github-actions[bot] commented on code in PR #65571:
URL: https://github.com/apache/doris/pull/65571#discussion_r3576125792
##########
fe/fe-core/src/main/java/org/apache/doris/httpv2/rest/LoadAction.java:
##########
@@ -367,7 +367,12 @@ private final synchronized int
getLastSelectedBackendIndexAndUpdate() {
}
private String getCloudClusterName(HttpServletRequest request) {
- String cloudClusterName =
request.getHeader(SessionVariable.CLOUD_CLUSTER);
+ String cloudClusterName =
request.getHeader(SessionVariable.COMPUTE_GROUP);
Review Comment:
Here `compute_group` is honored for FE redirect selection for every
`selectRedirectBackend` caller, including `/api/_http_stream`
(`streamLoadWithSql` calls it before redirecting). The target BE for that
endpoint is `HttpStreamAction`, not `StreamLoadAction`:
`HttpStreamAction::process_put` builds `TStreamLoadPutRequest` with `load_sql`,
auth, label, group commit, and backend id, but never copies
`HTTP_COMPUTE_GROUP` or `HTTP_CLOUD_CLUSTER` into `cloud_cluster`. Normal
stream load does that mapping before calling FE. So a SQL stream load with
`compute_group: cg_b` can be redirected to a BE in `cg_b`, then the final FE
planning RPC receives an empty cluster and falls back to the user's
default/policy cluster. Please either exclude `_http_stream` from this new
header path or pass the header through `HttpStreamAction` the same way normal
stream load does.
--
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]
