CXPACKET waits may or may not be an issue for a system, but on an OLTP style system, they can indicate costly queries that are executing and requiring parallelized plans.
An alert is not defined, because the normal value for this metric can vary greatly from system to system, but you could add an alert once the baseline for your environment has been established.
OS Wait Time: CXPACKET
CXPACKET waits indicate coordination of multiple threads from queries that have been parallelized and one or more of the subtasks have completed before others. These waits may or may not be an issue, but on an OLTP system, high CXPACKET waits may indicate too many complex queries needing parallel operations. Long wait times captured by this metric may indicate queries that are taking a long time to complete, even using parallel query plans. Note: Values are calculated/accumulated since the last restart of the server, or if they are reset using: DBCC SQLPERF ('sys.dm_os_wait_stats', CLEAR);
The T-SQL query that will collect data
Instances to collect from
Databases to collect from
Use collected or calculated values
Check the Use a calculated rate of change between collections check box
You could add an alert once the baseline for your environment has been established, and you have a better idea of what thresholds to set.