20:59:06 <flwang> #startmeeting zaqar
20:59:07 <openstack> Meeting started Mon Jan  4 20:59:06 2016 UTC and is due to finish in 60 minutes.  The chair is flwang. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:59:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:59:11 <openstack> The meeting name has been set to 'zaqar'
20:59:14 <flwang> #topic roll call
20:59:15 <flwang> o/
20:59:48 <vkmc> o/
21:00:02 <Eva-i> hello
21:00:45 <flwang> it's our first  meeting in 2016, i wish all you guys had a good xmas & new year holiday
21:02:07 <vkmc> merry xmas and happy new year for you too!
21:02:23 <ryansb> \o
21:02:28 <Eva-i> yeah, very merry
21:02:54 <flwang> ok, i don't have much new topics just those old ones
21:03:19 <flwang> client, UI, puppet and some new features
21:03:38 <flwang> #topic zaqar client
21:04:03 <flwang> i'm still working on the subscription patches and please review them asap
21:04:15 <flwang> since they're blocking the integration with ceilometer :)
21:04:24 <flwang> yep, ceilometer integration
21:05:13 <flwang> a core reviewer of ceiloemter just pinged me about this, they want to use zaqar as notifier of alarm
21:05:34 <flwang> but they can't do that since we're missing it in our client
21:07:19 <ryansb> can you link them?
21:07:20 <flwang> the mitaka M-2 is Jan 16-22
21:07:24 <ryansb> or is there a topic?
21:07:28 <Eva-i> flwang: I recently faced a problem when I was unable to use subscriptions, just because there were no defaul values for some settings related to subscriptions in zaqar.conf, hope ceilometer team know about this
21:07:43 <flwang> #link https://review.openstack.org/249395
21:07:56 <flwang> #link https://review.openstack.org/261280
21:08:05 <flwang> #link https://review.openstack.org/261282
21:08:12 <flwang> #link https://review.openstack.org/261303
21:08:15 <Eva-i> flwang: we noticed it with vkmc and I filled a bug report https://bugs.launchpad.net/zaqar/+bug/1529912
21:08:16 <openstack> Launchpad bug 1529912 in zaqar "No default values for pipelines in [storage] section in zaqar.conf" [Undecided,New]
21:09:26 <ryansb> thanks
21:09:28 <flwang> Eva-i: thanks, it's a 'valid' bug :)
21:10:37 <Eva-i> oki
21:10:50 <flwang> anything else we should talk about the zaqar client ?
21:11:39 <flwang> #topic UI
21:11:49 <flwang> no progress from my side
21:12:19 <Eva-i> no progress from me either
21:12:27 <flwang> i'm stuck at this patch https://review.openstack.org/#/c/258474/
21:12:44 <ryansb> Eva-i: wow, didn't know about lp#1529912
21:12:59 <flwang> if there is anybody know angular JS well, pls help take a look and feel free upload a new patch set
21:13:09 * ryansb does not sorry
21:13:50 <flwang> ryansb: no worries, i will bug horizon team to get some help
21:14:01 <flwang> jasondotstar: are you around?
21:15:26 <flwang> ok, seems we can skip the puppet topic
21:15:38 <flwang> #topic new features
21:15:46 <flwang> 1. delayed queues/messageas
21:16:01 <flwang> #link https://review.openstack.org/260859
21:16:13 <flwang> i got some good comments from ryansb
21:17:16 <flwang> in summary, the basic idea is creating a new delay field/attribute for message and queue, so that the messages in the queue can't be visible in the 'delay' seconds
21:17:47 <vkmc> oh angular :|
21:18:48 <flwang> as for the delay queues in SQS, pls ref http://docs.aws.amazon.com/AWSSimpleQueueService/latest/SQSDeveloperGuide/sqs-delay-queues.html
21:18:59 <flwang> any questions?
21:20:02 <Eva-i> flwang: no questions now, I need to investigate
21:21:44 <flwang> ryansb: could you pls explain your comment? re "Adding arbitrary delays doesn't seem like a "core" use case, and it'll  probably put more load on our storage backends than it appears."
