You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@impala.apache.org by "Yida Wu (Code Review)" <ge...@cloudera.org> on 2025/07/21 23:28:00 UTC

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Yida Wu has uploaded this change for review. ( http://gerrit.cloudera.org:8080/23204


Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure caused by a version mismatch when
the admissiond cluster membership is recovering after a statestore
restart. During recovery, the new cluster membership version could
be lower than the version associated with a queued query, violating
the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 86 insertions(+), 0 deletions(-)



  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/1
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newchange
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 1
Gerrit-Owner: Yida Wu <wy...@gmail.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 4:

(9 comments)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       // to current membership and cleared.
             :       new_state = recovering_membership_;
             :       if (current_membership_ != nullptr
             :        
> I think this is possible if we get a full membership update which starts fr
Maybe I can file another jira for the timestamp based approach for the membership versioning.


http://gerrit.cloudera.org:8080/#/c/23204/3/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/3/be/src/scheduling/cluster-membership-mgr.cc@305
PS3, Line 305:       }
> I'm not quite understanding why the if statement is necessary.  Wouldn't it
It can have a loop stuff for the recovery process, in line 313 it will bump the version , "new_state->version += 1;", so i guess we want it to keep increasing during the recovery process but not smaller than the current one, then when it quits the recovery process, it will be overwritten to the current membership.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2238
PS3, Line 2238:     This sequence can trigger a DCHECK hit in AdmissionD.
> Any idea what percentage of the time the DCHECK fails?  If we know that num
From my experience, the DCHECK fails more than 50% of the time, often on the first run. Since the test can run for a few minutes and the bug isn’t critical, it may not be worth the added time to run it multiple times. A single run should be sufficient to catch the issue in builds.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2243
PS3, Line 2243: select count(*) from functional.alltypes limit 1
> A coordinator-only query such as "select 1" would be better to eliminate an
"select 1" is a trivial query, can skip the admissiond. I think it might be better to use a bit complex short query, which could be better to ensure the cluster is working as expected after all the things.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2254
PS3, Line 2254: create_hs2_client()
> Nit:  Beeswax is deprecated, use HS2 if possible.  Also in line 2259.
Done


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2271
PS3, Line 2271:     statestore.wait_for_exit()
> Use statestore.wait_for_exit() instead of sleep
Done


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2274
PS3, Line 2274:     coord1.wait_for_exit()
> Use coord1.wait_for_exit() instead of sleep
Done


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2278
PS3, Line 2278:     handle3 = client2.execute_async(short_query)
> Does this query need a slight delay to have a better chance of failing the 
I think the current step is intentional, the error typically occurs when the new query and the statestore happen to start around the same time. Not sure if AC_BEFORE_ADMISSION would help to repro, but since the DCHECK fails more than 50% of the time, maybe it is good enough to catch it?


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2305
PS3, Line 2305:     except Exception:
> Why the need to restart coord1?  This test class appears to be restarting t
It might fail the teardown check earlier, but seems adding the closing handle can help. Removed the coord start stuff.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 4
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Mon, 04 Aug 2025 20:59:07 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Abhishek Rawat (Code Review)" <ge...@cloudera.org>.
Abhishek Rawat has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 2:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@297
PS2, Line 297: // The recovering membership is never exposed to clients and therefore requires no
             :       // copying.
The recovering snapshot could be used to update current snapshot and as a result used for admitting queries and notifying clients once statestore subscriber's post-recovery period has passed. I think this comment is misleading. Do we need to make deep copy?

https://github.com/apache/impala/blob/branch-4.5.1/be/src/scheduling/cluster-membership-mgr.cc#L455

`ss_is_recovering` could be false if statestore subscriber has come out of post recovery period and `recovering_membership` could be non null.


http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       if (current_membership_ != nullptr
             :           && current_membership_->version > new_state->version) {
             :         new_state->version = current_membership_->version;
             :       }
It looks like the issue is that we're maintaining separate snapshots in recovering_membership_ and current_membership_ but we expect a common snapshot version across these.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 00:40:43 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Riza Suminto (Code Review)" <ge...@cloudera.org>.
Riza Suminto has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 5:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG@9
PS5, Line 9: This patch fixes a DCHECK failure
Mention where the DCHEK is. Is it AdmissionController::ComputeGroupScheduleStates?

