15:00:52 <rhochmuth> #startmeeting monasca
15:00:53 <openstack> Meeting started Wed May 25 15:00:52 2016 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:55 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:57 <openstack> The meeting name has been set to 'monasca'
15:01:14 <rhochmuth> o/
15:01:17 <kamil_> o/
15:01:18 <FlintHPE> o/
15:01:20 <igorn> hi \o
15:01:20 <bklei> o/
15:01:22 <rbak> o/
15:01:22 <koji> o/
15:01:24 <laszloh> o/
15:01:24 <tomasztrebski> o/
15:01:25 <jayahn> o/
15:01:25 <hosanai> o/
15:01:25 <rbrndt> o/
15:01:26 <slogan> o/
15:01:36 <shinya_kwbt> o/
15:01:37 <rhochmuth> impressive
15:01:40 <tgraichen> o/
15:02:06 <rhochmuth> agenda is at, https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:02:14 <rhochmuth> 1. Discuss mid-cycle
15:02:14 <rhochmuth> 2. Discuss adding the ability to publish logs to Kafka topics based on a list of dimension filters and keystone roles. For example, if operational logs and audit logs need to be stored in separate Kafka topics (TSV - HPE)
15:02:14 <rhochmuth> 3. Reviews:
15:02:14 <rhochmuth> https://review.openstack.org/#/c/301355/ (agent hang fix)
15:02:14 <rhochmuth> https://review.openstack.org/#/c/286281/ (kv hint)
15:02:14 <rhochmuth> 4. Deterministic alarms
15:02:14 <rhochmuth> 5. Periodic notifications
15:02:43 <rhochmuth> So, it looks like i can travel to wherever we decide to host
15:02:49 <rhochmuth> the mid-cycle
15:03:09 <rhochmuth> but, from last week, it looks like we can't all travel to the same location at the same time
15:03:27 <rhochmuth> and for the most part it was split down the center
15:03:31 <bklei> will probably need to allow virtual either way
15:04:09 <rhochmuth> i think my preference is to run it all remote this time around
15:04:28 <rhochmuth> inclusivity is more important to me
15:04:31 <slogan> works for me
15:04:35 <bklei> ok with me too
15:04:38 <rhochmuth> and it is hard to justify travel for just half the team
15:05:05 <rhochmuth> after the barcelona summit the mid-cycles will be predefined
15:05:13 <rhochmuth> by the openstack organization
15:05:32 <rhochmuth> and i think at that point everyone should push hard to make the design summits
15:05:39 <slogan> I was against travel, but now that the TSA has vastly been improved....
15:05:43 <rhochmuth> i think the date is sometime in february
15:05:48 <rhochmuth> and forgot where
15:06:09 <jayahn> glad to hear TSA has been improved.
15:06:18 <rhochmuth> so, unless anyone objects, let's do another remote mid-cycle in july
15:06:24 <rhochmuth> we just need to decide on the week
15:06:51 <hosanai> let's fight against time zone :-)
15:07:16 <rhochmuth> well, we can also mixup the mid-cycles to do half morning
15:07:18 <rhochmuth> half night
15:07:35 <rhochmuth> so folks in Asia time-zone are taken care of too
15:07:56 <hosanai> thanks!
15:08:00 <jayahn> great!
15:08:00 <rhochmuth> np
15:08:32 <rhochmuth> are any of the weeks preferable?
15:08:59 <bklei> either works for me -- 7/18 or 7/25
15:09:29 <rhochmuth> well, let's wrap up on the date next week
15:09:36 <rhochmuth> not everyone is here
15:09:56 <rhochmuth> and maybe we can modify the doodle
15:10:02 <bklei> good idea
15:10:28 <rhochmuth> ok, so i'll close on this for now, and move on to agenda
15:10:40 <rhochmuth> tsv: are u here?
15:11:02 <tsv_> yes
15:11:16 <rhochmuth> #topic logging api
15:11:21 <rhochmuth> the floor is yours
15:11:28 <tsv_> thanks
15:12:06 <tsv_> i wanted to discuss adding support for publish logs to specific kafka topics based on filter criteria (using dimensions)
15:12:46 <tsv_> this is to differentiate audit logs from operational logs, for example, where the topic retention could be different for audit logs
15:13:19 <tomasztrebski> so instead of single topic in api - rather multiple topics detected from logs or fallback to default ?
