OGG-06530 Dirbdb On ZFS Corruption On GG Startup

Last updated on FEBRUARY 08, 2018

Applies to:

Symptoms

Datastore being corrupted every time when Goldengate is started.

The GG processes (manager, extract/replicat, jagent) then become unresponsive.
Recreating the datastore allows the processes to restart but then the datastore becomes corrupt again.

Goldengate Version : 12.2.0.1.170221 

Operating System : Linux, x64, 64bit 

Database Version: Oracle Database 12c Enterprise Edition Release 12.1.0.2.0 - 64bit Production

ggserr.log

=======
See the following error:

2017-12-12 08:46:52 WARNING OGG-06530 Oracle GoldenGate Command Interpreter for Oracle: Berkeley Database encountered a critical error.
2017-12-12 08:52:32 INFO OGG-00987 Oracle GoldenGate Command Interpreter for Oracle: GGSCI command (oracle): create datastore.
2017-12-12 08:54:30 INFO OGG-00987 Oracle GoldenGate Command Interpreter for Oracle: GGSCI command (oracle): delete datastore.
2017-12-12 08:55:03 INFO OGG-00987 Oracle GoldenGate Command Interpreter for Oracle: GGSCI command (oracle): create datastore mmap.
2017-12-12 08:55:16 INFO OGG-00987 Oracle GoldenGate Command Interpreter for Oracle: GGSCI command (oracle): create datastore mmap.
...
...
2017-12-12 06:28:10 WARNING OGG-06530 Oracle GoldenGate Command Interpreter for Oracle: Berkeley Database encountered a critical error.
2017-12-12 06:29:39 INFO OGG-00987 Oracle GoldenGate Command Interpreter for Oracle: GGSCI command (oracle): send manager restartstatus.
2017-12-12 06:29:40 WARNING OGG-06530 Oracle GoldenGate Command Interpreter for Oracle: Berkeley Database encountered a critical error.

ggsci:
GGSCI 1> send manager restartstatus
Sending RESTARTSTATUS request to MANAGER ...
ERROR: sending message to MANAGER (Timeout waiting for message).

Changes

Problem is happening when ZFS+Cluster VIP+CRS 

A non-CRS installation that is not using a VIP works fine on ZFS.

Also there is no issue on regular File system or DBFS.

Cause