When running gpbackup for large amount of data, it may fail with below error
gpbackup:gpadmin:hostname:xxxxx-[CRITICAL]:-ERROR: invalid snapshot identifier: "1794889C-1" (SQLSTATE 22023)
When gpbackup is started, it will start two pid at the master host. From the pg_stat_activity view, there will be two session. One is gpbackup_xxxx, one is copy . The gpbackup session is the main session, when the copy session is running, the gpbackup session will be in idle status. If there is an auto-clean script running in the backend, like automatically run pg_terminate_backend for any session being idle for more than certain period, then this gpbackup session could be stopped by this clean up script or by any manually run idle session clean up.
When gpbackup is running, it's better not to run any auto clean up script, or at least exclude gpbackup session from the script.