This patch was originally rejected because this is a duplicate to PostgreSQL. We found that XC needs a watchdog timer to monitor if coordinator/datanode is alive. Traditionally, select 1 or select now is used for this purpose. In XC, when GTM is down, these command returns error. We need another method to monitor just if coordinator/datanode is alive, regardless the GTM status.
We have many more means to detect if gtm/gtm_proxy/coordinator/datanode are running. PGping() is a typical one. From a bash script, we can specify \w command to psql to surpress the warning message that gtm is not available.
We don't have a good reason to do duplicate work with PG core.