By default ClustrixDB doesn't use tempfs/tmpfs for temporary allocations. It uses
device1-temp file, which resides on
/data/clustrix, hence on actual disk. There is no advantage in trying to use tempfs/tmpfs since ClustrixDB will send query data to disk anyway, so it is more beneficial to give ClustrixDB the extra memory as it knows best how to use RAM effectively for caching.
If you need to increase the temporary space for a ClustrixDB Cluster beyond the default (5GB per node), you can increase the global variable, device_temporary_space_limit_bytes accordingly.