Skip to content

Weekly Meeting 2016 07 28

Kristen Carlson Accardi edited this page Jul 28, 2016 · 3 revisions

Agenda

opens (5 minutes)

No P1 bugs this week.

Progress on P1 features for sprint 2:

https://github.com/01org/ciao/issues?utf8=%E2%9C%93&q=is%3Aissue%20is%3Aopen%20label%3AP1%20milestone%3A%22Sprint%202%22

Minutes

##Minutes

#ciao-project Meeting

Meeting started by kristenc at 16:00:36 UTC. The full logs are available at ciao-project/2016/ciao-project.2016-07-28-16.00.log.html .

Meeting summary

Meeting ended at 16:59:07 UTC.

Action Items

  • erick0zcr to write up what is needed to provide distro packages.
  • fuentess consider planning how to test volume attach/detach of volume to/from running workload
  • fuentess consider first the VM use case, later containers / clear-containers / bare metal
  • prior to face to face, kristenc to set up real ceph cluster
  • kristenc ask ahkok for delivery details on hardware order

Action Items, by person

  • fuentess
    • fuentess consider planning how to test volume attach/detach of volume to/from running workload
    • fuentess consider first the VM use case, later containers / clear-containers / bare metal
  • kristenc
    • prior to face to face, kristenc to set up real ceph cluster
    • kristenc ask ahkok for delivery details on hardware order
  • UNASSIGNED
    • erick0zcr to write up what is needed to provide distro packages.

People Present (lines said)

  • kristenc (57)
  • markusry (35)
  • tcpepper (30)
  • sameo (9)
  • mrkz (6)
  • _erick0zcr (5)
  • leoswaldo (3)
  • ciaomtgbot (2)
  • fuentess (1)

