13:01:19 #startmeeting 13:01:19 Meeting started Wed Jul 10 13:01:19 2013 UTC. The chair is aglitke. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:19 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:01:33 It added a _ after my nick name 13:01:53 * hlwanghl_ slaps shaohef around a bit with a large fishbot 13:01:55 FYI, the meeting will be logged and minutes will be uploaded to the kimchi website 13:02:18 #meetingname scrum 13:02:18 The meeting name has been set to 'scrum' 13:02:43 #info Agenda: 1) 1.0 Features and planning 13:02:55 #info 2) Roundtable / Open Discussion 13:03:00 any other items? 13:03:31 Customer user study, though we can hit that during the round table 13:03:38 ok 13:03:38 char in name? 13:04:13 shaohef: Let's do that in the round table too. There is a good discussion going on on the list as well. 13:04:24 aglitke: yes. 13:04:33 aliguori: do you have anything to add? 13:05:10 heh, guess not :) 13:05:20 I think it was something you said 13:05:26 #topic 1.0 Features and planning 13:05:27 aliguori_: left the room :) 13:05:43 #link https://github.com/kimchi-project/kimchi/issues?labels=enhancement&milestone=&page=1&state=open 13:06:18 The above link will take you to a list of all the enhancements we have planned. 13:06:29 You can filter by sprint and release. 13:06:35 I owe creating an enhancement to add "login" 13:06:42 As of now, no major features are tracked. 13:07:01 I think we also need one to finish the ISO scanning in the template creation flow 13:07:09 1.0 will be released on 9/30 13:07:13 It seems that most of the features in the page are UI enhancement. 13:07:27 AdamKingIT, agree 13:07:52 #action AdamKingIT add enhancement for login 13:08:07 #action aglitke add enhancement for iso scanning flow 13:08:08 that because UI is not strong enough 13:08:14 Federation and debugreports? 13:08:35 #link https://github.com/kimchi-project/kimchi/wiki/Todo 13:08:42 Here is our rough TODO list 13:08:59 both federation and debugreports are represented. 13:09:02 I have another action... 13:09:08 In fact, I'm wandering if bing_bu's get static config and my change static config are still on list of interest? 13:09:19 Right now we have about 20 days left in sprint one 13:09:34 AdamKingIT #action Add other user stories to the GitHub wiki 13:09:35 AdamKingIT: just go ahead and use #action to add it 13:09:49 put #action first 13:10:08 #action AdamKingIT Add other user stories to the GitHub wik 13:10:15 :) 13:10:22 lvroyce: your vm-rename still not merged? 13:10:46 guests: migration to remote system. Does this feature reach outside of Kimchi scope? 13:10:55 aglitke: hlwanghl_ need the template update 13:11:13 So in sprint 1 the only major enhancement we have is to complete the rename to Kimchi 13:11:15 edit template 13:11:32 shaohef: I don't really understand that. I can edit the template in the ui now. 13:11:35 bing_bu, I guess they are depending on yours, I want to make it commonly used for config change... 13:12:42 aglitke: that is because hongliang delete the template and add a new one ,but not edit api from back end 13:12:45 agiltke, sprint 1 will end in 9/30, the same as realease 1.0? 13:12:53 aglitke: I worked it around by 1) remove it 2) recreate it 13:13:06 aglitke, I'm doing the template create flow in dev-ui, do we still need that before release 1.0? 13:13:13 sming: https://github.com/kimchi-project/kimchi/wiki/Planning-1.0 13:13:18 Kimchi 1.0 - Sprint 1: 7/30/13 13:13:23 Kimchi 1.0 - Sprint 2: 8/30/13 13:13:27 Kimchi 1.0 - GA: 9/30/13 13:13:30 aglitke: yes. I also mentioned, combine GET, DELETE, CTEATE method, UI can edit template. But I have discusses with hlwanghl_, he need just one method. 13:14:00 shaohef: ok. Your patches may need to be resubmitted. 13:14:11 Hopefully we can support update in a generic fashion. 13:14:29 lvroyce: THat is not as important. 13:14:30 7/30 we mainly focus on rename to kimchi, right? 13:14:32 shaohef: we need add the first update method. 13:14:34 shaohef Combine? 13:14:37 Additional enhancements I suggest we add to sprint 1: 13:14:38 Create template by scanning for ISOs 13:14:38 Create template by selecting distro/versoin then stream ISO in the backend 13:14:42 I would rather see the template scanning via tasks done in the design ui 13:14:52 about the iso scanning feature... When I making up the dev-ui, no /iso exported really bothers, I think 13:14:59 aglitke: ok. need more  discussion 13:15:26 AdamKingIT: One thing I don't understand from the design UI is how we can organize newly imported templates. 13:15:35 people need to input string of server path 13:15:48 In the dev UI we had folders so the new templates could be placed in a folder called 'Import' 13:16:07 and the user could then move them around or delete the ones that were not needed,. 