09:00:55 <gsagie> #startmeeting dragonflow
09:00:56 <openstack> Meeting started Mon Dec 21 09:00:55 2015 UTC and is due to finish in 60 minutes.  The chair is gsagie. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:00:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:00:59 <openstack> The meeting name has been set to 'dragonflow'
09:01:07 <gsagie> #topic security groups
09:01:24 <gsagie> dingboopt: please share your progress on this topic
09:01:32 <dingboopt> ok
09:02:07 <dingboopt> I just set up a two node testing environment
09:02:33 <gsagie> do you have any code for security group or only design?
09:02:59 <dingboopt> Now I am do a bit research into ovs to see how I can map iptabls  into ovs stateful flow tables
09:03:11 <dingboopt> no code yet
09:03:17 <dingboopt> only design
09:03:48 <gsagie> ok, if you can you have some comments on the design document, please see if you can address them
09:04:12 <dingboopt> ok
09:04:17 <gsagie> if you have open things, its ok to put "TBD" if you dont know yet, but lets change the structure to be the formal one
09:04:37 <dingboopt> I will keep watching the design document
09:04:52 <gsagie> i also think that for first step we should focus on OpenStack security groups and then the spoofing stuff (ARP, DHCP)
09:05:21 <gsagie> #action dingboopt address design comments in https://review.openstack.org/#/c/257419/
09:05:32 <dingboopt> ok
09:05:56 <gsagie> dingboopt: do you need any help with something to move forward?
09:06:06 <gsagie> i think its important we have this feature ready for next release
09:06:49 <dingboopt> Now I can do it myself
09:06:51 <gsagie> i think the part that translate the security group from neutron to our DB is built, so thats less work for you
09:07:01 <dingboopt> yes
09:07:22 <gsagie> #info dingboopt,  kexiaodong, yuli_s, gsagie in meeting
09:07:46 <gsagie> ok, please reach out if you need help, but i think we need to start coding, even with single node
09:08:01 <gsagie> anything else on this topic?
09:08:45 <gsagie> ok
09:08:50 <dingboopt> ok. I do it soon
09:08:50 <gsagie> #topic testing
09:09:15 <gsagie> ok thanks dingboopt, its not an easy task so please reach out if you need anyone to join you/ help you out
09:09:53 <dingboopt> Ok
09:09:57 <gsagie> ok regarding testing, our fullstack tempest job has merged so in addition to tempest we can define tests that run on the gate vs a working dragonflow enviorment
09:10:09 <gsagie> #info Dragonflow fullstack job is not merged and working
09:10:15 <gsagie> #info Dragonflow fullstack job is now merged and working
09:10:54 <gsagie> The first part me and Yuli_S are working on to add tests that configure Neutron using Neutron client and tests everything is set correctly in the DB, we can then convert these to Rally tests as well
09:11:04 <gsagie> yuli_s: anything to add? what is your focus right now?
09:11:20 <yuli_s> I was working on tests
09:11:38 <gsagie> you creating tests for L3/Router APIs right?
09:11:48 <yuli_s> i think to add additional checks for db
09:11:51 <yuli_s> yes
09:12:13 <gsagie> #action yuli_s adding fullstack tests for L3 router interface add/delete
09:12:17 <gsagie> Hello Shlomo
09:12:28 <Shlomo_N> Hey
09:12:45 <gsagie> #action gsagie add fullstack tests for port creation/deletion
09:13:55 <gsagie> Shlomo, after you finish Rethinkdb integration we have different areas we might need work, unit tests, fullstack tests related to OVS (extracting the flows and checking them) and scale testing using shake and Rally, what you prefer to work on?
09:14:37 <Shlomo_N> Scale testing would be better
09:15:07 <gsagie> #action Shlomo_N to investigate scale testing and shaker performance testing
09:15:17 <gsagie> #action gsagie add unit tests for Dragonflow
09:16:01 <gsagie> Shlomo_N: Li Ma is also working on performance and scale testing, we will need to see how to combine actions, the following document is used to track this
09:16:17 <gsagie> #link https://docs.google.com/document/d/1TK10NfVhTv0-tsvWm4fhd1M0lftb27p_H7D1TsdV4Tg/edit?usp=sharing   - Dragonflow scale testing
09:16:46 <gsagie> We basically have few fronts here, both performance with few nodes and scale with many nodes
09:16:51 <gsagie> Aharon, Dima here by any chance?
09:17:29 <gsagie> #action gsagie to work with Aharon on DB testing framework
09:17:43 <gsagie> #action gsagie and Shlomo_N to work on shaker bring up
09:18:10 <gsagie> We also started to run tempest gate tests, we will need to start fixing some bugs there, i will take it on me
09:18:18 <gsagie> #action gsagie start fixing tempest bugs
09:18:42 <gsagie> #info anyone adding new feature need to add tests - unit/functional/fullstack
09:18:49 <gsagie> anything else on this topic from anyone?
09:19:07 <gsagie> #info Shlomo_N and CrayZee in the meeting as well :)
09:19:41 <gsagie> #info Waiting Dragonflow Rally integration patch to merge - https://review.openstack.org/#/c/258347/
09:19:57 <yuli_s> i think in addition to check using high level API
09:20:19 <yuli_s> in case on a failure we need to check if we got the same results from db directly
09:20:22 <gsagie> yuli_s : yeah thats what i call fullstack, what we both are doing, or you mean anything else?
09:20:43 <yuli_s> nop,
09:21:00 <gsagie> ok, yeah we do need to also start checking failures
09:21:14 <gsagie> and how the system react to them
09:21:21 <yuli_s> as part of full stack tests we neec to add addition direct to db level in case high level API returns empty result
09:21:32 <yuli_s> yes
09:21:36 <gsagie> yuli_s: yes i agree
09:22:04 <gsagie> #topic Kuryr and Dragonflow integration
09:22:18 <gsagie> kexiaodong: any update ?
