|Pivotal Greenplum (GPDB)||4.3.x|
If the master is rebooted while the GPDB cluster is up, this will leave all segment instances up and running while the master instance is down.
The following process was followed:
- gpstart -m (start the master instance only)
- gpstop -M fast -a (once the master instance has started, shut down the whole cluster for a clean startup)
- gpssh -h hostfile ' ps -ef |grep post' (ensure that no orphan process left at all hosts)
- gpstart -a (start the database again)
- Follow the directions in the KB article listed below to run a full gpcheckcat: