For example, we have seen the connector crash with an input timestamp value of -6795364578871000. The error is:
[row index 0] (location value.end_date, reason: invalid): Could not parse '1754-08-30 22:43:41.-871000' as a timestamp. Required format is YYYY-MM-DD HH:MM[:SS[.SSSSSS]]
As you can see, a negative fractional second is provided in a way that makes no sense and thus can't be parsed.
The expected date/time would be Friday, August 30, 1754 10:43:41.129 PM. So we should see the fractional second portion be .129000, not .-871000.
This timestamp conversion code does not account for timestamps with negative epochs: https://github.com/confluentinc/kafka-connect-bigquery/blob/c9aedfb47b5fb61ec67404adc4f14cf05a51d82b/kcbq-connector/src/main/java/com/wepay/kafka/connect/bigquery/convert/logicaltype/DebeziumLogicalConverters.java#L134-L136
For example, we have seen the connector crash with an input timestamp value of -6795364578871000. The error is:
As you can see, a negative fractional second is provided in a way that makes no sense and thus can't be parsed.
The expected date/time would be Friday, August 30, 1754 10:43:41.129 PM. So we should see the fractional second portion be
.129000
, not.-871000
.