#opendaylight-meeting: MD-SAL Interest Call
Meeting started by Sai at 16:02:08 UTC
(full logs).
Meeting summary
- - last meeting minutes (Sai, 16:02:50)
- - https://meetings.opendaylight.org/opendaylight-meeting/2016/md_sal_interest_call/opendaylight-meeting-md_sal_interest_call.2016-03-15-16.00.html
(Sai,
16:02:53)
- Agenda (Sai, 16:03:07)
- 1. Call for topics (Sai,
16:03:35)
- 2. Review the existing action items
(Sai,
16:03:39)
- 3. Bugs (Sai,
16:04:01)
- 4.Upgrades in Boron (Sai,
16:04:17)
- Review action items (Sai, 16:05:01)
- ACTION: - Ryan will
further investigate on DTCL over DCL (really just add children to
toaster example) - STILL IN PROGRESS (Sai,
16:05:23)
- - TomP to followup with colindixon regarding
upgrades- Done in TWS (Sai,
16:05:46)
- - TomP to link the release review plan with
upgrades & config subsystem - Done in TWS (Sai,
16:06:06)
- Bugs (Sai, 16:06:36)
- - https://bugs.opendaylight.org/show_bug.cgi?id=2825
(Sai,
16:08:36)
- - Anton raised it as an "umbrella bug"
(Sai,
16:09:28)
- - Tony suggested Robert may have more
details (Sai,
16:10:53)
- - Tony says utility classes have been added as
part of patch (Sai,
16:11:17)
- BUG 2825 has been partially addressed by
classes that bypass regex checks when building some IETF types from
bytes (colindixon,
16:11:19)
- - BUG 2825 has been partially addressed by
classes that bypass regex checks when building some IETF types from
bytes (Sai,
16:11:49)
- - Robert says they have two revisions of inets,
each of them have utility classes. (Sai,
16:13:25)
- - colindixon asks if it posssible to have
canonincal form of each type. Multuple versions of vsame value
results in .equal value doesn't work (Sai,
16:14:19)
- - Robert says gist is yang defined are being
dealt in strings and there is no other way of doing it. (Sai,
16:15:10)
- - Robert says original work uses binary
representation as internally which is really hard (Sai,
16:15:41)
- - Robert says Ipaddress / prefix/ Mac address
are defined here any modifications will lead to
complications. (Sai,
16:17:05)
- - colindixon says if you use canonical
ipaddress you need not worry about rest of the things. Are there any
best practices ? (Sai,
16:17:49)
- - If some one pushes it through restconf things
get interesting. (Sai,
16:18:22)
- colindixon asks if ther'es a way to sanely
canonicalize values that are set outside your purview, e.g., set via
RESTCONF (colindixon,
16:19:01)
- - Robert says changes for Ipv6 utils are not
done yet. (Sai,
16:19:17)
- rovarga_ says not at the moment, some classes
exist to help (colindixon,
16:19:26)
- - Robert says in case of Ipv6 it is hard to say
what exactly is canonical format (Sai,
16:19:41)
- - colindixon asks if there is one unit of
utility classes where canonical forms are defined that would be
great. (Sai,
16:20:23)
- - Robert says there is no way with out
interacting with datastore .. (have to convert it back ) .. so there
is not much value in creating it. (Sai,
16:21:39)
- - Robert says There is already a utility class
which has canonical forms with ipv4 (Sai,
16:23:34)
- - Robert says converting bytes result in
canonical , converting from bytes it could be another adddress or
byte . So it is possible to generate the end result as canonical
form (Sai,
16:25:49)
- Robert brings up MAC address issue in case of
canonical forms and says overriding the MAC address is the only
possible solution (Sai,
16:27:44)
- Bug 5396 (Sai, 16:28:03)
- - Regex processing of yang models is broken.
