* About to take debugging comments out
* Removed debugging statements and tested again
* Removed more debugging statements and unused method
* Made changes suggested by Tamilmani, tested again.
* Made spacing changes suggested by Tamilmani
* Triggering new tests to run
* Added AITelemetry support for CNI
* added new files
* added other configs in config file
* fixed ut
* updated disableall similar to cns
* added container name to report
* addressed review comments
* addressed review comments
* added check for azure environment
* added log
* close log handle in unit test
* addressed review comments
* addressed review comments
* fixed a condition
* keep the netagent channel for logs
* fixed error
* addressed review comments
Expose 2 APIs to publish and unpublish network containers from CNS.
a. PublishNetworkContainer
b. UnpublishNetworkContainer
DNC calls CNS to publish and unpublish the network containers.
This PR adds support for host NC bidirectional communication with windows HnsV2. This is supported in multitenant scenario only. AllowHostToNCCommunication and AllowNCToHostCommunication flags are used to enable Host to NC and NC to host communication respectively.
* added application insights telemetry package to send metrics and logs to appinsights
* moved a function common to both packages
acquired a lock before writing to file
* added read write lock as per chandan comment. Addressed jaeryn comments
* fixed telemetry unit test
* defined interface and added appinsights package as vendor to acn
* added vendor package.go(appinsights)
* dependencies of appinsights
* added AI dependencies
* updated unit tests
* addressed review comments
Loopback adapter creation operation comprises of two operations - createInterface and setWeakHostOnInterface. These operations take place inside the lock. If there are simultaneous requests, it interleaves these calls causing every loopback adapter creation to absorb the delay due to interleaving. createInterface can take time in seconds (typically 2 to 7 seconds based on the tests) while setWeakHostOnInterface finishes very quickly ( less than a second ). This change calls setWeakHostOnInterface within the same lock if createInterface succeeds. The tests show this improves the loopback adapter creation times for simultaneous requests.
* change telemetry to message queue and add npm
* remove [Azure-NPM] prefix
* remove npmreport url
* fair scheduling
* holds up to 1k reports for each type
* fix cap on reports
* 1. Start telemetry service as separate process for all components cns/networkmonitor
2. Added telemetry config through which we can configure reportTohost interval
3. Added unit tests for changes
* added new files for telemetry testing
* added tests for invalid cases
* updated with dummy subid
tamilmani1989
Jaeryn
Mathew Merrick
Paul Johnston
Ashvin Deodhar
Yongli Chen
Tamilmani Manoharan
Sushant Sharma
Yongli Chen
Onur Filiz