[
https://issues.apache.org/jira/browse/HADOOP-18908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17772268#comment-17772268
]
ASF GitHub Bot commented on HADOOP-18908:
-----------------------------------------
steveloughran commented on code in PR #6106:
URL: https://github.com/apache/hadoop/pull/6106#discussion_r1347608674
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -229,4 +254,49 @@ private static URI getS3Endpoint(String endpoint, final
Configuration conf) {
throw new IllegalArgumentException(e);
}
}
+
+ /**
+ * Parses the endpoint to get the region.
+ * If endpoint is the central one, use US_EAST_1.
+ *
+ * @param endpoint the configure endpoint.
+ * @return the S3 region.
+ */
+ private static Region getS3RegionFromEndpoint(String endpoint) {
+
+ if(!endpoint.endsWith(CENTRAL_ENDPOINT)) {
+ LOG.debug("Endpoint {} is not the default; parsing", endpoint);
+ return AwsHostNameUtils.parseSigningRegion(endpoint,
S3_SERVICE_NAME).orElse(null);
Review Comment:
weve had so much trouble in the past here with vpce, govcloud, cn that we
need a set of unit tests to verify this code does what we actually want -and
therea are no regressions
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -138,14 +157,21 @@ private <BuilderT extends S3BaseClientBuilder<BuilderT,
ClientT>, ClientT> Build
BuilderT builder, S3ClientCreationParameters parameters, Configuration
conf, String bucket)
throws IOException {
- Region region = parameters.getRegion();
- LOG.debug("Using region {}", region);
-
URI endpoint = getS3Endpoint(parameters.getEndpoint(), conf);
+ Region region = null;
+
if (endpoint != null) {
builder.endpointOverride(endpoint);
- LOG.debug("Using endpoint {}", endpoint);
+ region = getS3RegionFromEndpoint(parameters.getEndpoint());
+ LOG.debug("Using endpoint {} and region {}", endpoint, region);
+ }
+
+ if (region != null) {
+ builder.region(region);
+ } else {
+ builder.crossRegionAccessEnabled(true);
+ builder.region(getS3Region(conf));
Review Comment:
also, what about: env vars and sysprops? good or bad?
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -229,4 +254,49 @@ private static URI getS3Endpoint(String endpoint, final
Configuration conf) {
throw new IllegalArgumentException(e);
}
}
+
+ /**
+ * Parses the endpoint to get the region.
+ * If endpoint is the central one, use US_EAST_1.
+ *
+ * @param endpoint the configure endpoint.
+ * @return the S3 region.
Review Comment:
javadoc to say returns null if not resolvved
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -229,4 +254,49 @@ private static URI getS3Endpoint(String endpoint, final
Configuration conf) {
throw new IllegalArgumentException(e);
}
}
+
+ /**
+ * Parses the endpoint to get the region.
+ * If endpoint is the central one, use US_EAST_1.
+ *
+ * @param endpoint the configure endpoint.
+ * @return the S3 region.
+ */
+ private static Region getS3RegionFromEndpoint(String endpoint) {
+
+ if(!endpoint.endsWith(CENTRAL_ENDPOINT)) {
+ LOG.debug("Endpoint {} is not the default; parsing", endpoint);
+ return AwsHostNameUtils.parseSigningRegion(endpoint,
S3_SERVICE_NAME).orElse(null);
+ }
+
+ // endpoint is for US_EAST_1;
+ return Region.US_EAST_1;
+ }
+
+ /**
+ * Gets the region from configuration and returns.
+ * If configured region is an empty string, use
+ * the default SDK resolution chain.
+ *
+ * @param conf Configuration.
+ * @return the S3 region
+ */
+ private static Region getS3Region(Configuration conf) {
+ String region = conf.getTrimmed(AWS_REGION, AWS_S3_CENTRAL_REGION);
+ LOG.debug("fs.s3a.endpoint.region=\"{}\"", region);
+ if (!region.isEmpty()) {
+ // there's either an explicit region or we have fallen back
+ // to the central one.
+ LOG.debug("Setting region to {}", region);
+ return Region.of(region);
+ } else {
+ // no region.
+ // allow this if people really want it; it is OK to rely on this
+ // when deployed in EC2.
Review Comment:
if it is ok in ec2, then this will be very noisy in ec2 deployments?
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -138,14 +157,21 @@ private <BuilderT extends S3BaseClientBuilder<BuilderT,
ClientT>, ClientT> Build
BuilderT builder, S3ClientCreationParameters parameters, Configuration
conf, String bucket)
throws IOException {
- Region region = parameters.getRegion();
- LOG.debug("Using region {}", region);
-
URI endpoint = getS3Endpoint(parameters.getEndpoint(), conf);
+ Region region = null;
+
if (endpoint != null) {
builder.endpointOverride(endpoint);
- LOG.debug("Using endpoint {}", endpoint);
+ region = getS3RegionFromEndpoint(parameters.getEndpoint());
+ LOG.debug("Using endpoint {} and region {}", endpoint, region);
+ }
+
+ if (region != null) {
+ builder.region(region);
+ } else {
+ builder.crossRegionAccessEnabled(true);
+ builder.region(getS3Region(conf));
Review Comment:
if a region is set in the config, it must take priority over anything the
AWS SDK determines from the endpoint. without that we can never correct for sdk
issues
##########
hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/DefaultS3ClientFactory.java:
##########
@@ -138,14 +157,21 @@ private <BuilderT extends S3BaseClientBuilder<BuilderT,
ClientT>, ClientT> Build
BuilderT builder, S3ClientCreationParameters parameters, Configuration
conf, String bucket)
throws IOException {
- Region region = parameters.getRegion();
Review Comment:
restore the ability to pass in a region
> Improve s3a region handling, including determining from endpoint
> ----------------------------------------------------------------
>
> Key: HADOOP-18908
> URL: https://issues.apache.org/jira/browse/HADOOP-18908
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.4.0
> Reporter: Steve Loughran
> Assignee: Ahmar Suhail
> Priority: Major
> Labels: pull-request-available
>
> s3a now requires the fs.s3a.endpoint.region to be set; and while it can
> determine it from a network call, this takes time and doesn't work for third
> party stores.
> proposed
> * reinstate parsing of the fs.3a.endpoint url to automatically determine
> region from well known endoints (and vplink ones)
> * don't try to talk to AWS S3 if endpoint isn't an aws one: for that caller
> must declare (HADOOP-18673)
> * document this in v2 migration, including stack traces of falures
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
