence | 16:38 |
| lsmola | eglynn, yeah | 16:38 |
| lsmola | eglynn, and duration can be inferred from it | 16:38 |
| eglynn | lsmola: exactly | 16:39 |
| lsmola | eglynn, well, I will probably deep dive into docs :-D | 16:39 |
| eglynn | lsmola: right, if you want that total to be accurate, I think you'd need two things: | 16:39 |
| lsmola | eglynn, ach, it will be much easier with the hardware stats :-D | 16:39 |
| eglynn | (1) a relatively short period | 16:39 |
| eglynn | (2) walk the list of stats, and sum the volume * period length | 16:40 |
| eglynn | the relatively short period means the (duration_end-duration_start) is approximated by the period length | 16:40 |
| *** jsergent has joined #openstack-metering | 16:41 | |
| lsmola | eglynn, yeah | 16:41 |
| eglynn | lsmola: cool, I think we were talking past each other earlier | 16:41 |
| eglynn | lsmola: prolly my fault, I'm trying to hold two conversations at once | 16:42 |
| eglynn | lsmola: all straightened out now? | 16:42 |
| lsmola | eglynn, :-) | 16:42 |
| lsmola | eglynn, yes, thank you, it is much clearer now | 16:42 |
| eglynn | lsmola: cool, laters then ... | 16:42 |
| lsmola | eglynn, later | 16:42 |
| *** nsaje has quit IRC | 16:47 | |
| *** nsaje has joined #openstack-metering | 16:47 | |
| *** nprivalova has quit IRC | 16:49 | |
| eglynn | nprivalova: FYI I finally got to looking at https://etherpad.openstack.org/p/ceilometer-aggregation and left a few comments on the etherpad | 16:50 |
| *** nsaje has quit IRC | 16:52 | |
| *** nprivalova has joined #openstack-metering | 16:56 | |
| nprivalova | eglynn, thanks for comments! will look at this by tomorrow | 16:56 |
| eglynn | nprivalova: cool, np! | 16:56 |
| *** nprivalova has quit IRC | 17:09 | |
| *** nsaje has joined #openstack-metering | 17:25 | |
| *** jonnalley has joined #openstack-metering | 17:26 | |
| *** Bada has joined #openstack-metering | 17:32 | |
| *** sileht is now known as sileht_ | 17:40 | |
| *** sileht_ is now known as sileht | 17:40 | |
| *** jonnalley has quit IRC | 17:41 | |
| *** boris-42 has quit IRC | 17:42 | |
| *** nprivalova has joined #openstack-metering | 17:47 | |
| *** SergeyLukjanov has quit IRC | 17:48 | |
| *** nprivalova has quit IRC | 17:58 | |
| *** ruhe has joined #openstack-metering | 17:59 | |
| *** mirrorbox has quit IRC | 18:02 | |
| *** ruhe has quit IRC | 18:04 | |
| *** mirrorbox has joined #openstack-metering | 18:04 | |
| *** nprivalova has joined #openstack-metering | 18:04 | |
| *** sdake_ has joined #openstack-metering | 18:11 | |
| *** sdake_ has joined #openstack-metering | 18:11 | |
| *** eglynn is now known as eglynn-afk | 18:20 | |
| *** nprivalova has quit IRC | 18:21 | |
| *** ildikov has quit IRC | 18:26 | |
| *** nsaje has quit IRC | 18:32 | |
| *** nsaje has joined #openstack-metering | 18:33 | |
| *** nsaje has quit IRC | 18:33 | |
| *** nsaje has joined #openstack-metering | 18:33 | |
| *** Bada has quit IRC | 18:35 | |
| *** mgagne has joined #openstack-metering | 18:35 | |
| *** sandywalsh has quit IRC | 18:52 | |
| *** sandywalsh has joined #openstack-metering | 19:08 | |
| *** nadya_ has joined #openstack-metering | 19:14 | |
| *** nsaje has quit IRC | 19:41 | |
| *** nadya_ has quit IRC | 19:49 | |
