[
https://issues.apache.org/jira/browse/HADOOP-14138?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15976768#comment-15976768
]
Steve Loughran commented on HADOOP-14138:
-----------------------------------------
those JIRAs are so old they are implicitly dead.
right now things are, well, confused, as we have both code and XML defaults.
The XML ones win. Benefit: easy place to see what is there, one single spot to
track down the declarations without having access to the source
price:
* the defaults in the source files need to be kept in sync, when they aren't
they diverge and people loading in the configs without the defaults get
undefined outcomes
* sometimes those defaults are actually broken (S3AOutputStream temp dir setup
prior to HADOOP-13560), and nobody notices
* if someone has mistyped an entry (HADOOP-12735) then its not valid and nobody
knows why.
Personally, I like the XML file for its one-stop shop. But I don't like the
maintenance/sync problem
Thinking about it, maybe the way to go is an IDL mech which generates the XML,
the java constant names and the default value strings. That way: one single
thing to work off.
Either XML -> XSLT -> .java constants file, or see if we can't use protoc,
avroc or even, dare I say it, CORBA and the {{idlc}} compiler, to go from a
.idl declaration to java code. Though none of the protoc/avroc/idlj workflows
would generate the core-default.xml file. Our own XML file format, adding in
name of generated constant, maybe desired value type if not String, could be
used via <xslt> to generate everything. It could be done, though it'd need XSL
skills
> Remove S3A ref from META-INF service discovery, rely on existing core-default
> entry
> -----------------------------------------------------------------------------------
>
> Key: HADOOP-14138
> URL: https://issues.apache.org/jira/browse/HADOOP-14138
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 2.9.0
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Priority: Critical
> Fix For: 2.8.0, 2.7.4, 3.0.0-alpha3
>
> Attachments: HADOOP-14138.001.patch, HADOOP-14138-branch-2-001.patch
>
>
> As discussed in HADOOP-14132, the shaded AWS library is killing performance
> starting all hadoop operations, due to classloading on FS service discovery.
> This is despite the fact that there is an entry for fs.s3a.impl in
> core-default.xml, *we don't need service discovery here*
> Proposed:
> # cut the entry from
> {{/hadoop-aws/src/main/resources/META-INF/services/org.apache.hadoop.fs.FileSystem}}
> # when HADOOP-14132 is in, move to that, including declaring an XML file
> exclusively for s3a entries
> I want this one in first as its a major performance regression, and one we
> coula actually backport to 2.7.x, just to improve load time slightly there too
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
