21:00:54 <danwent> #startmeeting quantum
21:00:55 <openstack> Meeting started Mon Mar 18 21:00:54 2013 UTC.  The chair is danwent. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:56 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:58 <openstack> The meeting name has been set to 'quantum'
21:01:11 <danwent> yeah, salvatore is out as well
21:01:17 <garyk> gongysh: hi
21:01:22 <arosen> hiya
21:01:29 <danwent> #info agenda: https://wiki.openstack.org/wiki/Network/Meetings
21:01:31 <gongysh> garyk: helo
21:01:37 <nati_ueno> Hi!!
21:01:47 <danwent> #topic announcements
21:01:58 <danwent> #info master is now open for havana commits.
21:02:13 <danwent> I was OOO, and have not yet removed all -2s from grizzly freeze.  Will do that after meeting
21:02:27 <danwent> #todo #danwent remove all -2s from grizzly freeze
21:03:11 <danwent> #info we expect a Grizzly RC2, but as always, we want to keep the set of changes very limited to reduce the likelihood of regressions
21:03:22 <danwent> any other announcements?
21:03:43 <danwent> #topic Grizzly RC2 bugs
21:03:46 <danwent> markmcclain: ?
21:04:21 <markmcclain> like dan said, we'll be pretty strict with the changes we're cherry-picking
21:04:35 <markmcclain> have two changes that are up for review now
21:04:47 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1154895
21:04:48 <uvirtbot> Launchpad bug 1154895 in quantum "Quantum not logging anymore on stderr" [High,Fix committed]
21:05:02 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1153983
21:05:03 <uvirtbot> Launchpad bug 1153983 in heat "Consolidate SQLAlchemy requirements to 0.7.x" [Medium,Fix committed]
21:05:46 <markmcclain> There are 4 more that might make good candidates for back porting to RC2:
21:05:52 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1155121
21:05:54 <uvirtbot> Launchpad bug 1155121 in quantum "DB tables not loaded with Cisco Plugin" [Undecided,In progress]
21:06:16 <markmcclain> This one has a review here: https://review.openstack.org/24653
21:07:13 <markmcclain> there is active discussion on the review
21:07:16 <danwent> yeah, seems like cisco plugin is broken without that change, so seems like an obvious candidate
21:07:32 <danwent> but looks like this fix may still need some discussion
21:07:44 <markmcclain> yeah.. I can follow it, but we'll need another core
21:07:53 <danwent> i'll be on it as well
21:07:58 <danwent> but others are welcomed :)
21:08:05 <markmcclain> cool
21:08:24 <markmcclain> next bug:
21:08:25 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1155748
21:08:26 <uvirtbot> Launchpad bug 1155748 in quantum "SubnetRoute changes don't propagete to dhcp agent" [High,In progress]
21:08:47 <markmcclain> the review is here: https://review.openstack.org/24569
21:09:03 <markmcclain> arosen is currently working on revision
21:09:09 <markmcclain> arosen: any thing else to add?
21:09:20 <arosen> markmcclain: nope still working on it.
21:09:44 <danwent> arosen: any fundemental blockers?
21:09:54 <danwent> or do you just need to rework hte unit tests?
21:10:04 <arosen> nope just haven't had the cycles yet, i'll have it up by the end of the day.
21:10:04 <danwent> (i.e., why is it in WIP?)
21:10:07 <arosen> or out of WIP
21:10:12 <arosen> I put it in WIP due to the tests.
21:10:13 <danwent> k
21:10:39 <gongysh> arosen: ping me if u need my review.
21:10:48 <arosen> gongysh: k will do
21:11:12 <markmcclain> next bug
21:11:13 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1155063
21:11:14 <uvirtbot> Launchpad bug 1155063 in quantum "dhcp-agent uses an interface which is down" [Undecided,New]
21:11:41 <markmcclain> this one is interesting because the report is from the Folsom version
21:11:51 <markmcclain> but the behavior has not changed in Grizzly
21:12:16 <danwent> yeah, it seems like it must be something to do with the env. or incorrect setup
21:12:23 <garyk> markmcclain: the problem is that in ubuntu the ovs-cleanup utility is not used.
