POC: EE-2493: mtls for edge agent #234
Conversation
…s _and_ --help Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
… a feature? Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
…e.ID Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
…0, directly into the docker daemon Signed-off-by: Sven Dowideit <[email protected]>
Signed-off-by: Sven Dowideit <[email protected]>
SvenDowideit
force-pushed
the
feat/EE-2493/add-mtls-for-edge-agent
branch
from
96b0e2d to
d577451
Compare
| manager.key.PortainerInstanceURL, | ||
| manager.key.EndpointID, | ||
| manager.agentOptions.EdgeID, | ||
| GetNewHttpClient(10, manager.agentOptions), |
There was a problem hiding this comment.
maybe it's cleaner to pass the manager.agentOptions as param and call GetNewHttpClient internally
There was a problem hiding this comment.
atm, there are too many things created internally, breaking both the sharing, and hiding the high number of internal members - so I've been pulling it appart and making it more obvious.
Annoyingly, what I was originally trying to acheive started to be too big a refactor, so I'm currently doing the async thing in at a PortainerClient level, so its possibly true.
Main thrust is that the actual http requests are now in the pluggable PortainerClient, and this set of 3 managers all use the same code - except for the different timeouts, they would use the same PortainerClient instance - but that's something where I don't have time to work out the consequenses of the two 10second constants vs the User tuned one.
There was a problem hiding this comment.
yep I get that, what I had in mind was to at least pass the timeout value as a parameter and use that to call GetNewHttpClient internally, but it surely can be changed later, it's mostly a readability issue
|
Replaced by #234 |
Replaced by #249
again, based on #228 and #229
and requires the matching portainer PR