15:13:36 <tsv_> correct
15:13:49 <rhochmuth> so, basically we could use a diemsion such as log_type = log or audit
15:13:51 <tsv_> sdake, i see we already have support for multiple topics, but cannot filter
15:13:59 <tsv_> yes
15:14:02 <rhochmuth> and audit logs would end-up on a separate topic
15:14:17 <tomasztrebski> yes - there are multiple topics supported but all messages goes for all topics
15:14:28 <tomasztrebski> as you mentioned - that's not selective
15:14:30 <rhochmuth> audit topic would potentailly have a different retention period
15:14:55 <slogan> sounds like a nice idea
15:14:57 <rhochmuth> so, i think the idea is pretty simple, and justification and pretty clear
15:15:01 <tomasztrebski> +1
15:15:14 <rhochmuth> thanks tomasztrebski
15:15:17 <rhochmuth> +1
15:15:27 <kamil_> and who will consume this new topics?
15:16:07 <tsv_> kamil, in our implementation, logstash filter would read from this topic and forward to a different Elasticsearch index
15:17:29 <kamil_> could you not filter directly in kibana?
15:17:30 <koji> you mean that the difference between the audit log and current log is only the retention, right?
15:17:48 <tomasztrebski> I think thay could but that does not resolve problem of retention
15:17:57 <tsv_> koji, if Kafka security is enabled, the ACLs could be different too
15:18:35 <rhochmuth> i think there are multiple differences, with retention being probably the main motification
15:18:42 <rhochmuth> motiivation
15:18:44 <kamil_> tomasztrebski, yeah that's true
15:18:46 <koji> at first, i thought that you propose the feature for the customer, like cloud trail of AWS
15:19:00 <rhochmuth> no, this is all internal
15:19:10 <koji> ok, thank you
15:19:28 <tsv_> tomasztrebski, why do you say that ? kafka supports per topic retention settings right ?
15:20:30 <kamil_> tsv: tomasztrebski was answering my question
15:20:55 <tomasztrebski> I am just thinking on using dimensions for that
15:20:59 <tsv_> kamil, got it thanks
15:21:19 <tomasztrebski> if that's the routing - perhaps more verbose approach and simple add routing property into the log
15:22:00 <tomasztrebski> also Kamil - there's a question of an idea of log-metrics where logs needs to go through particular topic in order to be transformed into metrics if certain severity happens
15:23:01 <rhochmuth> tomasztrebski: are you proposing that we add a spefic field, or just use dimensions
15:24:02 <kamil_> a think a field is easier to analyze in logstash
15:24:27 <slogan> almost seems like it might be a better thing as a field
15:25:02 <slogan> but only because when I think of dimensions, I think of metadata about my metric, not details about how it is handled
15:25:12 <rhochmuth> slogan: agree
15:25:16 <jayahn> i agree on that, might be better as a field.
15:25:24 <rhochmuth> the dimensions have been really about identity
15:25:39 <slogan> if we find the need for more control data, perhaps there is some other thing that is a sibling to dimensions for that
15:25:56 <rhochmuth> so, a "type" field might be preferred
15:26:07 <tomasztrebski> in metrics there is value_meta, but I don't think that's suitable
15:26:18 <rhochmuth> tomasztrebski: agree
15:26:47 <slogan> or "handling": { "type": xyz, "retention": abc, ...}
15:27:01 <slogan> something like that
15:28:18 <tsv_> in addition to the field (or dimension), don't we also need a keystone role ? for example, if we want to restrict write access to the audit topic ?
15:29:15 <rhochmuth> not sure we need a role, but that could be a separate discussion
15:29:25 <rhochmuth> so, i would like to get to rest of agenda
15:29:30 <tsv_> rhochmuth, ok
15:29:37 <rhochmuth> tsv: can you submite  blueprint
15:29:45 <tsv_> thanks, sure will do
15:29:49 <tomasztrebski> also I think that that might suit monasa-common quite nice, if you don't see any objections
15:29:49 <rhochmuth> then we can discuss, comment further
15:30:07 <tomasztrebski> oh...ok sorry, I didn't see that we are done with this for now...sry
15:30:15 <rhochmuth> np
15:30:20 <tsv_> tomasztrebski: sure, thanks all for the support
15:30:27 <rhochmuth> i'm rushing us through
15:30:40 <kamil_> please distribute the blueprint to us. thx
15:30:40 <rhochmuth> i think a field is the general concensus
15:30:58 <rhochmuth> then slogan's idea is a good one too
15:31:10 <tsv_> rhochmuth: ok
15:31:12 <rhochmuth> possibly making it a dictionary to allow for additional attributes
15:31:40 <rhochmuth> although, i would prefer calling the new field something like attributes, rather than handling
15:31:47 <slogan> yep
15:31:57 <slogan> or just "meta"
15:32:03 <rhochmuth> but, in gneral, i think we hace some concensus/direction
15:32:17 <rhochmuth> meta, also might be nice
15:32:27 <rhochmuth> ok, need to move on
15:32:41 <rhochmuth> #topic reviews
15:32:48 <rhochmuth> https://review.openstack.org/#/c/301355/ (agent hang fix)
15:33:03 <bklei> that's me -- how do you guys think that's looking?
