Russell Sayers created KAFKA-10685:
--------------------------------------
Summary: --to-datetime passed to kafka-consumer-groups getting
interpreted as a timezone
Key: KAFKA-10685
URL: https://issues.apache.org/jira/browse/KAFKA-10685
Project: Kafka
Issue Type: Bug
Affects Versions: 2.7.0
Reporter: Russell Sayers
If you pass more than 3 decimal places for the fractional seconds of the
datetime, the microseconds get interpreted as milliseconds.
{{kafka-consumer-groups --bootstrap-server kafka:9092 \}}
{{ --reset-offsets \}}
{{ --group webserver-avro \}}
{{ --topic driver-positions-avro \}}
{{ --to-datetime "2020-11-05T00:46:48.002237400" \}}
{{ --dry-run}}
Relevant code
[here|https://github.com/apache/kafka/blob/2.7/clients/src/main/java/org/apache/kafka/common/utils/Utils.java#L1304].
Experimenting with getDateTime:
* getDateTime("2020-11-05T00:46:48.000") -> 1604537208000
* getDateTime("2020-11-05T00:46:48.000+0800") -> 1604508408000 - correct the
formatting string allows for ZZZ timezones
* getDateTime("2020-11-05T00:46:48.000123") -> 1604537208123 - note this ends
with 123 milliseconds.
The pattern string is "yyyy-MM-dd'T'HH:mm:ss.SSS". So SimpleDateFormat
interprets "000123" as 123 milliseconds. See the stackoverflow answer
[here|https://stackoverflow.com/a/21235602/109102].
The fix? Remove any digits after more than 3 characters after the decimal
point, or raise an exception. The code would still need to allow the RFC822
timezone, i.e Sign TwoDigitHours Minutes.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
