Improve goal node start error reporting in case of algod errored (alg…

…orand#1513)

Improve node error reporting in case of algod errors

nodecontrol/algodControl.go

@@ -185,9 +185,8 @@ func (nc *NodeController) StartAlgod(args AlgodStartArgs) (alreadyRunning bool,
 		errLogger.SetLinePrefix(linePrefix)
 		outLogger.SetLinePrefix(linePrefix)
 	}
-
 	// Wait on the algod process and check if exits
-	algodExitChan := make(chan struct{})
+	algodExitChan := make(chan error, 1)
 	startAlgodCompletedChan := make(chan struct{})
 	defer close(startAlgodCompletedChan)
 	go func() {
@@ -202,14 +201,14 @@ func (nc *NodeController) StartAlgod(args AlgodStartArgs) (alreadyRunning bool,
 			}
 		default:
 		}
-		algodExitChan <- struct{}{}
+		algodExitChan <- err
 	}()
-
 	success := false
 	for !success {
 		select {
-		case <-algodExitChan:
-			return false, errAlgodExitedEarly
+		case err := <-algodExitChan:
+			err = &errAlgodExitedEarly{err}
+			return false, err
 		case <-time.After(time.Millisecond * 100):
 			// If we can't talk to the API yet, spin
 			algodClient, err := nc.AlgodClient()

nodecontrol/nodeControlErrors.go

@@ -20,6 +20,20 @@ import (
 	"fmt"
 )
 
-var errAlgodExitedEarly = fmt.Errorf("node exited before we could contact it")
 var errKMDDataDirNotAbs = fmt.Errorf("kmd data dir must be absolute path")
 var errKMDExitedEarly = fmt.Errorf("kmd exited before we could contact it")
+
+type errAlgodExitedEarly struct {
+	innerError error
+}
+
+func (e *errAlgodExitedEarly) Error() string {
+	if e.innerError == nil {
+		return "node exited before we could contact it"
+	}
+	return fmt.Sprintf("node exited with an error code, check node.log for more details : %v", e.innerError)
+}
+
+func (e *errAlgodExitedEarly) Unwrap(err error) error {
+	return e.innerError
+}

test/e2e-go/cli/goal/expect/goalNodeTest.exp

@@ -36,6 +36,25 @@ if { [catch {
     # Stop node
     ::AlgorandGoal::StopNode $TEST_PRIMARY_NODE_DIR
 
+    # "break" the node by replacing it's ledger data files with "broken" ones.
+    lassign [exec find $TEST_PRIMARY_NODE_DIR -name "ledger.tracker.sqlite"] PRIMARY_TRACKER_DATABASE_FILE
+    exec find $TEST_PRIMARY_NODE_DIR -name "ledger.tracker.sqlite*" -delete
+    exec -- echo "1234" > $PRIMARY_TRACKER_DATABASE_FILE
+
+    # try to start the primary node, and observe the expected failuire
+    set timeout 15
+    spawn goal node start -d $TEST_PRIMARY_NODE_DIR
+    expect {
+        timeout { close; ::AlgorandGoal::Abort "starting node exceeded timeout" }
+        -re {^Algorand node failed to start: node exited with an error code, check node\.log for more details : exit status 1} {puts "Expected failuire : node failed to start"; close}
+        eof { ::AlgorandGoal::Abort "eof received before the expected output " }
+    }
+    lassign [::AlgorandGoal::CheckProcessReturnedCode 0] ERROR_CODE
+    if {$ERROR_CODE != 1} {
+        puts "Node was expected to fail with 1 due to invalid ledger file, but returned error code $ERROR_CODE instead"
+        exit 1
+    }
+
     # Shutdown the network
     ::AlgorandGoal::StopNetwork $NETWORK_NAME $TEST_ALGO_DIR $TEST_ROOT_DIR