This might occur in case of crash on host or database node where goldengate is running.
Goldengate maintains information about the process status in the file system, when the server crashes the file that has PID data is not released, and the process status in GGSCI shows erroneous information.
In GGSCI, process status it’s OK:
GGSCI (greporaLAB.net) 7> info mgr Manager is running (IP port greporaLAB.net.7809, Process ID 42064).
If you try to stop process, you will not be able to:
GGSCI (greporaLAD.net) 5> stop mgr Manager process is required by other GGS processes. Are you sure you want to stop it (y/n)?y Sending STOP request to MANAGER ... ERROR: opening port for MGR MGR (Connection refused).
If you try to kill process: PID does not exist.
GGSCI (greporaLAB.net) 8> kill mgr ERROR: 3, No such process killing process (42064) for MGR.
To resolve, as per (Doc ID 1314220.1), is recommended to move it to another name, and then get back to the original name.
greporaLAB.net> cd $GG_HOME/dirchk greporaLAB.net> mv MGR.cpe MGR.cpe.old greporaLAB.net> cp –p MGR.cpe.old MGR.cpe
Sometimes is need to perform ETROLLOVER in the extract/pump process, because a trail file has a resource busy status.