Generated by MeetBot_ 0.1.4

.. _MeetBot: http://wiki.debian.org/MeetBot

###Full IRC Log

16:00:36 <kristenc> #startmeeting
16:00:36 <ciaomtgbot> Meeting started Thu Jul 28 16:00:36 2016 UTC.  The chair is kristenc. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:36 <ciaomtgbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:00:46 <kristenc> o/
16:00:50 <mrkz> o/
16:00:50 <markusry> o/
16:00:51 <_erick0zcr> o/
16:01:37 <sameo> o/
16:01:40 <leoswaldo> o/
16:01:49 <tcpepper> o/
16:02:43 <kristenc> #topic Opens
16:02:43 <kristenc> anyone have any opens?
16:03:00 <_erick0zcr> I have one o/
16:03:15 <kristenc> _erick0zcr, ok.
16:03:17 <tcpepper> I have a small FYI:  I'm adding an item to the template agend for reviewing prior meeting's ACTION items in the minutes
16:03:22 <mrkz> not an open just a reminder obedmr is on vacations until Monday 01 Aug
16:03:44 <kristenc> tcpepper, ok - did you want to do that today then?
16:04:01 * kristenc digs up last weeks minutes.
16:04:43 <kristenc> all the action items were assigned to mcastelino, and I believe he's out.
16:05:01 <tcpepper> #LINK https://github.com/01org/ciao/wiki/Weekly-Meeting-2016-07-14 has a few also
16:05:11 * kristenc looks
16:05:39 <tcpepper> of the 7 there, the last three are done, but I'm not sure about the first 4
16:06:08 <kristenc> leoswaldo, ACTION: leoswaldo will update the API wiki with API's not implemented
16:06:26 * kristenc obedmr and leoswaldo will put together the list of candidate APIs for heat for the team
16:06:38 * kristenc leoswaldo will investigate automation using swagger and godoc
16:06:41 <_erick0zcr> Regarding packaging, the specfile for packaging CIAO in rpm is finished. The thing is, some docker dependencies are not in the fedora repositories and I think it will take more time to have ciao packaged in fedora official repos. Have you considered us to provide the .deb or .rpm packages?
16:06:47 <tcpepper> I just marked the trailing three on the list with " [done]"
16:06:54 * kristenc leoswaldo will update the API docs on the wiki to two sections: ciao apis and nova apis
16:07:44 <kristenc> speak of the devil.
16:08:24 <kristenc> leoswaldo, did you get the list of the action items assigned to you?
16:08:39 <mrkz> leoswaldo: check https://github.com/01org/ciao/wiki/Weekly-Meeting-2016-07-14#action-items-by-person
16:09:35 <leoswaldo> update the API wiki with API's not implemented: DONE
16:10:09 <tcpepper> leoswaldo: can you edit the notes there to capture " [done]" on the line so we see it plz :)
16:10:19 <leoswaldo> okas
16:10:40 <tcpepper> thx :)
16:11:48 <tcpepper> we can move on if obedmr and leoswaldo are updating the items there
16:12:02 <tcpepper> _erick0zcr has a question ^^
16:15:43 <tcpepper> _erick0zcr: it's easy for us to assume package creation and deliver is done by distributions.  if that's going to be too slow to achieve, we need to consider CI for package creation and testing and where to host the packages.
16:17:15 <_erick0zcr> tcpepper: yes, it will be slow. So we can consider to provide the packages, at least temporarily, I think.
16:18:14 <tcpepper> _erick0zcr: can you write up what all you think is needed? we can size that versus other things and decide if we wait for distros or fill the gap
16:18:49 <tcpepper> does clearlinux automatically notice ciao releases, autospec, and auto test the release?
16:18:50 <_erick0zcr> tcpepper: sure
16:21:17 <tcpepper> shall we move on?
16:22:02 <kristenc> #action erick0zcr to write up what is needed to provide distro packages.
16:22:25 <kristenc> yes - let's move on. any more opens?
16:22:44 <mrkz> no from my side
16:23:06 <kristenc> ok - normally we'd do a bug scrub, but we have no P1 bugs this week.
16:23:20 <tcpepper> o/
16:23:39 <kristenc> #topic Sprint 2 progress
16:23:55 <kristenc> #link https://github.com/01org/ciao/issues?utf8=0.000000E+002    16:24:11 <kristenc> that is a query with all P1 sprint 2 features.
16:24:32 <kristenc> I added a new label this week to indicate which ones have been staged into the topic/storage branch.
16:25:34 <kristenc> I don't think markusry is here.
16:25:41 <markusry> he's here
16:26:24 <kristenc> great. would it be possible for you to mark which of the P1 items you are working on have been staged already?
16:27:02 <markusry> sure
16:27:25 * tcpepper will also at the point he's got something
16:27:31 * sameo closed issue #386
16:27:36 <kristenc> markusry, you are done with sprint 2 work after friday, correct? (until the face-to-face)
16:28:02 <tcpepper> kristenc: can we do merges into your branch when you're out?
16:28:11 <markusry> Yes.  I'm not sure I'll get everything finished by Friday though
16:28:22 <markusry> I'll try to get AttachVolume and DetachVolume working
16:28:25 <kristenc> tcpepper, yes - it is not on my fork, it's on the main ciao tree.
16:28:35 <sameo> tcpepper: We technically can at least.
16:28:38 <markusry> Had a bit of a set back today.
16:28:54 <markusry> An upgrade to Ubuntu 16.04 failed
16:29:00 <kristenc> markusry, if we can get attach/detach working end to end by the end of sprint 2 I will feel like we accomplished something.
16:29:26 <kristenc> it would be nice to have more of the other flows working, but considering the time limitations, maybe we should just focus on that flow.
16:29:26 <markusry> kristenc: Okay, so I should be able to get that working by tomorrow, once I've fixed my dev environment.
16:29:30 <markusry> That's my goal.
16:30:29 <kristenc> I am almost done with the error handling for attach/detach in ciao-controller, and expect to send a PR for that today.
16:30:59 <kristenc> markusry, which part of the attach/detach flow is still missing for you? actual block driver and stats?
16:31:10 <markusry> Exactly.