Please elaborate the problem statement more. Why the DCHECK exist in the first place? Does Admission Control have strong requirement that cluster membership should not change between query queued and admitted?

Why failing-followed-by-retrying is not an option?



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 5
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 06 Aug 2025 21:26:43 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Riza Suminto (Code Review)" <ge...@cloudera.org>.
Riza Suminto has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

(3 comments)

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG@9
PS5, Line 9: This patch fixes a DCHECK failure
> Yes, the DCHECK is in AdmissionController::ComputeGroupScheduleState. Becau
Ack


http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2231
PS7, Line 2231: test_dcheck_issue_coord_statestore_kill_order
nit: This can use more clarifying test name. Maybe "test_cluster_membership_after_statestore_recovery" ?


http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2259
PS7, Line 2259:     client2 = coord2.service.create_hs2_client()
              :     handle2 = client2.execute_async(short_query)
              :     # Wait until the second query is confirmed to be queued.
              :     self._wait_for_change_to_profile(handle2, "Admission result: Queued", client=client2)
Have you try run this test again after rebase?
I did, and the test hang in this part, timing out for not getting queued.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 24 Sep 2025 17:49:11 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#6). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure in ComputeGroupScheduleStates() for
DCHECK_GE(current_membership_version, previous_membership_version)
that occurs during cluster membership recovery after a statestore
restart.

The DCHECK enforces the assumption that the cluster membership
version should be monotonically increasing. A new version that
is smaller than a version associated with a queued query is
considered an abnormal case, which this DCHECK is designed to
catch during development. However, during recovery, the new
cluster membership version could be lower than the version
associated with a queued query, violating the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 98 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/6
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 6
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 5:

Rebased.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 5
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 05 Aug 2025 17:41:25 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 5:

Build Successful 

https://jenkins.impala.io/job/gerrit-code-review-checks/20043/ : Initial code review checks passed. Use gerrit-verify-dryrun-external or gerrit-verify-dryrun to run full precommit tests.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 5
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 05 Aug 2025 18:06:42 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Abhishek Rawat (Code Review)" <ge...@cloudera.org>.
Abhishek Rawat has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 2:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@297
PS2, Line 297: // The recovering membership is never exposed to clients and therefore requires no
             :       // copying.
> Can we say that, "The recovering membership is not exposed to clients unles
I was more concerned about if we could try to dereference the recovering_membership_ shared_ptr after underlying snapshot has been removed?


http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       if (current_membership_ != nullptr
             :           && current_membership_->version > new_state->version) {
             :         new_state->version = current_membership_->version;
             :       }
> Yeah, one way could be removing the DCHECK in https://github.com/apache/imp
I think we should not remove the DCHECK. The ComputeGroupScheduleStates() could be called multiple times for a queued query and it's important for the current_snapshot version to be higher than previous ones so that the schedule state is based on current cluster membership.

I'm however not sure how we could get into situation where current_membership_->version could decrease?

We basicaly have:
recovering_membership_ copied from current_membership_
...
current_membership_ gets updated with recovering_membership_



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 19:59:49 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#2). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure caused by a version mismatch when
the admissiond cluster membership is recovering after a statestore
restart. During recovery, the new cluster membership version could
be lower than the version associated with a queued query, violating
the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 87 insertions(+), 0 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/2
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Abhishek Rawat (Code Review)" <ge...@cloudera.org>.
Abhishek Rawat has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 2:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       if (current_membership_ != nullptr
             :           && current_membership_->version > new_state->version) {
             :         new_state->version = current_membership_->version;
             :       }
> The version from recovering_membership_ starts at 0, and this seems to be t
I think this is possible if we get a full membership update which starts from version 0. I feel like the current solution doesn't work in all cases and we should probably use timestamps instead of versions which get reset every time there is a full membership update (say due to statestored restarting).



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 23:33:16 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Abhishek Rawat (Code Review)" <ge...@cloudera.org>.
Abhishek Rawat has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 2:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/23204/2/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/2/tests/custom_cluster/test_admission_controller.py@2296
PS2, Line 2296:         continue
Probably also update the test to actually change cluster membership snapshot i.e use different number of executors when query was queued vs when it actually executed.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Thu, 31 Jul 2025 23:58:19 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 1:

Build Failed 

https://jenkins.impala.io/job/gerrit-code-review-checks/19934/ : Initial code review checks failed. See linked job for details on the failure.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 1
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Comment-Date: Mon, 21 Jul 2025 23:55:08 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#7). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure in ComputeGroupScheduleStates() for
DCHECK_GE(current_membership_version, previous_membership_version)
that occurs during cluster membership recovery after a statestore
restart.

