Primeiro precisamos definir a variável de ambiente NLS_DATE_FORMAT no padrão solicitado pela Oracle:

export NLS_DATE_FORMAT='DD-MON-YYYY HH24MISS'

Agora podemos reproduzir o erro com o RMAN em modo debug, também passando o log para um arquivo de saída:

rman debug all trace=srdc_rman_debug_09-FEB-2021.trc log=srdc_rman_output_09-FEB-2021.log

Exemplo tentando executar recover em um datafile:

RMAN> recover datafile 37;
RMAN> exit;

O diretório atual é /home/oracle, então os arquivos .trc e .log foram gerados nesse diretório.

Agora podemos gerar uma coleta com TFA, informando o srdc “dbrman”:

tfactl diagcollect -srdc dbrman

O TFA solicita o nome do banco de dados que será analisado, assim como se foram gerados os arquivos .trc e .log ao reproduzir o problema com o RMAN. Informamos “Y” e passamos os caminhos absolutos dos respectivos arquivos:

Enter the Database Name [Required for this SRDC] : PRD
 Have you executed your rman command in debug mode? [Y|N]  [Required for this SRDC]: Y
 Provide the full path and name of your rman debug trace file:  [Required for this SRDC]: /home/oracle/srdc_rman_debug_09-FEB-2021.trc
 Provide the full path and name of your rman output file:  [Required for this SRDC]: /home/oracle/srdc_rman_output_09-FEB-2021.log

Se tiver ocorrência, o TFA identifica e solicita confirmação para incluir a coleta referente a erros internos no banco de dados, basta confirmar:

 Feb/09/2021 09:39:03 : [prd_gru12n] ORA-00600: internal error code, arguments: [723], [142008], [8749872], [memory leak], [], [], [], [], [], [], [], [] 
 Please choose the event : 1-1 [1]
 Selected value is : 1 ( Feb/09/2021 09:39:03 )
 Scripts to be run by this srdc:  cp_rman_log cp_rman_trace srdc_rman_08137_info.sql srdc_rman_CDBinfo.sql srdc_rman_dbinfo.sql srdc_fra_details.sql srdc_rman_diag
 Components included in this srdc: OS CRS DATABASE NOCHMOS
 Collecting data for local node(s)
 Scanning files from Feb/09/2021 03:39:03 to Feb/09/2021 15:39:03
 WARNING: End time entered is after the current system time.
 Collection Id : 20210209102455dbprd
 Detailed Logging at : /u01/app/oracle.ahf/data/repository/srdc_dbrman_collection_Tue_Feb_09_10_24_56_03_2021_node_local/diagcollect_20210209102455_dbprd.log
 2021/02/09 10:24:59 BRT : NOTE : Any file or directory name containing the string .com will be renamed to replace .com with dotcom
 2021/02/09 10:24:59 BRT : Collection Name : tfa_srdc_dbrman_Tue_Feb_09_10_24_56_-03_2021.zip
 2021/02/09 10:25:00 BRT : Scanning of files for Collection in progress…
 2021/02/09 10:25:00 BRT : Collecting additional diagnostic information…
 2021/02/09 10:25:35 BRT : Getting list of files satisfying time range [02/09/2021 03:39:03 BRT, 02/09/2021 10:25:00 BRT]
 2021/02/09 10:25:44 BRT : Completed collection of additional diagnostic information…
 2021/02/09 10:26:56 BRT : Collecting ADR incident files…
 2021/02/09 10:26:56 BRT : Completed Local Collection
 .-------------------------------------.
 |          Collection Summary         |
 +-----------+-----------+------+------+
 | Host      | Status    | Size | Time |
 +-----------+-----------+------+------+
 | dbprd | Completed | 50MB | 116s |
 '-----------+-----------+------+------'
 Logs are being collected to: /u01/app/oracle.ahf/data/repository/srdc_dbrman_collection_Tue_Feb_09_10_24_56_03_2021_node_local
 /u01/app/oracle.ahf/data/repository/srdc_dbrman_collection_Tue_Feb_09_10_24_56_03_2021_node_local/dbprd.tfa_srdc_dbrman_Tue_Feb_09_10_24_56_-03_2021.zip

O arquivo .zip gerado no final do processo pode ser anexado no chamado junto a Oracle.