21:12:24 <danwent> otherwise we'd be noticing it more frequently :)
21:12:35 <garyk> we need to document this and it will solve a number of problems
21:13:00 <markmcclain> garyk: can you create a doc bug?
21:13:09 <garyk> markmcclain: sure.
21:13:12 <danwent> garyk: interesting… is it that you think there was a reboot, or something else?
21:13:25 <gongysh> garyk:  has the bug said it is after rebooting?
21:13:25 <danwent> garyk: i know you've helped troubleshoot several related issues.
21:14:01 <garyk> the problem is that when the host reboots the ovs creates the device and it is down. the l3 agent and dhcp do not deal with this well
21:14:18 <garyk> the cleanup purges the ovs of the interfaces and all works like a charm afterwards
21:14:32 <markmcclain> the bug was not specific about the state of the system when the bug occurred
21:14:34 <garyk> we just need to make sure it is invoked in ubuntu prior to all other quantum services
21:14:45 <garyk> we have done this in fedora and rhel
21:14:47 <danwent> is there a harm to also having IP lib set it to up?
21:14:49 <gongysh> we will have to ask the bug submitter to provide more description.
21:15:05 <garyk> i'll ask for a clarification
21:15:22 <danwent> ok, garyk, are you ok if we assign the bug to you for triage?
21:15:32 <danwent> since you are most familiar with related issues?
21:15:34 <garyk> danwent: sure. i'll take it
21:16:01 <danwent> thx
21:16:14 <markmcclain> #todo #garyk to follow up on bug 1155063
21:16:14 <uvirtbot> Launchpad bug 1155063 in quantum "dhcp-agent uses an interface which is down" [Undecided,New] https://launchpad.net/bugs/1155063
21:16:41 <garyk> #todo #garyk to document
21:17:03 <markmcclain> last one…
21:17:03 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1156504
21:17:05 <uvirtbot> Launchpad bug 1156504 in quantum "Plugin doesn't get called when explicit rules are auto-created." [Undecided,New]
21:18:01 <danwent> arosen should comment in more detail, but it seems like some plugins were automatically creating the default rules, while others were expecting to get explicit create rule calls
21:18:14 <danwent> nvp plugin works, as it auto-creates the rules.
21:18:35 <arosen> Yup right now we are just adding the rules in the db base class but we should probably call add rule in the plugin to do this.
21:18:38 <danwent> i'm not sure about iptables based plugins
21:18:44 <danwent> sounds like midokura plugin does not
21:19:23 <arosen> I think we should push a patch that fixes the midokura plugin and then refactor this code for havana.
21:19:24 <danwent> arosen… is this a problem for any plugin using iptables based security groups (e.g., ovs + linux bridge)?
21:19:34 <danwent> ok, that's what I was trying to figure out.
21:19:46 <arosen> danwent:  i'll have to check, i'll reply to the bug report if so.
21:19:52 <tomoe_> danwent: i'm working on a patch now
21:19:55 <danwent> if midokura plugin is only one broken right now, for grizzly i'd rather just target a fix to that plugin.
21:19:59 <danwent> tomoe_: ok, great
21:20:09 <tomoe_> midonet plugin doesn't rely on iptables
21:20:23 <danwent> tomoe_: yeah, i was referring to other plugins like OVS + linux bridge
21:20:34 <danwent> trying to figure out if those plugins were currently broken in the same situation
21:20:39 <danwent> (it sounds like they are not)
21:20:49 <danwent> ok, so sounds like we're clear on that bug?
21:20:56 <markmcclain> think so
21:21:04 <tomoe_> i thought those would be ok as they are using db
21:21:24 <danwent> tomoe_: yup, i was just confirming that
21:21:43 <danwent> markmcclain, gongysh, do we think https://bugs.launchpad.net/python-quantumclient/+bug/1152427 needs to be in 2.2.1 for the client?
