This is an automated email from the ASF dual-hosted git repository.

yiguolei pushed a commit to branch branch-4.1
in repository https://gitbox.apache.org/repos/asf/doris.git


The following commit(s) were added to refs/heads/branch-4.1 by this push:
     new 95f0eb79be1 [fix](regression) keep variant_doc_mode_p2 load in the p2 
group with its queries (#64533)
95f0eb79be1 is described below

commit 95f0eb79be14bc330496e386504aa9ed1dbce651
Author: shuke <[email protected]>
AuthorDate: Wed Jun 17 12:56:22 2026 +0800

    [fix](regression) keep variant_doc_mode_p2 load in the p2 group with its 
queries (#64533)
    
    ### What problem does this PR solve?
    
    In `regression_test_variant_doc_mode_p2`, `load.groovy` (which creates
    and loads the `github_events` table) is tagged `nonConcurrent,p2`, while
    the 76 `sql/*.sql` query cases in the same directory fall back to the
    path-derived default group `variant_doc_mode_p2` (no `nonConcurrent`).
    
    `nonConcurrent` is a **separate pipeline**, so the load step and the
    queries get routed to different runs/clusters. In the P2 run the load
    never executes, and every query fails instantly with:
    
    ```
    Table [github_events] does not exist in database 
[regression_test_variant_doc_mode_p2]
    ```
    
    This produced **66 cascade failures in a single branch-4.1 P2 build**
    (e.g. build #35 / 198091), all from the same root cause.
    
    ### Root cause
    
    The `nonConcurrent` tag was applied only to `load.groovy`, not to the
    dependent queries. The correct nonConcurrent sibling
    (`variant_github_events_nonConcurrent_p2`) encodes it in the **directory
    name** so the whole suite (load + queries) routes together;
    `variant_doc_mode_p2` instead split the data-prep away from its queries.
    
    The load only sets **session variables** and does no global-state
    mutation (no `set global` / FE-BE config / debug points), so it does not
    actually need `nonConcurrent` isolation.
    
    ### Fix
    
    Drop `nonConcurrent` so `load.groovy` shares the `p2` group/pipeline
    with its queries. `load.groovy` sorts before `sql/`, runs first, and
    `github_events` is present when the queries execute.
    
    ```diff
    -suite("test_doc_value_p2", "nonConcurrent,p2"){
    +suite("test_doc_value_p2", "p2"){
    ```
    
    > Note: the identical structure exists on master (`#59183`); a follow-up
    there is advisable.
    
    ### Release note
    
    None
    
    🤖 Generated with [Claude Code](https://claude.com/claude-code)
    
    Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
---
 regression-test/suites/variant_doc_mode_p2/load.groovy | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/regression-test/suites/variant_doc_mode_p2/load.groovy 
b/regression-test/suites/variant_doc_mode_p2/load.groovy
index e9a926f380e..03bd0e3f4cf 100644
--- a/regression-test/suites/variant_doc_mode_p2/load.groovy
+++ b/regression-test/suites/variant_doc_mode_p2/load.groovy
@@ -17,7 +17,7 @@
 
 import org.codehaus.groovy.runtime.IOGroovyMethods
 
-suite("test_doc_value_p2", "nonConcurrent,p2"){
+suite("test_doc_value_p2", "p2"){
     sql """ set default_variant_enable_doc_mode = true """
     boolean random = new Random().nextBoolean()
     if (random) {


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to