* added a bit of logging
* added update messages to mqtt (every second with a timestamp)
* added isConnectionOpen to the interface and analysis this during execution
* added connection loss handler and status check
* put some global objects into an objects to avoid inter service testing issues
* TestingLocker Mutex to RWMutex (increase performance)
* termintationMutex became a sync.RWMutex
* added logging to the mqtt mock client
* made monitor Mutex a RWMutex
* added Mutex protection to the numberOfElements varialbe
* added another Waiting block to the machine attestation methods (CI tests)
* had to adjust the test cases to the impact of that change.
Signed-off-by: Jürgen Eckel <juergen@riddleandcode.com>
* added a MqttMonitor module with levelDB and periodic cleanup
* initialized in the app
* passed to dao keeper
* added conversion methods (string2unixtime, byte ToJSON)
* removed obsolete keeper code
* maded RDDLToken.Factor public
* added explicit mqtt client to the monitor module
* restart mqtt connection in mqttmonitor on connection loss
* adjusted mqttmock structure to be compatible
* added some linter exclusions to let the monitor tool pass
* created a MockMqttMonitor interface and mock object
* used this to pass tests
* made the MockMqttMonitor a global object so that it can be easily mocked
* removed MockMqttMonitor from the app/keeper initialization
* adjusted test cases to register "active machines" to the mqttmonitor
* added mutex in mocks to protect against data races
* defined mocks for the dao tests
* clear separation between interface and mqtt-Monitor
* added another waiting block to ensure the tx went through (multi-threading issue, race condition) during tests this failed sometimes
* added memstorage to test instead of a file based DB
Signed-off-by: Jürgen Eckel <juergen@riddleandcode.com>
* Added more sync objects to avoid accidental data races
* made lib.config.Config variables are private, so they cannot be tampered with (multithreading).
Please introduce Get-methods to retrieve the varialbes outside of the package if you need it.
This way, the race conditions and unexpected change of the global object state can be protected
* added reverse takedown of the validators to avoid the following error
[app] PoP broadcast tx failed: node0.info: key not found
* moved util.TerminationWaitGroup.Wait()
to the central network cleanup method
* removed mappings for private variables
Signed-off-by: Jürgen Eckel <juergen@riddleandcode.com>
* added wait for another block to have more reliable tests
* Added one network reload in case the address/port was still bound during the last try
test related
* separated Dao test suites (faster output on the CI)
* added 2 seconds wait time to avoid errors of still blocked ports/bindings (network)
* removed parallel flag from test/e2e/asset
* renamed a bunch of things to please the linters
* moved the reset of the libConfing clientctx back to the validator context before waiting for some blocks.
This should prevent the
panic: Log in goroutine after TestE2EMachineTestSuite has completed: [app] PoP broadcast tx failed: node0.info: key not found
on the CI.
Signed-off-by: Jürgen Eckel <juergen@riddleandcode.com>