The DCHECK enforces the assumption that the cluster membership
version should be monotonically increasing. A new version that
is smaller than a version associated with a queued query is
considered an abnormal case, which this DCHECK is designed to
catch during development. However, during recovery, the new
cluster membership version could be lower than the version
associated with a queued query, violating the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 98 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/7
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Jason Fehr (Code Review)" <ge...@cloudera.org>.
Jason Fehr has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 3: Code-Review+1

(8 comments)

http://gerrit.cloudera.org:8080/#/c/23204/3/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/3/be/src/scheduling/cluster-membership-mgr.cc@305
PS3, Line 305:       }
I'm not quite understanding why the if statement is necessary.  Wouldn't it be better to always set new_state->version = current_membership_->version in a recovery scenario?  That will happen anyways after recovery has completed.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2238
PS3, Line 2238:     This sequence can trigger a DCHECK hit in AdmissionD.
Any idea what percentage of the time the DCHECK fails?  If we know that number, then we should loop this test that many times to be more confident the issue has been fixed.

For example, if the DCHECK fails 10% of the time, then loop this test 10 times.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2243
PS3, Line 2243: select count(*) from functional.alltypes limit 1
A coordinator-only query such as "select 1" would be better to eliminate any issues with executor communication.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2254
PS3, Line 2254: create_beeswax_client
Nit:  Beeswax is deprecated, use HS2 if possible.  Also in line 2259.


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2271
PS3, Line 2271:     sleep(2)
Use statestore.wait_for_exit() instead of sleep


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2274
PS3, Line 2274:     sleep(3)
Use coord1.wait_for_exit() instead of sleep


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2278
PS3, Line 2278:     handle3 = client2.execute_async(short_query)
Does this query need a slight delay to have a better chance of failing the DCHECK?  Possibly the "AC_BEFORE_ADMISSION" debug action could be used?


http://gerrit.cloudera.org:8080/#/c/23204/3/tests/custom_cluster/test_admission_controller.py@2305
PS3, Line 2305:     coord1.start()
Why the need to restart coord1?  This test class appears to be restarting the minicluster after each test, thus the next test run will take care of restarting the entire minicluster.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 3
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Thu, 31 Jul 2025 23:18:04 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Riza Suminto (Code Review)" <ge...@cloudera.org>.
Riza Suminto has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2254
PS7, Line 2254:     client1 = coord1.service.create_hs2_client()
I suggest to use with-as statement so you can skip closing clients manually at L2311-L2312 below.

    with coord1.service.create_hs2_client() as client1, \
      coord2.service.create_hs2_client() as client2:


http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2259
PS7, Line 2259:     client2 = coord2.service.create_hs2_client()
              :     handle2 = client2.execute_async(short_query)
              :     # Wait until the second query is confirmed to be queued.
              :     self._wait_for_change_to_profile(handle2, "Admission result: Queued", client=client2)
> Umm, I don't see any issue running the test locally or in the build with th
Test works after I rebase. It took 120s to complete though. Is there any flags that can be set to speed up the test a bit?



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Sun, 28 Sep 2025 03:06:54 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 1:

(1 comment)

gerrit-auto-critic failed. You can reproduce it locally using command:

  python3 bin/jenkins/critique-gerrit-review.py --dryrun

To run it, you might need a virtual env with Python3's venv installed.

http://gerrit.cloudera.org:8080/#/c/23204/1/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/1/be/src/scheduling/cluster-membership-mgr.cc@300
PS1, Line 300:       if (current_membership_ != nullptr && current_membership_->version > new_state->version) {
line too long (96 > 90)



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 1
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Comment-Date: Mon, 21 Jul 2025 23:29:49 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#5). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure caused by a version mismatch when
the admissiond cluster membership is recovering after a statestore
restart. During recovery, the new cluster membership version could
be lower than the version associated with a queued query, violating
the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 98 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/5
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 5
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Jason Fehr (Code Review)" <ge...@cloudera.org>.
Jason Fehr has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2249
PS7, Line 2249:     assert len(all_coords) >= 2, "Test requires at least two coordinators"
Modify parameters to the `@CustomClusterTestSuite.with_args` annotation instead of using this assert to enforce two coordinators.  That way, future changes to this annotation will not cause this test to fail.