13:16:20 With the design UI they all appear in a flat namespace. 13:16:29 aglitke: templates can be imported from server side? 13:16:36 They are, thouhg new ones are to be marked new: pending 13:17:07 After that. there are a number of search/filter mechanisms to be implemented 13:17:13 ok so we need some enhancements to support tagging in the backend and UI 13:17:31 :)I like that 13:17:52 While I like the idea of tagging, we may be ok w/ filers based on distro/version 13:18:05 lvroyce: regarding /isos... I think reserving a storage domain for isos only is a mistake. 13:18:07 I think tags would be easy enough to add on top 13:18:46 aglitke, I would argue that dispersing ISO images around is also a mistake. 13:18:58 sming: why? 13:19:19 aglitke, there is no benifit to disperse the images to different directories. 13:19:28 do we still want a user story for upload an iso? 13:19:29 sming: agree 13:19:30 But that is where they happen to be. 13:19:42 aglitke, but input long string path is error prone, if only we had way to tell users what iso we've got and make them choose 13:19:49 Kimchi should not require users to relocate their files just so they can work with this software,. 13:20:04 lvroyce: scan uses a path '/' 13:20:10 or something else. 13:20:21 We can default it to '/' 13:20:29 Uploading ISO files is slow through HTTP 13:20:40 yes 13:20:46 hlwanghl_: We are just talking about scanning the local server for what may be on the box. 13:20:56 aglitke, but we can require the user to upload the iso images to a specific Kimchi directory if they want to use Kimchi. 13:21:01 Rather than requiring a user to add specific isos 13:21:10 sming: Why? 13:21:21 It's just as easy to not have that restriction. 13:21:21 aglitke: reading 13:21:33 aglitke, after they uninstall Kimchi, the iso images also removed. 13:21:40 no! 13:21:50 hlwanghl_: we can install the VM from remote ISO by steaming. aglitke right? 13:21:57 sming: the iso images are user data 13:22:04 you should absolutely not remove them :) 13:22:31 shaohef: / hlwanghl_: eventually that is another feature we are shooting for. 13:22:35 Hopefully by 1.0 13:22:51 #action aglitke add an enhancement for web templates 13:23:02 Local first and then remote 13:23:08 yes 13:23:15 aglitke, uploading the iso images to a specific Kimvhi directory will clone the iso images and those images belong to Kimchi. 13:23:22 remote requires very new versions of qemu and libvirt 13:23:35 sming: cloning wastes space 13:23:51 since they are readonly there is no reason to clone. 13:23:55 Can we detect if qemu/libvirt is at a new enough level? 13:24:14 sming: Have you seen the iso repository we have on ltcphx? It's 10s of GBs of ISOS 13:24:25 that would take forever to clone to each individual server. 13:24:35 AdamKingIT: yes. 13:24:39 We don't want to offer the user a UI choice only to quickly fail to stream 13:24:44 AdamKingIT: sure. libvirt support a API 13:24:54 I believe libvirt capabilities will provide that info to us. 13:25:24 Great. We'll want to think about a design to query the capability, along w. others... 13:25:29 AdamKingIT: right. We will disable the web templates feature when the system is too old to support it. 13:25:46 should dovetail nicely into an authorization model... 13:26:21 Eventually some users will be authorized to actions, and some not. Noone will be authorized if the server can't support the action 13:26:21 aglitke: a new URL for query the capability? 13:26:29 ok. I don;t really have anything else for Planning right now. We are going to add some enhancements and work to assign them to the proper sprints. 13:26:42 shaohef: maybe. 13:26:55 Lets start a discussion thread. A new URL may be clumsy 13:26:57 probably. 13:27:38 #topic Roundtable / Open Discussion 13:28:39 Any takers? 13:28:48 #info alinefm Closed bug https://github.com/kimchi-project/kimchi/issues/21 (get open VNC port in mockmodel) 13:28:51 David Wang in CDL is arranging a customer visit to get some Kimchi feedback. I have asked him to take Ming and Xin w/ him... 13:29:06 #info alinefm Investigated bug https://github.com/kimchi-project/kimchi/issues/6. The problem is with Firefox 10.0.2 ESR. We should list the browsers will be supported by kimchi. 13:29:17 #info alinefm I am unable to reproduce https://github.com/kimchi-project/kimchi/issues/20. Need more details in how do that. 13:29:22 Info I hope to get back is documented on the wiki 13:29:27 #info shaohef test kimchi on FC19. shaohef work with zhoumeina to post her new patch. shaohef closed bug 16 and 18 13:29:34 #info alinefm Still need to apply review in the patches to disable 'create' button in create vm window and resend it to review. 