18.3 GI is installed successfully. +ASMn is run on every node and -MGMTDB is on one node. Each node has two private networks, ens224 and ens256. When running dbca to create a database, it always crashed toward the end with ORA-03113, leaving one instance up on one node instead of each node having its instance running. Then I disabled one private interface with `oifcfg delif'. Dbca finally created the database. Once I added the second private network back, the database crashed again except for one instance on one node.
The lmon trace file shows
doprlcora13a trace $ grep "Local Address" dbcp131_lmon_15989.trc
IPCLW:[0.4]{-}[WAIT]:UTIL: [1557953522748780] Local Address: 169.254.22.194:61986 Remote Address: 169.254.25.141:10880
This is strange because the local address is on one private network while the remote address is on the other (ens224 trying to talk to ens256 remotely). These two network interfaces obviously do not cross-talk. What could be the reason the remote IP is not on the same interface, which *is* pingable from the local node when I specify source interface (-I option of `ping')? According to
Only One Instance of a RAC Database Can Start at a Time: Second Instance Fails to Start due to "No reconfig messages from other instances" - LMON is terminating the instance (Doc ID 2528588.1)
a firewall between the nodes would cause this problem. We double-checked. We don't have a firewall.
We have some 11gR2 RAC databases up and running where each node also has two private networks. Ping'ing from one source interface to the IP of the other interface on a remote node does not work either, as expected. The old version Oracle probably does not provide the "IPCLW local address - remote address" info in the lmon trace. But if it were to, I'm sure the two addresses from the same network would be shown.