02:59:40 <fengshengqin> #startmeeting zun
02:59:41 <openstack> Meeting started Tue Apr  3 02:59:40 2018 UTC and is due to finish in 60 minutes.  The chair is fengshengqin. Information about MeetBot at http://wiki.debian.org/MeetBot.
02:59:42 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
02:59:44 <openstack> The meeting name has been set to 'zun'
02:59:51 <fengshengqin> #topic Roll Call
02:59:52 <hongbin> o/
03:00:06 <hongbin> i knew kevin is on vacation today
03:00:42 <fengshengqin> we may have a short meeting
03:00:48 <hongbin> sure
03:00:55 <fengshengqin> #topic Blueprints
03:01:03 <fengshengqin> 1. OpenStack as a virtual Kubernetes node (assignee: kevinz)
03:01:10 <fengshengqin> #link https://blueprints.launchpad.net/zun/+spec/zun-connector-for-k8s
03:01:18 <fengshengqin> let's skip it
03:01:26 <fengshengqin> 2. Introduce quota for containers (assignee: TBD)
03:01:34 <fengshengqin> #link https://blueprints.launchpad.net/zun/+spec/quota-support
03:01:40 <hongbin> this one is assigned to kien
03:01:53 <hongbin> it looks kien is not here
03:01:59 <hongbin> so we can skip this as well :)
03:02:11 <fengshengqin> 3. Manage the reverse disk for container images (assignee: fengshengqin)
03:02:18 <fengshengqin> #link https://blueprints.launchpad.net/zun/+spec/manage-container-image-disk
03:02:35 <fengshengqin> i have some ideas for it
03:02:48 <hongbin> what ideas?
03:03:07 <fengshengqin> first,we would  reverse disk for container images
03:03:18 <hongbin> yes
03:03:34 <fengshengqin> 2. we may get the image disk info by 'docker system df'
03:04:10 <fengshengqin> 3. check the size of free_disk for loading a new image
03:04:27 <fengshengqin> 4. update the disk if the images is deleted
03:04:41 <fengshengqin> 5. as we know, a image may has multilayers.
03:05:09 <fengshengqin> so if avail_disk < image_size, we may return exception
03:05:37 <fengshengqin> or  parse the layers of image, and check which has loaded, which not.
03:06:13 <fengshengqin> then we can calculate the size we need
03:06:24 <fengshengqin> that's all.
03:06:45 <hongbin> since image are with layers, i don't think you can get the exact disk size that will be taken by the image
03:07:32 <hongbin> however, you can run the 'df' command again after the image is pulled
03:07:33 <fengshengqin> i will check "docker system df"
03:07:50 <hongbin> sure
03:08:03 <fengshengqin> yes, we can update database by df
03:08:08 <hongbin> what i am commenting is "if avail_disk < image_size, we may return exception"
03:08:44 <hongbin> i don't think it is necessary to raise the exception based on the image size
03:09:22 <hongbin> because it is hard to estimate the size to be taken
03:09:28 <fengshengqin> i know
03:09:33 <hongbin> however, this is minor
03:10:00 <hongbin> for this one, i think we can do this step by step
03:10:10 <fengshengqin> we may add a congifure to choose
03:10:25 <hongbin> which config?
03:11:12 <fengshengqin> whether need to parse mutil layers
03:11:37 <hongbin> i think it will be hard to parse
03:11:45 <hongbin> anyway
03:12:01 <hongbin> a question though
03:12:14 <fengshengqin> yes, i will think about it more
03:12:24 <hongbin> if the disk is not enough, how about deleting some image?
03:13:09 <hongbin> you can do "docker images" to check which image is not used
03:13:11 <fengshengqin> no,we may choose other host to create container
03:13:36 <hongbin> ok
03:13:49 <fengshengqin> but it is not safe
03:13:58 <fengshengqin> delete images
03:14:12 <hongbin> i think you can find which image is unused
03:14:30 <hongbin> if an image is not used by any container, is it safe to delete ?
03:15:14 <hongbin> #link https://forums.docker.com/t/command-to-remove-all-unused-images/20/2
03:15:35 <fengshengqin> how about adding a task to delete unused images
03:15:50 <hongbin> yes, it can be a periodic task, that is fine
03:16:48 <fengshengqin> we can create a new bp
03:16:54 <hongbin> sure
03:17:29 <fengshengqin> anything else for this?
03:17:35 <hongbin> for this image reserve feature, my suggestion is to make the design/implementation as simple as possible
03:18:02 <fengshengqin> ack
03:18:25 <hongbin> ok
03:18:34 <hongbin> nothing else from me, go ahead
03:18:47 <fengshengqin> #topic Bugs
03:18:54 <fengshengqin> 1. Cannot create container with kata runtime (assignee: hongbin)
03:19:00 <fengshengqin> #link https://bugs.launchpad.net/zun/+bug/1754552
03:19:02 <openstack> Launchpad bug 1754552 in Zun "Cannot create container with kata runtime" [Critical,Triaged] - Assigned to hongbin (hongbin034)
03:19:24 <hongbin> for this one, i am still waiting for the feedback from kata team
03:19:31 <fengshengqin> ok
03:19:34 <fengshengqin> 2. Error on running privsep helper command (assignee: hongbin)
03:19:34 <hongbin> we can revisit this bug at the next meeting
03:19:41 <fengshengqin> #link https://bugs.launchpad.net/zun/+bug/1749342
03:19:43 <openstack> Launchpad bug 1749342 in Zun queens "Error on running privsep helper command" [Undecided,Fix committed] - Assigned to hongbin (hongbin034)
03:19:52 <hongbin> for this one, all hte patches are merged, so this bug is fixed
03:20:02 <hongbin> thanks for reviewing the patches
03:20:04 <fengshengqin> great!
03:20:17 <fengshengqin> #topic Open Discussion
03:20:46 <hongbin> any question ?
03:20:50 <fengshengqin> no
03:20:58 <hongbin> then we finish :)
03:21:12 <fengshengqin> see you next time
03:21:19 <hongbin> see you
03:21:26 <fengshengqin> #endmeeting