18:02:36 <renuka> #startmeeting
18:02:37 <openstack> Meeting started Thu Dec  1 18:02:36 2011 UTC.  The chair is renuka. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:02:38 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:02:49 <renuka> #topic openstack volumes
18:03:18 <renuka> I mainly had to raise one point,about the usage of the host flag
18:03:40 <renuka> so before that, do either of you have any updates or would like to discuss anything today?
18:04:12 <jdg_> Nothing really from me.  I've been trying to set up a cloud in house the past week or so and that's about it.
18:04:13 <vladimir3p> nope, I have no updates. stuck with integrating our stuff with keystone
18:05:14 <renuka> vladimir3p: so this week or so, you mentioned you can start on the scheduler, is that still on track?
18:05:43 <vladimir3p> hope to start it 1st week of Dec (next week)
18:06:07 <vladimir3p> but only if I'll succeed to bring up our keystone :-)
18:06:52 <renuka> right, keep us in the loop :)
18:07:04 <renuka> #topic host flag
18:07:26 <renuka> so I just wanted to ensure that everyone is aware of this...
18:07:44 <renuka> so far we have been using the host flag in the code as if it is necessarily an IP address
18:08:05 <renuka> recently with some upgrade work, it was proposed that it should be a GUID for a machine
18:08:18 <renuka> that began to break some iscsi tests
18:08:28 <renuka> i believe we have a fix here: https://review.openstack.org/#change,1982
18:08:39 <renuka> #link https://review.openstack.org/#change,1982
18:09:17 <vladimir3p> let me take a look, I'm not sure if we ever experienced such issue...
18:09:46 <renuka> so please keep in mind that the host flag, and therefore volume['host'] and iscsi_target['host'] may not be IP addresses and could well be some form of IDs
18:10:16 <renuka> you may not have experienced it yet, because John from citrix is doing the upgrades changes... so they may not be upstream
18:10:17 <vladimir3p> hmm... volume['host'] is always a hostname, isn't it?
18:10:28 <renuka> well it has been, so far
18:10:32 <jdg_> So forgive me for the ignorant question, but where is the coorelation made back to an IP?  Assuming some link back to the database?
18:10:38 <renuka> I am just pointing out that in the future it need not be
18:11:31 <renuka> in that patch on gerrit, we are populating "provider location" with the details of where to find the iscsi target
18:11:46 <jdg_> Yes, sorry... I just read it in the change link you sent.
18:12:13 <renuka> right, so the definition of "host" has been clarified now
18:12:35 <vladimir3p> ok, I see...
18:13:01 <renuka> since I would most certainly have read the code assuming volume['host'] is an IP address, I wanted to make it a bit more explicit that it need not be
18:13:11 <vladimir3p> this is irrelevant for us, because we never really used it. The IP is filled up from host-specific SW we install on every host and from there we create provider_location
18:13:49 <vladimir3p> renuka, I still suppose that volume['host'] is just a hostname ...
18:14:32 <renuka> last conversation I had about this change, it was decided that volume['host'] should continue to be whatever is in the host flag
18:14:45 <vladimir3p> ok
18:14:49 <renuka> in fact it has always been that... i dont think this change changes it
18:14:56 <renuka> let me check
18:15:52 <renuka> right, it doesn't
18:16:49 <renuka> so whenever someone starts the services with the host flag set to something else that is not an IP address, we run the risk of misusing volume['host'] if we are not careful
18:17:32 <vladimir3p> unless this is set to default value, which is hostname
18:17:37 <vladimir3p> and supposed to be unique
18:18:18 <renuka> it will be unique in either case...we just don't have a guarantee that it can be used as an IP address
18:18:49 <renuka> for example, with the recent changes, the iscsiadm commands were failing, because the code does a "-p volume['host']"
18:19:00 <renuka> that assumes it is an IP address
18:19:08 <renuka> so any such lines of code, we need to watch out for
18:20:14 <vladimir3p> renuka, just to understand ... if you override FLAGS.host with an IP address it means that on all levels (schedulers, API, etc) you are using IP and not a hostname
18:20:24 <vladimir3p> which is fine, but it is on all levels
18:20:34 <renuka> yes
18:20:39 <vladimir3p> FLAGS.host is kind of "global" param for Manager
18:20:46 <renuka> yes
18:20:49 <vladimir3p> it means that all compute services and others are using it
18:20:49 <vladimir3p> ok
18:21:37 <renuka> this caveat does apply to any other coding that you are doing also
18:21:59 <vladimir3p> hmm... not really
18:22:10 <vladimir3p> we don't care what exactly is set in host field
18:22:18 <jdg_> cd vsa
18:22:18 <vladimir3p> it just should be unique
18:22:18 <jdg_> ls
18:22:27 <jdg_> sorry
18:22:30 <vladimir3p> :-)
18:22:33 <renuka> #info FLAGS.host and therefore volume['host'] need not always be an IP address
18:22:45 <vladimir3p> yeah :-)
18:22:45 <jdg_> :)
18:23:38 <vladimir3p> ok, do we have anything else for today?
18:23:41 <renuka> right, i didn't really have anything more for the meeting
18:24:05 <vladimir3p> unless you could tell me where I could find normal & stable keystone for Natty
18:24:06 <jdg_> Nothing here really, just trying to actually stand up instances and connect to iscsi storage.
18:24:19 <vladimir3p> (preferably in form of packages)
18:24:22 <jdg_> I'll be finishing my blue-print and submitting properly next week.
18:24:47 <jdg_> I've been using Kiall's script (which uses packages)
18:25:02 <renuka> vladimir3p: I haven't installed from packages before... have you looked at devstack? not sure if they use keystone
18:25:13 <renuka> aah ok
18:25:22 <jdg_> Trouble is I'm trying to do physical machines (multi-node)
18:25:45 <jdg_> Kialls script gives keystone, glance, dashboard etc.
18:26:11 <jdg_> Now I'm hosed though because none of the euca2ools stuff is configured.... anyway, that's going to be the rest of my week.
18:26:21 <renuka> #keystone packages
18:26:37 <renuka> this, for oniric http://packages.ubuntu.com/oneiric/keystone does not help?
18:27:01 <vladimir3p> yes, I've seen this one for oneiric, but we are using Natty
18:27:07 <vladimir3p> not sure if it will work
18:27:45 <vladimir3p> also, I was trying to use the latest stable/diablo branch from devstack, but it is behaving weird... the one from 1-2 weeks ago worked perfectly
18:27:49 <renuka> maybe we should keep this out of the meeting and on the #openstack chat room where more people can help
18:28:05 <vladimir3p> yeah, let's finish the volume one
18:28:11 <renuka> #endmeeting