Attempt to stabilize OPC UA unit tests #1797
Conversation
splatch
force-pushed
the
stabilize-opc-ua-tests
branch
3 times, most recently
from
d00c866
to
1737941
Compare
|
@chrisdutz Looks like tests after above changes are better. Would you risk merging this to see how it will go with Jenkins? Can you give it a try on Mac/Parallels? |
|
How about manually re-running the tests for 4-5 times ... if they remain unflaky, please merge them ... if they remain flaky, please don't enable them. I'd really like to have our build in a form again, that we can trust it and not blindly click "rerun failed" for 2-3 times before we start looking if it's really a problem. |
|
Hmmm ... odd ... I had that error a few days ago ... as if Java forgot it's own options ... maybe just run again later. |
|
The good thing is - build didn't fail due to opcua :-) |
splatch
force-pushed
the
stabilize-opc-ua-tests
branch
2 times, most recently
from
1ff40e0
to
4d8db15
Compare
First and foremost - this commit switches OPC-UA tests to rely on dynamic port. This implies also slight change to endpoint selection method and possibility to override desired endpoint port and host which might differ from host/port reported by OPC-UA server. Because server is not aware of actual tcp port and host used by client, it is client duty to make proper decission when looking for endpoint. Related to #1764. Signed-off-by: Łukasz Dywicki <[email protected]>
splatch
force-pushed
the
stabilize-opc-ua-tests
branch
from
4d8db15
to
896d869
Compare
|
@chrisdutz Ping - looks like tests are a bit better on github, there are fewer failures coming from OPC-UA. Would you risk merging them now? |
|
Well ... let's see ... if they start randomly failing the build there, I'll deactivate them again ... so let's consider them "on probation" ;-) |
|
I see chunk tests are failing again, will have a closer look on them. Milo seems to be fine with testcontainers and randomized network port. |
Related to #1764.