| openstackgerrit | John Herndon proposed a change to openstack/ceilometer: Add a specialized Event Type model and db table https://review.openstack.org/41316 | 20:00 |
| *** nadya_ has joined #openstack-metering | 20:01 | |
| *** SergeyLukjanov has joined #openstack-metering | 20:08 | |
| *** r4nd_ has quit IRC | 20:10 | |
| *** sdake_ is now known as randallburt | 20:10 | |
| *** nadya_ has quit IRC | 20:12 | |
| *** randallburt is now known as sdkae | 20:19 | |
| *** ildikov has joined #openstack-metering | 20:19 | |
| *** sdkae is now known as sdake_ | 20:19 | |
| *** asalkeld is now known as are_we_all_chang | 20:20 | |
| *** are_we_all_chang is now known as angusss | 20:20 | |
| *** sdake_ is now known as randallburt | 20:21 | |
| *** angusss is now known as asalkeld | 20:26 | |
| *** randallburt is now known as sdake_ | 20:26 | |
| *** sandywalsh has quit IRC | 20:30 | |
| *** r4nd_ has joined #openstack-metering | 20:35 | |
| *** sandywalsh has joined #openstack-metering | 20:44 | |
| *** SergeyLukjanov has quit IRC | 20:49 | |
| openstackgerrit | Claudiu Belu proposed a change to openstack/ceilometer: Fixes Hyper-V Inspector disk metrics bug https://review.openstack.org/57014 | 20:50 |
| *** jergerber has quit IRC | 20:57 | |
| *** eglynn-afk is now known as eglynn | 20:58 | |
| *** shardy has quit IRC | 21:01 | |
| *** terriyu has joined #openstack-metering | 21:01 | |
| *** shardy has joined #openstack-metering | 21:03 | |
| *** litong has joined #openstack-metering | 21:06 | |
| *** herndon_ has joined #openstack-metering | 21:12 | |
| *** nati_ueno has joined #openstack-metering | 21:46 | |
| *** sdake_ has quit IRC | 21:57 | |
| *** zul has quit IRC | 22:00 | |
| dragondm | Erf, it looks like next weeks Ceilometer meeting is on Thanksgiving day? US folks might miss out on that... | 22:05 |
| *** ryanpetrello has quit IRC | 22:05 | |
| *** ryanpetrello has joined #openstack-metering | 22:12 | |
| thomasem | Yeah, is there any way we can shift that around? I know that's abnormal. | 22:13 |
| *** ekarlso has quit IRC | 22:15 | |
| *** terriyu has quit IRC | 22:15 | |
| *** ekarlso has joined #openstack-metering | 22:17 | |
| gordc | thanksgiving was last month ;) | 22:19 |
| *** ryanpetrello_ has joined #openstack-metering | 22:23 | |
| *** ryanpetrello has quit IRC | 22:26 | |
| *** ryanpetrello_ is now known as ryanpetrello | 22:26 | |
| *** ildikov has quit IRC | 22:30 | |
| *** thomasem has quit IRC | 22:36 | |
| *** changbl has quit IRC | 22:38 | |
| *** gordc has quit IRC | 22:42 | |
| *** zul has joined #openstack-metering | 22:45 | |
| *** litong has quit IRC | 22:46 | |
| *** eglynn has quit IRC | 22:56 | |
| *** changbl has joined #openstack-metering | 23:17 | |
| *** herndon_ has quit IRC | 23:20 | |
| openstackgerrit | John Herndon proposed a change to openstack/ceilometer: Add a specialized Event Type model and db table https://review.openstack.org/41316 | 23:26 |
| openstackgerrit | John Herndon proposed a change to openstack/ceilometer: Add a Trait Type model and db table https://review.openstack.org/42407 | 23:33 |
| *** sdake_ has joined #openstack-metering | 23:37 | |
| *** yfujioka has joined #openstack-metering | 23:43 | |
| *** thomasem has joined #openstack-metering | 23:53 | |
| *** julim has quit IRC | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!