13:29:54 #info alinefm Now I am working on https://github.com/kimchi-project/kimchi/issues/4. I was able to reproduce the bug and the problem seems to be with screenshots. 13:30:09 #info hlwanghl worked on vm name issue 13:30:26 #info BingBu Sent 4 patches: 13:30:27 #info BingBu 1.Disable the VNC button and give warning message to prompt user 13:30:27 #info BingBu 2.Raise an exception in skipUnless() to abort the test on Python2.6 13:30:27 #info BingBu 3.Bug Fix: Get correct exception info when run unittest on Python2.6 13:30:27 #info BingBu 4.Add another path for SLES11 to find the qemu binary 13:30:52 hlwanghl_: Did you originally report https://github.com/kimchi-project/kimchi/issues/20 13:31:42 #info lvroyce fixed some bugs about testcase, template, unicode, are doing the dev-ui for template, screenshot flicker patch merged, did some review 13:31:51 #info aglitke Reviewed and applied lots of patches. 13:32:23 #info aglitke is hoping to implement an autotools based build and the iso scanning flow 13:32:31 aglitke: shaohef and I found the issue 13:32:42 aglitke: yes. 13:32:51 hlwanghl_: ok, thanks. Can you please provide some more information on how to reproduce it. 13:32:51 But still not resolved 13:33:20 Is it the first vnc opened when kimchi starts? Does it need a specific browser? 13:33:20 hlwanghl_, shaohef, I am not able to reproduce it 13:33:22 etc. 13:33:23 #info sming rebased async task to v9 and debugreports to V7 13:33:24 #info sming verified the RPM packages in RHEL 6.4 . 13:33:25 Ok I'll try. Not happens every time 13:33:39 alinefm: I see the problem sometimes 13:33:49 #info zhoumeina is hoping to discuss the design UI together, so that everyone can get a good understand of it. and make the UI design better 13:34:00 Usually if I start a VM and then try to vnc to it quickly. 13:34:25 aglitke, ok 13:34:26 aglitke alinefm still don't know the exact steps 13:34:30 zhoumeina: This is a great place to do that. Do you have some specific areas to discuss> 13:34:48 aglitke, do you remember which distro? 13:35:50 I see it on fedora 13:36:11 alinefm: FC17 and firefox 13.0 can reproduce. iterate open noVNC and close, 13:36:12 aglitke: Some better tools? I think netmeeting is OK.. 13:36:47 shaohef, aglitke, ok! I will try again on Fedora 13:37:01 zhoumeina: no netmeeting for linux. I can meet you tomorrow and you can show me 13:37:03 We have found that using skype or netmeeting is not very effective with a global team. We have network latency and language barriers. 13:37:11 I see 13:37:49 We can try a google hangout sometime for a small sub-group though. 13:38:04 sorry to interrupt... repost: aglitke, but input long string path is error prone, if only we had way to tell users what iso we've got and make them choose 13:38:05 we may also want to add template disable when it can't create vm because of the iso disapeared... 13:38:20 But just seeing the person doesn't really help to discuss UI look and feel. 13:38:44 lvroyce_, good point. 13:38:52 lvroyce_: Yes, we are going to add scanning. 13:39:03 All found isos will be added as templates. 13:39:11 We will tag them as 'new' 13:39:26 no need to see the person, but share the screen, and can operate by everybody 13:39:28 so the UI can enable a 'new' filter after the scan completes 13:39:32 but we still need a path input or use a default one? 13:39:41 It would be interesting to add a 'files' selection path, where they remote user can browse, within some scope 13:39:42 lvroyce_: default is '/'. 13:39:49 If someone knows a better path, then can use it. 13:40:16 we can avoid some common bad places to scan (ie. /proc, /sys) 13:40:21 aglitke: how should we handle the spice display. you know kimchi can manager VMs created by other VM manger tool. https://github.com/kimchi-project/kimchi/issues/35 13:40:25 scanning / may take longer than we would prefer depending on NFS mounts and such 13:41:01 shaohef: We don't support spice. If a vnc port is not found, then the 'vnc_port' field in the VM resource should be blank. 13:41:11 In that case, the UI should disable the VNC button. 13:41:15 I hope default can be a place like /usr/share/kimchi/iso as we don't want to expose our server directory structure 13:41:35 aglitke: VNC button change to gray? 13:41:42 shaohef: yes 13:41:44 It would be useful if we revist spice 13:41:59 aglitke: got it. 13:42:06 AdamKingIT: Until spice supports an HTML5 client we cannot support it. 13:42:18 supposedly they are working on it. 13:42:34 aglitke: I think if spice, we can tell user the port and give some hint how to show the display,such as using 'spicec' 13:42:34 AdmaKingIT, some times the nfs mounted path in the Kimchi server is pretty big, I think we should avoid scan that directory. 