16:31:29 <markusry> I've figured out which QMP commands I need to make it work
16:31:35 <markusry> I just need to modify launcher to send them
16:31:55 <markusry> And then I need to do stats
16:32:00 <markusry> but that should be easy
16:32:15 <kristenc> from the controller side - I'm missing error handling and stats handling.
16:32:20 <markusry> As far as I can tell we'll need QMP 2.5 for detach to work
16:32:35 <markusry> Sorry QEMU 2.5
16:32:35 <tcpepper> are ya'll focusing on the attach/detach ahead of launch, or also after launch?
16:32:48 <markusry> tcpepper: Yes
16:32:54 <markusry> At least I am
16:33:04 <sameo> markusry: ahead of launch ?
16:33:10 <markusry> Yes
16:33:23 <kristenc> not sure what "ahead of launch" means?
16:33:39 <markusry> volumes specified in workloads I guess
16:33:44 <tcpepper> yes
16:34:09 <sameo> yes.
16:34:25 <sameo> i.e. not attaching a volume when the instance is running.
16:34:26 <kristenc> so - to clarify - markusry is working on attach/detach after a workload has been launched, not at workload launch time, correct?
16:34:41 <markusry> Exactly.
16:34:49 <markusry> This is the trickiest piece
16:34:58 <markusry> Which is why I started on this first
16:35:06 <kristenc> so that is the first flow we are working on - and likely to be the only one we'll get done at this point.
16:35:13 <markusry> Agreed.
16:35:26 <tcpepper> ok so plan is to skip the easy stuff and focus on hard
16:35:26 <markusry> Once I have it done, the other workflows should be easy to add
16:36:03 <kristenc> sameo, you merged the cli changes for the block and image api?
16:36:10 <sameo> kristenc: Yes.
16:36:15 <kristenc> (merged into master)
16:36:21 <sameo> yes.
16:36:39 <tcpepper> kristenc: does your storage branch need rebased on master then?
16:36:55 <kristenc> tcpepper, yes.
16:37:04 <sameo> kristenc: Ah, sorry about that.
16:37:42 <kristenc> tcpepper, I assume you are working on the image and block apis now. Can you focus on the ones that are used in the attach/detach flow?
16:38:30 <tcpepper> in the flow for attach/detach to/from running instance, sure
16:39:31 <tcpepper> is anybody assigned yet to user documentation?  docu would help me here as I work up test of code I'm writing. eg:
16:40:31 <tcpepper> what needs run where to create a volume, which ciao-cli commands are the ones to attach it to an instance, any caveats and how to validate that persistent data was seen in the running instance
16:41:06 <tcpepper> eg: do we have a test vm that reads a file and increments the number in it or something like that?  an external program that mounts the volume when the instance is not running to see it was incremented?
16:41:32 <markusry> not that I'm aware of
16:41:42 <kristenc> ciao-cli has --help
16:44:48 <tcpepper> #ACTION fuentess consider planning how to test volume attach/detach of volume to/from running workload
16:45:17 <tcpepper> and in our focus, markus has mentioned qemu, are we focused on VM's not containers?
16:45:28 <kristenc> tcpepper, yes - correct.
16:45:31 <markusry> yes.
16:45:37 <markusry> Haven't even looked at containers yet
16:45:46 <fuentess> tcpepper: sure, need to add image and storage tests to the test plan
16:46:19 <markusry> The tricky thing is that there is no configuration at the moment
16:46:39 <tcpepper> #ACTION fuentess consider first the VM use case, later containers / clear-containers / bare metal
16:46:42 <markusry> So we'll need to make sure that /etc/ceph is correctly set up on the compute nodes
16:46:58 <mrkz> albertom: ^
16:47:01 <kristenc> markusry, yes - I think we should do small scale testing for now.
16:47:05 <mrkz> that's easy with ansible I'd say
16:47:24 <markusry> right.  At the moment I'm testing launcher on a single device, which is about as small as you can get
16:47:25 <kristenc> it should reduce the amount of manual config we have to do.
16:47:33 <kristenc> markusry, me too.
16:47:44 <kristenc> I'm using the ceph container.
16:48:01 <markusry> Yes me too.  Just wanted to point this out.
16:48:55 <kristenc> #action prior to face to face, kristenc to set up real ceph cluster
16:49:18 <kristenc> I'll make sure when we are all together we have a real storage cluster to do our integration with.
16:49:33 <tcpepper> #ACTION kristenc ask ahkok for delivery details on hardware order
16:50:10 <kristenc> I have 4 machines now I can use. they just aren't ideally configured. if we don't get the hw we need, we can always configure these 4 non-ideally.
16:50:52 <tcpepper> when ya'll say "the ceph container", which specifically do you mean?  There's a number on docker hub.
16:52:00 <markusry> ceph/demo
16:52:01 <kristenc> tcpepper, I created a document with some notes on setting up the ceph container. let me paste the link again for you.
16:52:54 <tcpepper> https://github.com/01org/ciao/wiki/Storage  ?
16:53:17 <kristenc> tcpepper, many of these notes are specific to our corporate infrastructure, so it is internal.
16:53:20 <kristenc> I posted it to you there.
16:55:00 <kristenc> ok - we have about 5 more minutes. I think we are about done with this topic. any other issues/comments about sprint 2?
16:55:34 <kristenc> I think the only other thing to sort of keep track of is that we're going to have to rethink sprint 3 now.
16:56:04 <kristenc> since there's going to be a lot of basic flows that don't get done for sprint 2.
16:56:21 <kristenc> I'll alert amy today.
16:56:29 <markusry> Okay.  I hope from the launcher side they should be pretty quick
16:56:50 <markusry> But I agree, they'll fall into sprint 3 now
16:57:23 <kristenc> i'm going to see if I can look at our sprint 2 list and adjust it to reflect reality.
16:58:58 <kristenc> ok, if nobody has any thing else, I'm calling it a meeting.