09:23:09 <gsagie> i saw you submitted the OVS unbind patch
09:23:10 <kexiaodong> for unbind ,update first patch
09:23:35 <gsagie> and its working?
09:23:42 <gsagie> i added some comments on your patch
09:24:03 <kexiaodong> I test it in my env, it works
09:24:19 <gsagie> ok, just few comments i gave you there to address
09:24:28 <kexiaodong> I will check it later
09:24:40 <gsagie> the next step is we need to finish the integration with Dragonflow
09:25:14 <gsagie> the missing part, is that Kuryr will need to use the df-db cli tool to set the logical port hypervisor
09:25:24 <gsagie> the Dragonflow bind/unbind script
09:26:36 <gsagie> as Dragonflow currently needs the binding information to be set
09:26:47 <gsagie> kexiaodong: you think you will have time to work on this?
09:27:07 <kexiaodong> I will let dingboopt give me some detail about df-db cli tool
09:27:31 <gsagie> ok, i will update you offline, it needs some modification
09:28:02 <gsagie> #action kexiadong to work on bind/unbind for dragonflow, basically use df-db cli tool to set the port hypervisor in the binding information
09:28:52 <gsagie> the problem in a nutshell, is that when Kuryr calls add-port to Neutron it doesnt set the binding information, and Dragonflow needs this information
09:29:05 <gsagie> thats the only missing part beside the OVS bind/unbind
09:29:28 <gsagie> kexiaodong: anything else?
09:30:41 <gsagie> #topic DB
09:30:52 <gsagie> Shlomo_N: any update on RethinkDB integration?
09:31:11 <Shlomo_N> I am working on multinode support
09:31:34 <gsagie> ok
09:31:36 <Shlomo_N> We need to decide if we want to support Fedora as well
09:32:02 <gsagie> #action Shlomo_N finish RethinkDB integration , patch should merge
09:32:24 <gsagie> We do, but we will need to see if thats the most urgent task right now
09:33:23 <gsagie> #info Bug added for supporting RethinkDB in fedora https://bugs.launchpad.net/dragonflow/+bug/1527217
09:33:23 <openstack> Launchpad bug 1527217 in DragonFlow "RethinkDB installation only works in Ubuntu" [Undecided,New]
09:33:47 <Shlomo_N> ok
09:34:07 <gsagie> We also need to relook at the rejoin stack, i dont think its working with etcd, i will do it
09:34:22 <Shlomo_N> Yuli worked on this
09:34:24 <gsagie> #action gsagie make rejoin_stack.sh work
09:34:40 <gsagie> yes, but things have changed in etcd and we need to do some changes
09:34:47 <gsagie> in the installation script
09:35:05 <Shlomo_N> There are few bugs with the rejoin, for RAMCloud and RethinkDB as well
09:35:51 <gsagie> for RAMCloud we cant address this, the problem is that all data is stored on RAM and hence if we only have one server we need to perform a sync operation with Neutron to make it work
09:36:27 <gsagie> but thats not the usual usecase for RAMCloud deployments, you will usualy have cluster with some nodes always alive, so i think we should wait with this
09:36:39 <gsagie> for Rethinkdb, i will put this action on you, you know how to solve it?
09:36:42 <Shlomo_N> https://bugs.launchpad.net/dragonflow/+bug/1527970
09:36:42 <openstack> Launchpad bug 1527970 in DragonFlow "rejoin-stack doesn't support RethinkDB" [Undecided,New]
09:37:02 <gsagie> #link https://bugs.launchpad.net/dragonflow/+bug/1527970
09:37:11 <Shlomo_N> Didn't checked yet, but I can check it.
09:37:12 <gsagie> #action Shlomo_N working on RethinkDB rejoin stack bug
09:37:17 <gsagie> okie, thanks
09:37:42 <gsagie> #action for RAMCloud rejoin stack will not be supported at this time, we will address this in the future if needed with a sync process
09:37:49 <gsagie> #info for RAMCloud rejoin stack will not be supported at this time, we will address this in the future if needed with a sync process
09:38:02 <gsagie> anything else on the DB?
09:38:18 <Shlomo_N> nop
09:38:23 <gsagie> #topic Bugs
09:38:39 <gsagie> Anyone has any remarkable bugs that they are blocked in?
09:38:52 <gsagie> beside the rejoin stack part
09:39:00 <gsagie> and there is a traceroute bug which i will check
15:36:56 <devvesa> apuimedo: you are welcome
15:37:00 <apuimedo> devvesa: you'll be taking a look at rdo or ubuntu packaging next?
15:37:16 <devvesa> Yup. add it in an action if you want to
15:37:31 <devvesa> Let's start with RDO :)
15:37:32 <apuimedo> devvesa: which of them?
15:37:34 <apuimedo> ok
15:37:50 <apuimedo> #action devvesa to work on rdo packaging
15:37:57 <apuimedo> this should also be for mitaka-2
15:38:04 <apuimedo> Any other topic?
15:38:06 <devvesa> This is not up to me
15:38:38 <devvesa> I'll need to fill a bug in bugzilla and see the priority they are giving to it
15:38:44 <devvesa> *will give to it
15:39:05 <apuimedo> #action apuimedo to use his dusty Red Hat to help push it
15:39:39 <apuimedo> s/it/the bug for rdo packaging/
15:41:16 <apuimedo> alright then
15:41:21 <apuimedo> thank you all for joining
15:41:32 <apuimedo> have a nice holidays those who have them
15:41:46 <apuimedo> #endmeeting