Hello,
I encountered a problem on replicas after the primary crashed for lack of disk space.
After the problem I had a constant flow of "invalid contrecord" logs and replication ceased working.
The only way I found to make it work again was to completely restart the replica.
The logs:
Jun 11 12:44:28 myhostreplica postgres[20302]: [7-1] 2019-06-11 14:44:28 CEST [20302-2] FATAL: could not receive data
fromWAL stream: SSL SYSCALL error: EOF detected
Jun 11 12:44:28 myhostreplica postgres[19166]: [6-1] 2019-06-11 14:44:28 CEST [19166-1] FATAL: could not connect to
theprimary server: FATAL: the database system is in recovery mode
Jun 11 12:44:33 myhostreplica postgres[19167]: [6-1] 2019-06-11 14:44:33 CEST [19167-1] LOG: started streaming WAL
fromprimary at 2C/6A000000 on timeline 1
Jun 11 12:44:33 myhostreplica postgres[20298]: [10-1] 2019-06-11 14:44:33 CEST [20298-6] LOG: invalid contrecord
length4571 at 2C/69FFF648
Jun 11 12:44:33 myhostreplica postgres[19167]: [7-1] 2019-06-11 14:44:33 CEST [19167-2] FATAL: terminating walreceiver
processdue to administrator command
Jun 11 12:44:33 myhostreplica postgres[20298]: [11-1] 2019-06-11 14:44:33 CEST [20298-7] LOG: invalid contrecord
length4571 at 2C/69FFF648
Jun 11 12:44:33 myhostreplica postgres[20298]: [12-1] 2019-06-11 14:44:33 CEST [20298-8] LOG: invalid contrecord
length4571 at 2C/69FFF648
François J.