Models that worked on Li does not work on Be (Sai,
16:28:45)
- - Peter able to reproduce the bug (Sai,
16:29:20)
- - Robert says thisyang model has particular
pattern like union, this is a specific scenario. (Sai,
16:30:14)
- - Bala says Behavioral change of adding quotes
is observed in Beryllium. (Sai,
16:30:52)
- ACTION: - Ryan &
Bala will link the Bug (behavioral constraints in Be 5484) to Bug
5396 (Sai,
16:31:50)
- - Bug 5396 is critical when compared to Bug
5484 (Sai,
16:32:16)
- - Bug 5484 - Tony says the leaf vhost is
valid (Sai,
16:33:56)
- - Bala says by adding quotations to patterns in
5484 we could make it work (Sai,
16:34:35)
- - Bala requests Tony to add comments on Bug
5396 which helps peter (Sai,
16:35:56)
- - Robert says Bala may need to ask Peter about
resolution time (Sai,
16:37:22)
- - Tony says Bug may not be in code and he will
sync up with Peter (Sai,
16:37:52)
- ACTION: - Tony will
sync up with Peter wrt Bug 5396 (Sai,
16:38:07)
- - Filter framework for netconf (Sai, 16:38:50)
- -Ryan added a patch - standard framework for
modyfying /adding . (Sai,
16:40:05)
- Ryan requests netconf experts tcere to have a
look (Sai,
16:40:38)
- - tcere says he will go through it (Sai,
16:40:51)
- Upgrades in Boron (Sai, 16:41:13)
- - colin asks if robert & tony had time to
go through the TWS meeting minutes. (Sai,
16:41:42)
- - TomP says he can have a meeting with Robert
and Tony regarding the upgrades (Sai,
16:42:14)
- - colindixon says we don't have any concrete
plan as of now (Sai,
16:42:36)
- - colindixon asks what happens if revision of
model change but it is not backward compatible ? (Sai,
16:42:54)
- - colindixon say as long as we get things right
in Boron which we did wrong in previous releases that should be good
start (Sai,
16:43:57)
- - TomP says his focus as of now is on config
subsystem (Sai,
16:44:22)
- - TomP asks for volunteer on the models
removal/addition/changes in process of upgrades (Sai,
16:45:12)
- - colindixon inutuition is better to make
manually at application layer that could work (Sai,
16:46:34)
- - colindixon asks for faster (two version of
copies get loaded) (Sai,
16:48:00)
- Tony says karaf experts don't suggest so
(Sai,
16:48:13)
- - colindixon says if users modify something, it
is going to be their responsibility to change it back. (Sai,
16:49:24)
- colindixon asks if we delete the data
directory, unzip the new release over the old one, and just go for
it, do we expect things go wrong (colindixon,
16:50:35)
- - colindixon asks about deleting data
directory issues (Sai,
16:51:13)
- ttkacik says talking to the Karaf folks, they
think thalt this unzipping over is probably a bad idea (colindixon,
16:51:54)
- - Tony says if you override the existing the
logs is a bad idea (Sai,
16:52:20)
- instead people recommend unzipping to a new
location and then moving the stuff you want to keep over
(colindixon,
16:52:38)
- Tony says we need to know the previous
logs/data so that we can move them back, possible solution is to put
new logs in a different location (Sai,
16:53:26)
- TomP and others seem to think that an install
script which copies over just what you want might be the right
approach (colindixon,
16:53:47)
- Ryan asks Tony if there any recommendations
from karaf guys (Sai,
16:53:50)
- Tony says he'd go with that, but notes that the
Karaf folks didn't have a strong suggestion other than recommending
to avoid unzipping over (colindixon,
16:54:21)
- Tony says zipping is not an ideal solution,
moving files is not much more complex. (Sai,
16:54:46)
- - TomP says moving and then merge or may be
rebioult features cfg may be a feasible solution (Sai,
16:55:24)
- - Ryan says there is already an compatibility
matrix where we can add the upgradability matrix (Sai,
16:57:06)
- - colindixon says once we have scripts for
upgrades, compatibility matrix should be doable (Sai,
16:57:59)
- - colindixon says checking data before and
after upgrades would be first step (Sai,
16:59:31)
- - colindixon asks TomP to poll with each
project if the data is not is as expected (Sai,
17:00:25)
- Participants (Sai, 17:01:41)
- - Sai/Phil/Ryan/Alexis/An Ho/ Atul Gosain/
Bala/ Colin/ Robert/Stephen/Tcere/TomP/Tony (Sai,
17:02:41)
Meeting ended at 17:02:45 UTC
(full logs).
Action items
- - Ryan will further investigate on DTCL over DCL (really just add children to toaster example) - STILL IN PROGRESS
- - Ryan & Bala will link the Bug (behavioral constraints in Be 5484) to Bug 5396
- - Tony will sync up with Peter wrt Bug 5396
People present (lines said)
- Sai (82)
- colindixon (8)
- odl_meetbot (7)
Generated by MeetBot 0.1.4.