[CQLReplicator on Glue] Handling of replication write error

[hungfaileung]

If during the replication, the writes to the target Amazon Keyspaces table failed for some records for some reason, how would cql-replicator handles that?

Example error:

• https://repost.aws/knowledge-center/keyspaces-http-5xx-errors

[nwheeler81]

@hungfaileung the CQLReplicator retires with the exponential backoff strategy (256 retires with the based delay 25ms and 1.1 multiplier), but you should plan ahead the incoming traffic and pre-warm/provision target resources to avoid retires.

[hungfaileung]

@nwheeler81 Is the retry talking about Glue job or individual write to target table?

[nwheeler81]

@nwheeler81 Is the retry talking about Glue job or individual write to target table? first Individual retries and after Glue job retries (default 1, but you can increase it)

[hungfaileung]

@nwheeler81 Thanks