21:21:46 <uvirtbot> Launchpad bug 1152427 in python-quantumclient "Support refresh token if token is passed to the constructor" [Undecided,In progress]
21:21:57 <markmcclain> I added to the client section :)
21:22:00 <markmcclain> yes
21:22:04 <danwent> markmcclain: ah, sorry
21:22:06 <danwent> will wait :)
21:22:23 <markmcclain> had one late bug:
21:22:24 <markmcclain> https://bugs.launchpad.net/quantum/+bug/1156782
21:22:25 <uvirtbot> Launchpad bug 1156782 in quantum "LoadBalancerCallbacks.get_ready_devices() returns incorrect set of devices" [Undecided,New]
21:22:38 <danwent> markmcclain: there was also the policy.json typo bug, let me look at that.
21:22:46 <danwent> markmcclain: yeah, that just came in, but seems worthy
21:23:15 <danwent> https://bugs.launchpad.net/bugs/1155379
21:23:16 <uvirtbot> Launchpad bug 1155379 in quantum "Typo in policy.json" [Medium,In progress]
21:23:27 <markmcclain> yeah.. that should have been in my list
21:23:48 <danwent> salv-orlando says this isn't technically a security concern, but since it appears to be without some explanation, probably best to fix it.
21:24:45 <danwent> ok, anything else on current RC2 targeted bugs?
21:24:57 <danwent> otherwise, I want to spend a big chunk of the meeting on docs
21:25:07 <danwent> as I think we're in rougher shape there than we are on bugs :)
21:25:29 <markmcclain> danwent: that's it for me
21:26:02 <danwent> #topic grizzly documentation
21:26:28 <danwent> #info grizzly doc impact tracker: https://wiki.openstack.org/wiki/Network/GrizzlyDocImpact
21:26:42 <danwent> please pull up that link, as I want to cover it pretty methodically in the meeting
21:26:56 <danwent> #info current admin-doc bugs: https://bugs.launchpad.net/openstack-manuals/+bugs?field.tag=quantum
21:27:09 <danwent> #info current api-doc bugs: https://bugs.launchpad.net/openstack-api-site/+bugs?field.tag=netconn-api
21:27:37 <danwent> unfortunately, salv-orlando it out today, so we won't be able to cover all of the API stuff.  I will try to sync with him one-on-one tomorrow to identify key gaps
21:27:45 <danwent> I want to focus on admin guide for today
21:28:54 <danwent> my assumption is that most of the service-framework, service-type, etc capabilities will be documented in the admin guide only to the degree that they are necessary for setting up Lbaas, and thus will be tracked under lbaas install.
21:28:59 <danwent> (a later section)
21:29:24 <danwent> does anyone here disagree?
21:30:16 <danwent> ok, anything we need to doc in the admin guide on XML or pagination?
21:30:19 <garyk> sounds reasonable
21:30:29 <danwent> or anything we need to doc in terms of the CLI?
21:30:36 <danwent> (i.e., force the CLI to use XML
21:30:46 <danwent> ?)
21:30:47 <gongysh> I will look after the CLI doc.
21:31:10 <danwent> gongysh: ok, thanks.  I added a section below on the page for CLI, but it is currently empty
21:31:39 <gongysh> I forgot the wiki page name. I added it to doc project and our meeting agenda page.
21:31:40 <danwent> ok, on metadata
21:32:07 <gongysh> I will update this page too.
21:32:18 <danwent> gongysh: thx
21:32:34 <danwent> we will need metadata agent install
21:32:56 <danwent> and config.  this is being tracked with: https://bugs.launchpad.net/openstack-manuals/+bug/1099573
21:32:57 <uvirtbot> Launchpad bug 1099573 in openstack-manuals "Quantum Metadata Proxy should be documented" [High,Confirmed]
21:33:27 <danwent> i have a separate bug i'm working on updating nova.conf flags for quantum, which will cover metadata settings there among other things.
21:34:16 <markmcclain> I was out of the office Friday, but I can work on the Quantum side of this
21:34:20 <danwent> markmcclain: we'll also need to update the later metadata section in the quantum admin guide, currently under L3
21:34:35 <danwent> markmcclain: hehe, good.  I think i already assigned the bug to you :P
21:34:44 <gongysh> I think DHCP agent can do it too.
