Windows: Support local drivers internal, l2bridge and nat#3121
Conversation
corhere
left a comment
There was a problem hiding this comment.
I messed up. I'm really sorry. The Linux/Darwin manager drivers are not all the same, despite my earlier claims. The dynamic manager driver will only work for the Windows network types.
do that by utilizing new dynamic manager Signed-off-by: Olli Janatuinen <olli.janatuinen@gmail.com>
5a0da59 to
5ffbc07
Compare
Codecov Report
@@ Coverage Diff @@
## master #3121 +/- ##
==========================================
- Coverage 61.73% 61.62% -0.11%
==========================================
Files 153 154 +1
Lines 31099 31142 +43
==========================================
- Hits 19198 19192 -6
- Misses 10364 10402 +38
- Partials 1537 1548 +11 |
I is not all that bad. As far I can see those was added by moby/libnetwork#1742 to support moby/moby#32981 it is just those two more parameters which was needed and we can use same logic for Linux too. Starting to look good for me as this will actually remove more vendored code than add new one. EDIT: Oh, I can see it now when looked removed code in details so kicked it back to containing Windows only change for now. |
5ffbc07 to
9177d5a
Compare
corhere
left a comment
There was a problem hiding this comment.
LGTM! @dperny @neersighted PTAL
|
@dperny PTAL |
- What I did
This PR takes generic driver manager which @corhere built on moby/moby#44983 and includes it to Swarmkit instead of libnetwork like proposed on moby/moby#44983 (comment)
In additionally that driver manager is used to register
internal,natandl2bridgedrivers for Windows so scenarios explained on moby/moby#44983 can be solved.- How I did it
Copy/paste from that another PR, minimal refactoring to existing code to support it and revendor to remove libnetwork code which is not needed here anymore.
- How to test it
If we get CI to 🟢 it should enough.
- Description for the changelog