While testing a 3 node percona cluster we often get:
Errcode: 28 - No space left on device
After login to the servers and checking the defined
tmpdir I notice that is empty and clients can connect again without any issues, therefore wondering if there is an auto purge/expunge variable or setting that without rebooting the servers is cleaning the
tmpdir or if there is any sql command that could help on demand clean/flush the tmp directory.