08:02:28 <huzhj> #startmeeting daisycloud
08:02:29 <openstack> Meeting started Fri Dec  9 08:02:28 2016 UTC and is due to finish in 60 minutes.  The chair is huzhj. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:02:30 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:02:33 <openstack> The meeting name has been set to 'daisycloud'
08:02:42 <huzhj> #topic Roll Call
08:02:49 <luyao2> #info luyao
08:02:51 <huzhj> #info Zhijiang Hu
08:03:02 <zeyu> #info zeyu
08:03:06 <huzhj> Agenda
08:03:17 <huzhj> 2) Shanghai Telecom Support (Storage & Bare Metal)
08:03:17 <huzhj> 3) OPNFV: Daisy CI Progress
08:03:17 <huzhj> 4) AoB
08:03:19 <zhouya> #info zhouya
08:03:48 <huzhj> #topic Shanghai Telecom Support (Storage & Bare Metal)
08:05:13 <huzhj> #info The setup plan is clear now
08:06:01 <huzhj> #info single controller/network/storage node, single compute node. compute node can be added future.
08:06:56 <huzhj> The Disk array is AMS2300 or HP 2012
08:07:08 <zhouya> so we should use lvm?
08:07:23 <huzhj> #info Disk Array not supported by any Cinder driver, so LVM will be used.
08:08:13 <huzhj> After installing the OS, we need to pause, then setup cinder-volumes VG and go on deploy OpenStack
08:09:03 <luyao2> can we add in daisy code？
08:09:34 <huzhj> I prefer install OS for a host and connect it to the disk array first, just try to test the lvm to see if it is functional. then destroy the host which will be re-deployed by daisy.
08:09:58 <huzhj> luyao2: that is a Good idea.
08:10:27 <huzhj> We need to do some temp modification to the daisy code.
08:11:29 <huzhj> #info Keep a record of the  temp modification to the daisy code. basically , lvm setup and kolla confg file modification
08:12:29 <zhouya> These code need to add into disk_array.py ?
08:12:51 <huzhj> zhouya: your https://review.openstack.org/#/c/406036/ can be a temp modification
08:13:04 <huzhj> zhouya: no
08:13:15 <huzhj> we do not touch disk_array.py
08:13:21 <zhouya> OK
08:13:31 <zhouya> we just add this into kolla/config.py
08:14:21 <huzhj> #info About how to setup LVM: #link http://docs.openstack.org/developer/kolla/cinder-guide.html#create-a-volume-group
08:15:13 <huzhj> #info about the network, each target node use only 2 port is enough. One for neutron interface, the other for anything else.
08:15:57 <huzhj> Since we use VxLAN instead of VLAN, there is no need to config VLAN on the switch at all.
08:16:18 <huzhj> #info Since we use VxLAN instead of VLAN, there is no need to config VLAN on the switch at all.
08:17:58 <zhouya> OK
08:18:20 <huzhj> #info Reason why only one controller: 1) Cinder in LVM mode has bug which restrict us to use on controller. 2) No need to share Glance image storage over more than one controller.
08:18:59 <huzhj> #info for reason 1), you can refer to http://docs.openstack.org/developer/kolla/cinder-guide.html
08:20:21 <huzhj> For reason 2), because Glance images are stored on controller. So say we have 2 controllers A and B. If one upload an image to A, no method now to sync it to B, when A is down and B take over, the image will disappear.
08:21:51 <huzhj> The solution is 1) using swift as the object storage shared by all controllers 2) use NFS to share /var/lib/glance/images directory over all controller. 3) uses only one controller....
08:21:58 <huzhj> We choose 3) :)
08:22:44 <huzhj> #info The Blade system may be HP C7000
08:23:15 <huzhj> C7000 is look the same as E9000
08:23:49 <huzhj> We need to grab some manual for blade setup as a prepareation
08:23:52 <zhouya> So we should test C7000>
08:23:54 <zhouya> ?
08:24:32 <huzhj> zhouya: test C7000 what? We do not have C7000 in hand
08:24:57 <zhouya> we only have E9000?
08:25:04 <huzhj> Yes
08:25:52 <zhouya> Are they operation method the same?
08:26:17 <huzhj> I am getting familiar  with C7000/E9000, but need more time.
08:26:23 <zhouya> Or we should reference the manual book.
