Replication fails on timestamps with BC years

[jasobrown-rs]

Summary

When a timestamp with a BC year is inserted into the upstream database, replication fails and gets into a loop.

Steps to reproduce

CREATE TABLE t3(id int, ts timestamp);
INSERT INTO t3 values(3, '214-02-05 15:40 BC');

Then look at the readyset logs:

2024-03-25T19:32:40.771241Z ERROR replicators::noria_adapter: Will stop replicating a table due to table error table="public"."t3" error=Error during replication: WAL error: TimestampParseError
2024-03-25T19:32:40.771364Z  INFO replicators::noria_adapter: Removing table state from readyset table="public"."t3"
2024-03-25T19:32:40.774066Z DEBUG migrate: readyset_server::controller::sql: num_queries=8 named_queries=8 dry_run=false
2024-03-25T19:32:40.774232Z TRACE migrate: readyset_server::controller::sql: removing base table table_name=public.t3 dry_run=false
2024-03-25T19:32:40.774293Z DEBUG migrate: readyset_server::controller::sql::mir: Removing base node name=public.t3 dry_run=false
2024-03-25T19:32:40.774485Z DEBUG migrate: readyset_server::controller::sql: Adding non-replicated relation with TableDropped name=public.t3 dry_run=false
2024-03-25T19:32:40.784261Z  INFO replicators::table_filter: denying replication schema=public table=t3
2024-03-25T19:32:40.790451Z  WARN replicators::noria_adapter: Restarting adapter after error encountered error=Internal error: in replicators/src/postgres_connector/connector.rs:822:21: invariant failed: lsn == cur_pos.commit_lsn;
left = 0/270E800;
right = 0/270E6E0
2024-03-25T19:32:40.790538Z ERROR replicators: Error in replication, will retry after timeout error=Internal error: in replicators/src/postgres_connector/connector.rs:822:21: invariant failed: lsn == cur_pos.commit_lsn;
left = 0/270E800;
right = 0/270E6E0 timeout_sec=1

ReadySet version

Upstream DB type and version

Instance Details

Deployment Details

\[Docker | OSS K8s | OSS binary | RS-Cloud\]

OS Information

Logs