13:42:41 One of the design tenets of kimchi is that it requires nothing of the client except for a browser. 13:42:46 AdamKingIT: maybe spice need a plugin. java plugin. 13:43:04 sming: THat's where the isos are though sometimes :) 13:43:21 We can have a 'quick scan' and a deep scan 13:43:32 maybe quick scan looks only on local fs 13:43:44 If the client has spice capability, it would be useful to give the user a quick path to launch it 13:43:46 and deep scan does '/' and searches everything 13:43:47 aglitke, also nfs scanning may block the scanning thread and block the whole process. That is the lesson from VDSM. 13:44:30 sming: good point. 13:44:36 AdamKingIT: we can do. but we like HTML5 spice client. aglitke, right? 13:44:51 shaohef: yes. 13:45:07 aglitke: I don't know if we decided that "spcie" vm can not be show in the list of VMS? 13:45:13 I think there are much higher priority items than supporting legacy spice clients at the moment. 13:45:42 IMO we don't have to constrain ourselves to a browser only soln, other applications can be launched from the browser if avail 13:45:51 and if we make kimchi great, the spice guys will have another reason to accelerate their html5 client work. 13:46:07 AdamKingIT: I strongly disagree. 13:46:15 That is why virt-manager currently sucks. 13:46:23 oVirt spice connectivity sucks 13:46:51 aglitke: I mean the issue "VNC port" can not be found 13:46:53 because of the need to fetch a plugin and install it. 13:46:54 aglitke: which characters should be allowed in VM name? https://github.com/kimchi-project/kimchi/issues/36 13:47:02 I also agree that we have higher priority items than adding ANOTHER remote display technology 13:47:28 shaohef: We should strive to support the same set of chars as libvirt does. 13:47:33 now kimchi can show the spice vms on the list of vms, but can not VNC 13:47:46 shaohef: Can you try to use virt-manager to create a vm with a chinese name? 13:47:56 zhoumeina: Yes, and that's ok 13:48:10 aglitke: not try. I will try it later. 13:48:34 #action shaohef use virt-manager to try and create a vm with a chinese name and report results to list 13:48:40 thanks 13:55:10 And IE 13:55:13 alinefm: we can make a update package and a tool to clean the catch,after update 13:55:44 zhoumeina: we can discuss that tomorrow 13:55:46 yikes, no. You can;t ask people to clean cache because we updated something 13:56:02 aglitke: kimchi-bot become kimchi-bot` 13:56:13 ok 13:56:18 Bot quit 13:56:31 shaohef: Yes. Strong storms here caused a brief power outage. 13:56:33 there is a pretty well established pattern to handle. I'll start the discussion on the list 13:56:41 The bot and I missed part of this conversation :( 13:57:17 I think we should to check whether cherrypy supports standard http cache config for static contents or not 13:57:21 aglitke1, which part did you read? 13:57:25 #action AdamKingIT start the discussion on the list about how to handle UI code updates wrt browser caches 13:57:33 shaohef: you can send the conversation to aglitke 13:57:42 I will paste the message before and after the outage: 13:57:44 aglitke1: sming: can we setup a irc proxy? 13:57:58 (09:48:39 AM) aglitke: thanks 13:58:00 aglitke1: yes. I will 13:58:07 (09:54:33 AM) alinefm: sming, in firefox is Ctrl+shift+Del 13:58:34 about 6 minutes 13:58:41 there is always a UI for clearing the cache 13:59:19 In firefox in under Tools -> options => advanced , Network 13:59:48 IE has similar 13:59:50 aglitke1, sent to you in pvt 14:00:02 as well as a path in dev tools - F12 14:00:16 AdamKingIT, I used that UI before to clean all the caches. 14:00:27 aglitke1: also send to you. 14:00:37 thanks for catching me up guys 14:00:55 There used to be a FF plugin called Johnny Cache that I loved, but it stopped being maintained 14:01:01 Well we are at the top of the hour and I want to respect everyone's time. 14:01:21 Feel free to continue discussions here 14:01:25 or on the list. 14:01:31 Thanks for joining everyone 14:01:34 #endmeeting 14:01:40 good luck w/ the storm 14:01:44 sming: will we setup a irc proxy? 14:01:47 thanks. 14:02:05 These old power lines aren't as resilient as the underground lines we had in rochester. 14:02:05 aglitke1, do you have a storm now in US? 14:02:12 irc proxy for China guys? 14:02:14 hlwanghl_: 14:02:25 lvroyce_: Yes, heavy thunderstorms and rain. Flash Flooding 14:02:28 we can discuss it now 14:02:28 irc for all? 14:02:47 aglitke1: take care. 14:02:59 aglitke1, that's scary... take care 14:03:39 Good luck, aglitke1 14:03:44 hlwanghl_: 14:04:26 you have better ideas?