After 2 months running without problems, my AWS DMS task failed with the following error:
Last failure message
Last Error Stream Component Fatal error. Task error notification received from subtask 0, thread 0 [reptask/replicationtask.c:2880] [1020101] Error executing source loop; Stream component failed at subtask 0, component; Stream component terminated [reptask/replicationtask.c:2888] [1020101] Stop Reason FATAL_ERROR Error Level FATAL
In my CloudWatch logs I found: NativeError: 1 Message: ERROR: could not access file "pglogical": No such file or directory; Error while executing the query
and later ERROR: relation "pglogical.replication_set" does not exist; No query has been executed with that handle
.
I noticed I didn't enable pglogical in my postgresql.conf
so I updated shared_preload_libraries
:
shared_preload_libraries = 'pglogical'
Postgres doesn't start with that setting, just fails. At the same time, I think that there might be another reason why the DMS task fails every time I'm trying to start it, as previously I didn't have that problem.
What else can I try to resolve this issue?
Here are some of the settings in my postgres.conf
wal_level = logical
max_wal_senders = 10 # max number of walsender processes
wal_sender_timeout = 0 # in milliseconds; 0 disables
max_replication_slots = 10 # max number of replication slots
Source: PostgreSQL 13.4 (Ubuntu 13.4-1.pgdg18.04+1) Target: Aurora PostgreSQL 13.7 DMS task type: Full load, ongoing replication