08:26:34 <huzhj> zhouya: No, not thesame in operating
08:27:15 <zhouya> It is really a big chellange.
08:27:30 <huzhj> May be some guys in SH Tel can give us a hand on how to setup, but we can not depend on that.
08:28:50 <zhouya> We should read some manual book befort they can give us help.
08:28:57 <zhouya> before
08:29:50 <zhouya> Is there any link we can reference?
08:30:41 <huzhj> #zhouya here is a pdf describing the hardware, you can take a look
08:30:44 <huzhj> #link https://www.hpe.com/h20195/v2/GetDocument.aspx?docname=4AA1-4286ENW&doctype=product%20family%20guide&doclang=EN_US&searchquery=&cc=us&lc=en
08:31:30 <huzhj> basically some blade servers and 1-2 switches  in a enclosure.
08:32:15 <huzhj> and manually can be found at #link http://h20566.www2.hpe.com/portal/site/hpsc/public/psi/home/?sp4ts.oid=1844065&ac.admitted=1481272145239.125225703.1938120508#manuals
08:32:39 <zhouya> It is a great help.
08:32:57 <huzhj> OK this is all for me on this topic, anything else?
08:33:50 <zeyu> no
08:33:57 <zhouya> No,maybe we should take a little time to  read these reference.
08:34:22 <huzhj> #topic OPNFV: Daisy CI Progress
08:34:57 <huzhj> luyao2: any status update on this topic?
08:36:16 <huzhj> Or zhouya ?
08:36:40 <luyao2> new daisy code have error when build pxe for os
08:36:54 <zhouya> We should resize the virtual mashine.
08:37:08 <zhouya> I add them to the deploy.yml file
08:37:43 <zhouya> and we should use yaml package analysis the fdate.
08:37:46 <luyao2> because now suppout diff node install diff os，and sub file in linuxlistall file file。
08:38:59 <huzhj> luyao: can you fire a bp/bug on launchpad so that we can take a look into this problem later offline?
08:39:05 <huzhj> luyao2: can you fire a bp/bug on launchpad so that we can take a look into this problem later offline?
08:39:21 <luyao2> ok
08:39:55 <huzhj> luyao2: can we fallback to use the old pxe_server_install.rpm?
08:40:15 <luyao2> yes
08:40:36 <luyao2> the script midify of build pxe for os
08:40:41 <huzhj> If ther is no time for us to fix this bug before go to SH Tel, we will fallback to the old one,
08:41:51 <luyao2> may be easy。just a line code
08:42:25 <huzhj> #info new code in pxe_server_install.rpm cause problem when installing OS, @luyao2 is working on that
08:43:09 <huzhj> zhouya: about the resize prlbolem
08:43:20 <huzhj> So you have added the size def in the yml?
08:43:39 <zhouya> yes
08:44:10 <huzhj> So the next step is impl. the parser code right?
08:44:18 <zhouya> and i can use size=`python filename.py` command to pass the parameter to yangyang's script
08:44:35 <huzhj> Good
08:44:41 <luyao2> good
08:44:49 <zhouya> Yes I am working on the parser code.
08:45:03 <huzhj> Great
08:45:36 <zhouya> Also,we move daisy password and ip into deploy.yml
08:45:51 <huzhj> Hope after finishing this work we can finally meet the virtual POD requirements
08:45:52 <zhouya> we can use this function to get them too.
08:46:02 <huzhj> #info we move daisy password and ip into deploy.yml
08:46:15 <huzhj> #info we move disk size into deploy.yml
08:47:42 <huzhj> Another thing , please take a look at the common config file requirement. see if we can follow it and is there any part of it not fit for us.
08:48:28 <huzhj> Anything else for this topic?
08:48:36 <zhouya> OK,I have receive that email,And will follow it.
08:48:38 <zhouya> NO
08:49:04 <huzhj> #topic AoB
08:50:23 <zhouya> May be escalator version query progress can be hang up.
08:50:58 <zhouya> For I should take some time to test deploy openstack via daisy.
08:51:35 <zhouya> #link https://gerrit.opnfv.org/gerrit/#/c/25571/ this commit also has some flake8 to deal with.
08:51:37 <huzhj> OK.
08:55:10 <huzhj> #info #link https://etherpad.opnfv.org/p/d6fCuBdOrH
08:55:18 <huzhj> this is for SH Tel prepareation
08:59:51 <huzhj> #endmeeting