15:33:03 <rhochmuth> so, i think this is ready for merging
15:33:19 <rhochmuth> there are several +1's from the hpe tam
15:33:22 <rhochmuth> team
15:33:22 <rbrndt> tested it over the weekend, didn't see any errors or missing metrics
15:33:34 <rhochmuth> tomasz: you had looked at earlier
15:33:41 <rhochmuth> are you ok with a merge?
15:33:48 <bklei> good to hear, we are hitting that bug a bunch, but we deployed that fix on a staging node, no hangs yet
15:33:50 <rhochmuth> thanks rbrndt
15:34:09 <tomasztrebski> I've had it running for 3 days and never saw anything that would suggest that agent has stopped
15:34:26 <rhochmuth> tomasztrebski: thx for all the testing
15:34:32 <bklei> +1
15:35:07 <rhochmuth> tomasztrebski: if you +1, i'll merge it
15:35:50 <rhochmuth> so, related to that review, we had talked about multi-processing/threading the plugins
15:36:04 <rhochmuth> so all plugins would run as their own thread/process
15:36:24 <rhochmuth> bklei: are you planning on continuing with that development
15:36:52 <rhochmuth> or is this an area where the joe/michael would work on?
15:36:56 <bklei> i'm not sure about that rhochmuth -- i thought that was joe
15:36:58 <bklei> yeah
15:37:07 <tomasztrebski> I think that the idea is in overall very good - to sandbox each plugin
15:37:20 <rbak> Yeah, I think the plan was for Joe to take over from here
15:37:33 <bklei> joe ok with that?
15:37:43 <rhochmuth> ok, i don't think joe is here, but will follow-up with him and hoppal
15:37:49 <bklei> cool
15:37:58 <rhochmuth> i think he is ok, just the time thing
15:38:05 <bklei> sure
15:38:17 <rhochmuth> tomasztrebski: agree
15:38:42 <rhochmuth> ok, so will close on this topic, merge current code, and hoe/hoppal to follow-up
15:38:45 <rhochmuth> joe
15:38:51 <bklei> cool, thx
15:39:04 <rhochmuth> https://review.openstack.org/#/c/286281/ (kv hint)
15:39:20 <bklei> me again -- i think this one's been sitting a while
15:39:24 <rhochmuth> so, i'll need to take another look, but it sounds like you think it is ready for a merge
15:39:36 <rhochmuth> will try and review again today
15:39:49 <bklei> yeah -- ryanb did some testing, didn't see any improvement, but didn't hurt either
15:39:59 <rhochmuth> right
15:40:05 <rbrndt> yup
15:40:06 <bklei> sorta taking vertica support at their word
15:40:13 <rhochmuth> but it is hard to test know in this case
15:40:22 <bklei> yeah
15:40:27 <rhochmuth> so, agree, if vertica recommends, then we should enable
15:40:36 <rhochmuth> hopefully you'll see an improvemnt if prod
15:40:39 <bklei> that's what i'm operating on
15:40:44 <rhochmuth> #topic Deterministic alarms
15:41:16 <rhochmuth> tomasztrebski: looks like  some code is ready for merging
15:41:48 <tomasztrebski> monasca-common here is ready to merge IMHO and without it I cannot proceed with thresh and api
15:41:55 <rhochmuth> https://review.openstack.org/#/c/292753/
15:42:11 <rbrndt> I notice thresh failed with the current common changes
15:42:21 <rbrndt> and I tried building locally with the same result
15:42:54 <tomasztrebski> for api - I've just fixed one last issue with tempests so that should be fine as well, thresh and ui are finished for me as well
15:42:55 <rbrndt> don't know if the error is in the thresh or common change though
15:43:15 <tomasztrebski> hmm...I haven't looked at thresh that much assuming that lack of certain fields in model is causing that issue
15:43:50 <tomasztrebski> and looking at log from gate right now it looks like that actually the problem
15:44:06 <tomasztrebski> for instance: cannot find symbol 2016-05-20 13:01:00.659 | [INFO]   symbol:   method isDeterministic()
15:45:03 <rhochmuth> so, if common is merged, gate should pass
15:45:32 <rhochmuth> as your review for common i'm assuming adds isDeterminstic
15:45:49 <rhochmuth> is that correct"
15:46:26 <tomasztrebski> that and couple other things like grammar
15:46:57 <rhochmuth> rbrndt: is it possible your local build didn't get installed into maven repo?