21:34:50 <danwent> we should probably pull that out from under L3, as L3 is no longer a requirement for metadtaa
21:35:27 <danwent> we should also rewrite the text, as we know have fewer assumptions about how things must be setup (i.e., the fact that we can route between nova-api and quantum-l3-agent host)
21:35:27 <markmcclain> danwent: yeah I think it makes sense to pull it into it's own section
21:36:03 <danwent> for reference, this is current section: http://docs.openstack.org/trunk/openstack-network/admin/content/adv_cfg_l3_agent_metadata.html
21:36:08 <danwent> i will add that to the bug
21:36:16 <gongysh> maybe under using quantum with nova section.
21:36:28 <danwent> i'd put the metdata agent as its own top level service, along with dhcp, l3, etc
21:36:45 <markmcclain> danwent: That was my thinking too
21:36:47 <danwent> but whatever, there are many ways to skin that cat :)
21:37:22 <gongysh> danwent: you are crude. :)
21:37:42 <danwent> ah, sorry, common figure of speech :)
21:38:09 <danwent> ok, next up on the doc-list, scheduler: https://blueprints.launchpad.net/quantum/+spec/quantum-scheduler
21:38:17 <danwent> i know we already have docs on this in admin guide
21:38:40 <danwent> gongysh, can you just link to the existing bug, and highlight any gaps that you think still need to be added to the docs?
21:38:51 <gongysh> sure
21:39:05 <gongysh> I will update our doc wiki page this time.
21:39:09 <danwent> https://bugs.launchpad.net/openstack-manuals/+bug/1156136
21:39:10 <uvirtbot> Launchpad bug 1156136 in openstack-manuals "Document quantum agent management and scheduler extension with multiple DHCP agents" [Undecided,In progress]
21:39:25 <danwent> gongysh: is that main bug?
21:39:29 <gongysh> yes
21:39:40 <danwent> k, will add that.  feel free to update wiki page further
21:39:48 <gongysh> ok
21:40:08 <danwent> on db migration
21:40:16 <danwent> https://bugs.launchpad.net/openstack-manuals/+bug/1099574
21:40:17 <uvirtbot> Launchpad bug 1099574 in openstack-manuals "Quantum DB migration" [High,Confirmed]
21:41:10 <danwent> markmcclain: this talks only about folsom->grizzly
21:41:17 <danwent> any doc needed for just a clean grizzly install?
21:41:31 <danwent> i will clean up wiki, as I was thinking that bug was about a clean install
21:41:33 <markmcclain> right now for a clean install.. the server will just create the tables
21:41:59 <markmcclain> but we should have them stamp the tables so that Havana migrations will just work
21:42:20 <danwent> markmcclain: yeah, that's what I figured
21:42:20 <markmcclain> Want to me file another bug or add more detail to this description?
21:42:31 <danwent> totally up to you
21:42:37 <danwent> just update the wiki accordingly
21:42:58 <markmcclain> will do
21:43:09 <danwent> on Nova / Quantum integration
21:43:29 <danwent> there are some misc update to nova.conf for quantum in grizzly.  I am working on those, and tracking with https://bugs.launchpad.net/openstack-manuals/+bug/1156822
21:43:30 <uvirtbot> Launchpad bug 1156822 in openstack-manuals "update nova config section of quantum admin guide for grizzly" [High,New]
21:43:56 <danwent> there are also changes to vif-plugging config, as plugins using the port-bindings no longer need explicit vif-plugging flags
21:44:19 <garyk> danwent: i need to address tis.
21:44:24 <danwent> garyk: do you want me to just fold that second change into my main patch?
21:44:26 <danwent> up to you
21:44:43 <garyk> if you can it will be great. i'll be happy to review
21:44:55 <danwent> ok, let's do that.  I will update the description of my bug
21:45:03 <garyk> thanks!
21:46:07 <danwent> ok, on security groups
21:46:25 <danwent> arosen mentioned that admin guide has had content for security groups CLI commands for a while.
21:46:46 <arosen> There are a few doc bugs left there that i've found reading though. I've added them to the doc impact wiki
21:47:08 <danwent> I will also be updating nova section instructing users to use Noop FW driver for nova, and to proxy calls to quantum.
21:47:25 <danwent> this will let us remove the limitation about security groups not working with overlapping IPs.
