We have a very large database of 37TB and we had to promote our standby to primary due to some disk failures. Now the issues are solved, we would like to make standby the old primary from a copy of the new primary which is already in place. Is it possible without using pg_basebackup? We have already a copy created than can be started but if we open as standby (standby.signal created) the log says when started:
[2022-10-28 21:27:12 CEST-]LOG: starting PostgreSQL 13.3 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 8.5.0 20210514 (Red Hat 8.5.0-10), 64-bit
[2022-10-28 21:27:12 CEST-]LOG: listening on IPv4 address "0.0.0.0", port 8300
[2022-10-28 21:27:12 CEST-]LOG: could not create IPv6 socket for address "::": Address family not supported by protocol
[2022-10-28 21:27:12 CEST-]LOG: listening on Unix socket "/tmp/.s.PGSQL.8300"
[2022-10-28 21:27:12 CEST-]LOG: database system was interrupted; last known up at 2022-10-28 20:31:08 CEST
[2022-10-28 21:28:06 CEST-]LOG: restored log file "00000003.history" from archive
cp: cannot stat '/PostgresWalLogArchive/new/00000004.history': No such file or directory
[2022-10-28 21:28:06 CEST-]LOG: entering standby mode
[2022-10-28 21:28:06 CEST-]LOG: restored log file "00000003.history" from archive
cp: cannot stat '/PostgresWalLogArchive/new/000000030000BF720000002F': No such file or directory
[2022-10-28 21:28:06 CEST-]LOG: restored log file "000000020000BF720000002F" from archive
[2022-10-28 21:28:06 CEST-]FATAL: requested timeline 3 is not a child of this server's history
[2022-10-28 21:28:06 CEST-]DETAIL: Latest checkpoint is at BF72/2F0309C0 on timeline 2, but in the history of the requested timeline, the server forked off from that timeline at B1D6/60000000.
[2022-10-28 21:28:06 CEST-]LOG: startup process (PID 1298266) exited with exit code 1
[2022-10-28 21:28:06 CEST-]LOG: aborting startup due to startup process failure
[2022-10-28 21:28:07 CEST-]LOG: database system is shut down
Do you know if we can make it work without having to run pg_basebackup for 37TB which will take about 3 days to copy? Why it is requesting a different timeline if it is a copy of the primary?
Primary replica parameters:
wal_level = replica
archive_mode = on
archive_command = 'cp -i %p /PostgresWalLogArchive/new/%f'
standby replica parameters:
primary_conninfo = 'host=XXX port=8300 user=postgres password=***'
restore_command = 'cp /PostgresWalLogArchive/new/%f %p'
archive_cleanup_command = '/home/postgres/software/postgresql/bin/pg_archivecleanup /PostgresWalLogArchive/new/%f %r'
Thank you very much for your help in advance.
Pilar de Teodoro
