nastra commented on code in PR #14767:
URL: https://github.com/apache/iceberg/pull/14767#discussion_r2675210808
##########
core/src/main/java/org/apache/iceberg/rest/RESTTableScan.java:
##########
@@ -151,6 +175,21 @@ private CloseableIterable<FileScanTask>
planTableScan(PlanTableScanRequest planT
this.planId = response.planId();
PlanStatus planStatus = response.planStatus();
+ List<Credential> storageCredentials = response.credentials();
Review Comment:
I think what matters to a server is to know in what context the credentials
were initially vended. If the credentials for the table's FileIO were vended
when loading the table and those credentials are being refreshed, then the
server knows that those are "normally" vended credentials. With remote scan
planning we want to give the server the additional context by adding the planId
when the creds were actually vended for remote scan planning. When those creds
are being refreshed, the server knows that those might be "special" vended
creds. This is because a server might decide to implement remote scan planning
in a way where it stores data in a temp location that it vends credentials for
and refreshing those creds is different from refreshing "normal" table creds.
Hence why I think we shouldn't be sending the planId when we use the table's
FileIO
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