21:48:19 <danwent> arosen: I will review your doc bug fix patches later today
21:48:31 <gongysh> nova secu* commands will be proxied to quantum?
21:48:38 <arosen> gongysh:  yup.
21:48:45 <gongysh> great.
21:48:50 <arosen> if you security_group_api = quantum in nova.conf :)
21:49:01 <danwent> ok, now lbaas
21:49:29 <danwent> i created a bug around porting the basic lbaas workflow to the admin gudie: https://bugs.launchpad.net/openstack-manuals/+bug/1156828
21:49:30 <uvirtbot> Launchpad bug 1156828 in openstack-manuals "q-admin: basic lbaas workflow" [High,New]
21:50:14 <danwent> this is what I would consider "bare minimum", and someone else who wants to really drive the lbaas docs are welcome to dive in, but I'm going to leave it at the basics in terms of my own work until I feel better about the rest of the quantum docs.
21:50:40 <danwent> we also need to create install/setup sections for lbaas on both ubuntu + fedora/rhel platforms.
21:51:17 <danwent> I will create bugs for both of those as well
21:51:38 <danwent> ok, on CLI, sounds like gongysh will update the wiki with pointers to some bugs he has already created.
21:51:48 <gongysh> yes.
21:51:53 <danwent> on horizon, amotoki is out today, but you can see he updated the general agenda with info
21:52:51 <danwent> #todo #danwent work with #amotoki to make sure there are DocImpact entries for all horizon/quantum doc issues
21:53:03 <danwent> most obvious is showing how to enable Lbaas in horizon
21:54:05 <danwent> other than that, there are plugin updates, and I will be adding a table describing support plugins and what virt layers they are compatible with
21:54:44 <danwent> please remember to tag all quantum related doc bugs with "quantum" so that they show up in the query
21:54:49 <danwent> https://bugs.launchpad.net/openstack-manuals/+bugs?field.tag=quantum
21:55:56 <danwent> #topic open discussion
21:56:02 <danwent> one items I wanted to bring up was packaging
21:56:21 <danwent> and the fact that as we get closer to release, its important to be testing with packages, not from source
21:56:53 <danwent> garyk: any comments on status or fedora/rhel packaging?
21:57:15 <danwent> i've been flighting through some issues with ubuntu packaging and filing some bugs
21:57:21 <garyk> danwent: we just need to pull the latest tarball and then we can do the fedora.
21:57:23 <danwent> or -> of :)
21:57:35 <garyk> rhel will be a little while later
21:57:53 <danwent> garyk: ok.  is there going to be another fedora test day, like we had in the past?
21:58:08 <garyk> danwent: not sure. i'll check and get back to you guys
21:58:20 <gongysh> obviously I need a guide how to test the packaging, in particular where to get
21:59:13 <garyk> gongysh: i'll send you links tomorrow for fedora
21:59:18 <danwent> gongysh: for ubuntu, i've been using https://launchpad.net/~openstack-ubuntu-testing/+archive/grizzly-trunk-testing
21:59:30 <gongysh> I updated slides on Technical Architecture of Quantum http://sdrv.ms/YdnQYS
21:59:43 <gongysh> garyk and danwent: thanks
22:00:03 <danwent> gongysh: cool
22:00:07 <garyk> i am going to crash. have a long day tomorrow. goodnight and thanks!
22:00:10 <danwent> ok, times up.  anything else?
22:00:20 <danwent> one more week of focus on bugs + docs
22:00:21 <Shree-HP> I have question regarding tempest coverage for Quantum. I have filled some blueprints last week and was wondering if we can get them reviewed./approved from the group so that we can submit the code
22:00:27 <danwent> and I think we'll be in good shape
22:00:55 <danwent> Shree-HP: there's a group of folks on the ML who have been working on tempest items
22:01:01 <danwent> include some tempest code that is already in review
22:01:17 <danwent> happy to have you and others contributing, just want to make sure we aren't duplicating effort
22:01:37 <danwent> i'd either look up the thread on the ML and reply, or send me an email I can put you in touch with the people leading this work.
22:01:58 <danwent> ok, thanks folks!  i'll be around after the meeting for a bit, but want to close the official proceedings :)
22:02:01 <danwent> #endmeeting