15:47:19 <rbrndt> i can try building again, perhaps it was misaligned
15:47:29 <rhochmuth> i think you need to pull monasa-common, and then mvn install
15:47:37 <rbrndt> yeah, I did that
15:48:15 <rhochmuth> so, i'll let you reverify and then see if we can resolve
15:48:26 <rhochmuth> if resolved, then will merge
15:48:42 <rhochmuth> tomasztrebski: sound good?
15:48:55 <rhochmuth> so, will wait for all clear from rbrndt
15:50:38 <rhochmuth> should we move on
15:51:07 <rhochmuth> sorry, just checking if we can close for now, and move on next topic
15:51:39 <tomasztrebski> i have nothing to add, will reverify that my self as well and post a comment with my results
15:51:53 <rhochmuth> tomasztrebski: thx
15:52:04 <rhochmuth> #topic Periodic notifications
15:52:28 <rhochmuth> so, mhoppal, what is status
15:52:40 <rhochmuth> and is this ready for merging, presumably
15:52:46 <mhoppal> the api review
15:52:49 <mhoppal> is ready for reviews
15:52:50 <rhochmuth> or more development required
15:52:51 <mhoppal> and merging
15:52:58 <mhoppal> the client, ui and notification
15:53:03 <mhoppal> have some comments to address
15:53:12 <rhochmuth> ok, thanks
15:53:16 <mhoppal> client should be ready today
15:53:32 <rhochmuth> thx
15:53:55 <rhochmuth> will move on, just wanted to give folks a heads-up on that bit of work since it is a new feature too
15:54:08 <rhochmuth> #topic open
15:54:25 <rhochmuth> we have about 5 minutes left
15:54:29 <FlintHPE> may I put in a quick shameless plug for the Monasca-Transform review (https://review.openstack.org/#/c/315245)?  ;-)
15:54:41 <rhochmuth> FlintHPE: thansk for reminder
15:54:48 <rhochmuth> turns out i didnt' have +2 privs
15:54:54 <rhochmuth> will get that today and merge
15:55:01 <FlintHPE> cool...thanks!
15:55:02 <rhochmuth> sorry about delay
15:55:13 <FlintHPE> no worries
15:55:48 <shinya_kwbt> I wrote blue print about monasca-ui pagination style. https://blueprints.launchpad.net/monasca/+spec/horizon-pagination-style
15:56:12 <rhochmuth> thanks shinya_kwbt
15:56:48 <shinya_kwbt> To adopt horizon pagination style needs to api sort option.
15:57:09 <shinya_kwbt> needs to chanage sort option or add new option.
15:57:34 <rhochmuth> so sorry, i didn't review, but i'll review, and we should discuss next week
15:57:42 <rhochmuth> does that sound ok
15:57:52 <shinya_kwbt> Thanks.
15:58:02 <rhochmuth> so, this can be first topic next week
15:58:43 <rhochmuth> rbrndt: you''ll want to review this one too
15:58:56 <rbrndt> alright
15:59:21 <rhochmuth> so, time has almost up again
15:59:40 <rhochmuth> any last gasp topoics
16:00:00 <rhochmuth> i keep thinking we need more time
16:00:25 <rhochmuth> i have a meeting after this, but i'll be in the monasca room
16:00:30 <rhochmuth> going to close down the meeting
16:00:34 <rhochmuth> thanks everyone
16:00:42 <hosanai> thanks & bye
16:00:50 <koji> thanks
16:00:58 <rhochmuth> bye hosanai, koji
16:01:12 <rhochmuth> #endmeeting