Skip stacktrace in info log for s3 region loading issue

[ywangd]

The stacktrace is rather excessive given the level of the log message. The information provided in the exception message should be sufficient for taking actions. Also adjust the wording to better match the contribuation guide for logging messages.

[elasticsearchmachine]

Pinging @elastic/es-distributed-coordination (Team:Distributed Coordination)

[DaveCTurner]

Hmm I think we need all the inner detail sometimes. But you're right that it'd be best not to emit this when starting the node since many nodes will not need this detail. Instead, I would rather we captured the exception here and then only emit the log message (once per node run) if org.elasticsearch.repositories.s3.S3Service#getClientRegion falls through to trying to use the default region.

[ywangd]

The stacktrace is something like the followings

[2025-07-18T11:14:02,464][INFO ][o.e.r.s.S3RepositoryPlugin][testMultiProjectSnapshots] failed to obtain region from default provider chain
software.amazon.awssdk.core.exception.SdkClientException: Unable to load region from any of the providers in the chain software.amazon.awssdk.regions.providers.DefaultAwsRegionProviderChain@64e725ed: [software.amazon.awssdk.regions.providers.SystemSettingsRegionProvider@8b63edf: Unable to load region from system settings. Region must be specified either via environment variable (AWS_REGION) or  system property (aws.region)., software.amazon.awssdk.regions.providers.AwsProfileRegionProvider@5895fb34: No region provided in profile: default, software.amazon.awssdk.regions.providers.InstanceProfileRegionProvider@34273343: Unable to retrieve region information from EC2 Metadata service. Please make sure the application is running on EC2.]
	at software.amazon.awssdk.core.exception.SdkClientException$BuilderImpl.build(SdkClientException.java:130) ~[sdk-core-2.31.78.jar:?]
	at software.amazon.awssdk.regions.providers.AwsRegionProviderChain.getRegion(AwsRegionProviderChain.java:70) ~[regions-2.31.78.jar:?]
	at org.elasticsearch.repositories.s3.S3RepositoryPlugin.getDefaultRegion(S3RepositoryPlugin.java:102) ~[repository-s3-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
	at org.elasticsearch.repositories.s3.S3Service.lambda$new$0(S3Service.java:132) ~[repository-s3-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
	at org.elasticsearch.common.util.concurrent.RunOnce.run(RunOnce.java:41) ~[elasticsearch-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
	at org.elasticsearch.repositories.s3.S3Service.doStart(S3Service.java:418) ~[repository-s3-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
	at org.elasticsearch.common.component.AbstractLifecycleComponent.start(AbstractLifecycleComponent.java:51) ~[elasticsearch-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
	at java.lang.Iterable.forEach(Iterable.java:75) ~[?:?]
	at org.elasticsearch.node.Node.start(Node.java:278) ~[elasticsearch-9.2.0-SNAPSHOT.jar:9.2.0-SNAPSHOT]
...

I don't see how they are useful other than the part that is preserved in this PR, i.e.:

[2025-07-18T11:14:02,464][INFO ][o.e.r.s.S3RepositoryPlugin][testMultiProjectSnapshots] unable to obtain region from default provider chain: Unable to load region from any of the providers in the chain software.amazon.awssdk.regions.providers.DefaultAwsRegionProviderChain@64e725ed: [software.amazon.awssdk.regions.providers.SystemSettingsRegionProvider@8b63edf: Unable to load region from system settings. Region must be specified either via environment variable (AWS_REGION) or  system property (aws.region)., software.amazon.awssdk.regions.providers.AwsProfileRegionProvider@5895fb34: No region provided in profile: default, software.amazon.awssdk.regions.providers.InstanceProfileRegionProvider@34273343: Unable to retrieve region information from EC2 Metadata service. Please make sure the application is running on EC2.]

Info logs are meant to be read by end-users. I'd think we don't want to burden them with the stacktrace which does not really add more value for them taking actions?

[DaveCTurner]

I suspect there may sometimes be more detail in some inner exceptions.

Info logs are meant to be read by end-users. I'd think we don't want to burden them with the stacktrace which does not really add more value for them taking actions?

Exactly, that's why I think we shouldn't log anything unless there's actually some action to take, which would only be because S3Service#getClientRegion fell through to the bottom.

ETA: I opened ES-12449 to track this work.