Gargi Jaiswal created HDDS-17141:
------------------------------------
Summary: S3 Gateway Range header parsing returns wrong status for
unsatisfiable and malformed ranges
Key: HDDS-17141
URL: https://issues.apache.org/jira/browse/HDDS-17141
Project: Apache Ozone
Issue Type: Task
Reporter: Gargi Jaiswal
Assignee: Gargi Jaiswal
{color:#00875a}RangeHeaderParserUtil#parseRangeHeader{color} mishandles some
edge-case Range values. Ozone returns 200 or 206 where AWS S3 returns 416
InvalidRange, and in two cases sends a malformed Content-Range header. The
parser is shared, so this affects *GetObject* today and *HeadObject* once
[HDDS-14224|https://issues.apache.org/jira/browse/HDDS-14224] is merged.
*Examples -> suppose 10-byte object is created*
{code:java}
|| Range || Ozone || AWS S3 ||
| bytes=10- | 200, full object | 416 InvalidRange |
| bytes=-0 | 206, Content-Range: bytes 10-9/10 | 416 InvalidRange |
| bytes=-5 (empty object) | 206, Content-Range: bytes 0--1/0 | 416 InvalidRange
|
| bytes=- | 206, Content-Range: bytes 1-9/10 (drops byte 0) | 200, full
object (header ignored) |{code}
For comparison, bytes=10-20 already returns 416 correctly.
*Reason*
An empty end is replaced with length-1 before the "end beyond object" check, so
bytes=10- is never marked invalid. Suffix ranges are not checked for a zero
suffix length, a zero-length object, or a missing suffix value.
*- Found while reviewing HDDS-14224 (PR
https://github.com/apache/ozone/pull/11612)*
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
