[ https://issues.apache.org/jira/browse/HDDS-791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16674012#comment-16674012 ]
Steve Loughran commented on HDDS-791: ------------------------------------- make sure that 0 byte ranges are supported, even on a 0-byte file. Source of pain. S3A client execute ranged gets when a file is opened in fadvise=random, or when it detects random IO. The benefit comes in being able to seek() fast to a new offset, and for positioned readable to explicitly set the range to read. In both cases, the HTTP connection can be reused: no need to either read to the end of the file (a real perf killer in production which doesn't show up in small-file tests), or two abort the TCP stream and renegotiate a new one > Support Range header for ozone s3 object download > ------------------------------------------------- > > Key: HDDS-791 > URL: https://issues.apache.org/jira/browse/HDDS-791 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task > Components: S3 > Reporter: Elek, Marton > Priority: Major > > Using s3 rest api smaller chunks of an object could be uploaded with using > Range headers: > For example: > {code} > GET /example-object HTTP/1.1 > Host: example-bucket.s3.amazonaws.com > x-amz-date: Fri, 28 Jan 2011 21:32:02 GMT > Range: bytes=0-9 > Authorization: AWS AKIAIOSFODNN7EXAMPLE:Yxg83MZaEgh3OZ3l0rLo5RTX11o= > Sample Response with Specified Range of the Object Bytes > {code} > Can be implemented with using the seek method on OzoneInputStream. > The Range header support is one of the missing piece for fully support s3a > interface. > References: > Range header spec: > https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.35 > Aws s3 doc: > https://docs.aws.amazon.com/AmazonS3/latest/API/RESTObjectGET.html -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org