21:23:47 <ryansb> yeah, so I meant two things. The latter is that having both visible (standard) and not-yet-visible (delayed) messages is likely to reduce the hit rate for paginated queries we use against our backends
21:24:00 <ryansb> esp ones with less advanced query capabilities (redis)
21:24:45 <flwang> ryansb: yep, the paginated query is a trouble
21:24:50 <ryansb> and that'd be sort-of alright since it would only affect folks that use that feature, so my comment was intended to be "if we implement this, we have to make sure it doesn't affect the normal case"
21:25:25 <flwang> technically, we won't impact the normal case
21:25:55 <flwang> currrently, we're using the .exists() method to check if the queue is eixsting before query messages
21:26:06 <ryansb> and for the former (it not being a core use case) I was trying to get across that I've not really seen many use cases where this is needed, and I don't know how popular the SQS version of this feature is
21:26:43 <ryansb> but given that I've used SQS frequently but never encountered a need for delayed messages, I'm not sure how many others have
21:26:57 <flwang> and we can replace it with .get() to get the metadata of queue to decide if it's a delay queue
21:27:15 <ryansb> none of this means I don't like the feature, I just want to make sure there's a clear use case before we go implementing the spec
21:27:15 <flwang> and then we can decide if add the 'delay' query
21:27:25 <ryansb> yeah, that's sensible
21:27:55 <flwang> but if we want to support delayed message in normal queues, then we may always add the query condition
21:27:56 <ryansb> (and was the type of solution I was hoping for when I said "don't affect the normal case")
21:28:19 <ryansb> yeah, I'm not much of a fan of interleaving delayed and nondelayed messages
21:28:54 <ryansb> if users need it, we can do it as long as we're aware of the tradeoff.
21:28:55 <flwang> ryansb: no problem, we can discuss it more :)
21:29:06 <flwang> ryansb:  really appreciated for the valuable comments
21:29:17 <flwang> ryansb: yep, cool
21:30:16 <flwang> 2. another spec is about adding more reserved attributes of queues, see https://review.openstack.org/#/c/257622/
21:31:07 <flwang> in summary, it will support 'ttl' and 'max msg size' in queue metadata so that we can have flexible queues which don't have to use the global settings defined in zaqar.conf
21:37:00 <vkmc> having the ttl and max_msg_size attributes sounds like a very good addition to me
21:37:07 <vkmc> we need to give better use of the metadata
21:37:09 <vkmc> as we have it right now
21:37:23 <vkmc> and the delay queue feature
21:37:30 <vkmc> it's something I see no harm in adding it as a flavor
21:38:17 <ryansb> +1 for more metadata on queues
21:38:26 <flwang> awesome
21:38:36 <Eva-i> this feature need to be implemented in a way that no users can abuse the system
21:38:36 <flwang> ok, i have done all my topics
21:38:40 <flwang> anything else?
21:39:15 <Eva-i> flwang: maybe docs, but you wanted this meeting to be short
21:40:03 <Eva-i> *needs to be implemented
21:41:12 <Eva-i> flwang: okay, we can discuss it in a chat
21:41:59 <flwang> no, go ahead
21:42:08 <Eva-i> okay
21:42:10 <flwang> Eva-i: we have 17 mins for sure
21:43:12 <Eva-i> I posted a patch with Zaqar config-reference. https://review.openstack.org/#/c/263405/
21:43:21 <Eva-i> You can render it and see if it's normal.
21:43:57 <Eva-i> As you can see one person doesn't like that Zaqar configuration tables can't be generated now, but let's wait for other reviewers
21:44:21 <Eva-i> *autogenerated
21:44:52 <ryansb> personally, I think manual docs > no docs
21:45:29 * ryansb doesn't have a meeting topic
21:46:21 <Eva-i> ryansb: thanks for your +1, I agree
21:48:22 <flwang> Eva-i: did you see the -1 from aj?
21:48:52 <Eva-i> flwang: that's what I'm saying now
21:49:03 <Eva-i> flwang: of course
21:49:50 <flwang> Eva-i: i think it would be nice if you can update the commit message to explain how it's generated
21:49:58 <flwang> it's really 'manual' actually
21:50:12 <flwang> it would be nice if we can add the reason
21:50:56 <Eva-i> okay, I'll update the commit message and now will mark this patch as WIP, alright?
21:51:22 <Eva-i> flwang: I'm gonna sleep after this meeting
21:51:56 <flwang> Eva-i: you don't have to update it as WIP, i think :)
21:52:05 <flwang> ok, we can discuss it later
21:52:26 <flwang> #topic open discussion
21:52:34 <flwang> anything else?
21:52:36 <Eva-i> flwang: hm, okay, then I think I'll try to update commit message now
21:53:56 <flwang> Eva-i: it's no rush, take your time
21:54:22 <flwang> ok, if there is no more topic, let's close the meeting and back to our channel
21:54:38 <flwang> thank you, guys
21:54:53 <ryansb> thanks folks :)
21:54:57 <Eva-i> ty
21:56:10 <flwang> #endmeeting