root page 3 of index "pg_class_oid_index" has level 0, expected 1
Hi, During major version upgrade from PostgreSQL v9.4 to v9.6 I facing following error: Performing Consistency Checks - Checking cluster versions ok connection to database failed: FATAL: root page 3 of index "pg_class_oid_index" has level 0, expected 1 could not connect to old postmaster started with the command: "/postgresql94/bin/pg_ctl" -w -l "pg_upgrade_server.log" -D "/data/directory/postgresql" -o "-p 5432 -b -c listen_addresses='' -c unix_socket_permissions=0700 -c unix_socket_directories='/home/postgresql_user'" start Failure, exiting Again, I was trying to take the pg_dumpall of the same database cluster(PostgreSQL v9.4) and it failed with error: pg_dump: [archiver (db)] connection to database "template1" failed: FATAL: root page 3 of index "pg_class_oid_index" has level 0, expected 1 pg_dumpall: pg_dump failed on database "template1", exiting Please suggest how should I proceed with major version upgrade in this case. Regards, Abhijit
PITR based recovery in a primary/standby cluster setup
Hi, We have primary and standby PostgreSQL cluster setup and also we have PITR enabled on it. To improve the recovery time we are thinking of recovering the database to both primary and standby at the same time. These are the steps we are following: 1. Restore the base backup to the both primary and standby server 2. Replay the WAL files on both primary and standby and once the recovery target is reached stop the servers 3. Start one of the server as Primary and other one as standby We have followed the above steps but when we are trying to start the servers as primary and standby we are having replication issues. Could you please suggest what should be done here so that we can recover the database in both primary as well as in standby server? We are using PostgrerSQL version: 9.6.12 and for PITR we are using WAL-E. Regards, Abhijit
