fix: do not wait for change to complete during scale up

Add it as a separate step, because this can take longer than the default 5 minutes. When the whole operation is retried
scale up request fails because there is an ongoing change. By splitting up into two steps, only query has to be retried.

go-chaos/cmd/cluster.go

@@ -111,7 +111,6 @@ func scaleUpBrokers(k8Client internal.K8Client, port int, brokers int, replicati
 	ensureNoError(err)
 	_, err = k8Client.ScaleZeebeCluster(brokers)
 	ensureNoError(err)
-	waitForChange(port, changeResponse.ChangeId)
 	return changeResponse, nil
 }
 

go-chaos/internal/chaos-experiments/camunda-cloud/scaling/broker-scaling.json

@@ -51,6 +51,16 @@
                 "arguments": ["cluster", "scale", "--brokers", "5"]
             }
         },
+        {
+            "type": "action",
+            "name": "Wait for scale up to complete",
+            "provider": {
+                "type": "process",
+                "path": "zbchaos",
+                "arguments": ["cluster", "wait"],
+                "timeout": 900
+            }
+        },
         {
             "name": "All pods should be ready",
             "type": "probe",