I was performing a database upgrade from 11.2 to 12.2 when my VPN crashed.
I realized that I forgot to start the upgrade process using linux screen terminal which means that my upgrade process was lost.
Well, Oracle 12.2 has the ability to resume a failed upgrade process from the failed step automatically !
From the Oracle Docs:
Oracle Database 12c release 2 (12.2) includes a new Resume option for Parallel Upgrade Utility. This option is available for both CDBs and Non-CDBs. You are not required to identify failed or incomplete phases when you rerun or restart the upgrade. When you use the Parallel Upgrade Utility using the resume option (-R), the utility automatically detects phases from the previous upgrade that are not completed successfully. The Parallel Upgrade Utility then reruns or restarts just these phases that did not complete successfully, so that the upgrade is completed. Bypassing steps that already completed successfully reduces the amount of time it takes to rerun the upgrade.
So I just ran $ORACLE_HOME/bin/dbupgrade -n 4 -R -l $ORACLE_HOME/diagnostics and the upgrade process was restarted.
I was following the steps from MOS note 12.2: How to Create GI Management Repository (Doc ID 2246123.1) and it failed with the error below:
[ 2019-04-02 10:15:05.945 BRT ] Registering database with Oracle Grid Infrastructure [ 2019-04-02 10:15:06.320 BRT ] PRCR-1006 : Failed to add resource ora.mgmtdb for mgmtdb PRCR-1071 : Failed to register or update resource ora.mgmtdb CRS-2510: Resource ‘ora.MGMTLSNR’ used in dependency ‘hard’ does not exist or is not registered. CRS-2514: Dependency attribute specification ‘hard’ is invalid in resource ‘ora.mgmtdb’ [ 2019-04-02 10:15:11.715 BRT ] DBCA_PROGRESS : DBCA Operation failed.
I could not found an exact match @MOS but to workaround this issue you can run srvctl add mgmtlsnr as grid user before running dbca and after that it will work properly.