Hi Saranee,
In general it is better to not have a node fail in the first instance but rather construct the logic so that the upstream node from the MetaCheck node outputs zero records when the relevant condition is encountered. You can then configure the MetaCheck node to test for zero input records and clock one sub-flow on success (> zero records) and a different sub-flow on failure (zero records).
If the graph is scheduled then you can split graph into different parts and the schedule to execute one sub-graph on success of the previous graph and a different graph on failure of the previous graph.
It would also be useful if you could provide some additional information on the node that is failing and the error message.