http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2259
PS7, Line 2259:     client2 = coord2.service.create_hs2_client()
              :     handle2 = client2.execute_async(short_query)
              :     # Wait until the second query is confirmed to be queued.
              :     self._wait_for_change_to_profile(handle2, "Admission result: Queued", client=client2)
> Test works after I rebase. It took 120s to complete though. Is there any fl
May need to create custom fair-scheduler.xml and llama-site.xml files to decrease timeouts.



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Thu, 16 Oct 2025 07:29:00 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 6:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       // to current membership and cleared.
             :       new_state = recovering_membership_;
             :       if (current_membership_ != nullptr
             :        
> Maybe I can file another jira for the timestamp based approach for the memb
I've considered the timestamp approach, and it should be feasible. I think that currently after the fix, the current approach seems working good, as the timestamp based approach may require more consideration, I prefer that we can have it later



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 6
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 23 Sep 2025 22:53:32 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 6:

Build Failed 

https://jenkins.impala.io/job/gerrit-code-review-checks/20566/ : Initial code review checks failed. See linked job for details on the failure.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 6
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 23 Sep 2025 22:54:42 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 4:

Build Successful 

https://jenkins.impala.io/job/gerrit-code-review-checks/20031/ : Initial code review checks passed. Use gerrit-verify-dryrun-external or gerrit-verify-dryrun to run full precommit tests.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 4
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Mon, 04 Aug 2025 21:25:27 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 6:

Build Failed 

https://jenkins.impala.io/job/gerrit-code-review-checks/20567/ : Initial code review checks failed. See linked job for details on the failure.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 6
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 23 Sep 2025 23:52:28 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 3:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@297
PS2, Line 297: // The recovering membership is not exposed to clients unless the post-recovery
             :       // grace pe
> The recovering snapshot could be used to update current snapshot and as a r
Can we say that, "The recovering membership is not exposed to clients unless the post-recovery grace period has passed. Before that, no copying is needed. After the grace period, it will be set to the current membership and cleared". I added this comment.


http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       // to current membership and cleared.
             :       new_state = recovering_membership_;
             :       if (current_membership_ != nullptr
             :        
> It looks like the issue is that we're maintaining separate snapshots in rec
Yeah, one way could be removing the DCHECK in https://github.com/apache/impala/blob/branch-4.5.1/be/src/scheduling/admission-controller.cc#L2224, since the version in the queued node could be higher. However, it seems no harm to bump the version here to match the current one, and doing so may be more consistent



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 3
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 17:53:49 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 3:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc
File be/src/scheduling/cluster-membership-mgr.cc:

http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@297
PS2, Line 297: // The recovering membership is not exposed to clients unless the post-recovery
             :       // grace pe
> I was more concerned about if we could try to dereference the recovering_me
Thanks for raising this. From what I see, recovering_membership_ is a shared_ptr and is only accessed before it’s reset, and new_state holds a shared reference, I don't see the underlying Snapshot object would be somehow destroyed. Also, this logic has been around for a while and we haven’t observed any crashes in this area. Let me know if I’m missing something, are you concerned that another thread might modify it unexpectedly?
The flow here looks like:
if (recovering_membership_.get() != nullptr) {
  new_state = recovering_membership_;
}
...
if (ss_is_recovering) {
  recovering_membership_ = new_state;
  return;
}
...
SetState(new_state); // Set to current
...
recovering_membership_.reset();


http://gerrit.cloudera.org:8080/#/c/23204/2/be/src/scheduling/cluster-membership-mgr.cc@300
PS2, Line 300:       // to current membership and cleared.
             :       new_state = recovering_membership_;
             :       if (current_membership_ != nullptr
             :        
> I think we should not remove the DCHECK. The ComputeGroupScheduleStates() c
The version from recovering_membership_ starts at 0, and this seems to be the case even when not using the standalone admissiond. I’ll need to take a closer look at where it’s coming from, it might just be that the version wasn’t copied over?



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 3
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 21:09:26 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 2:

Build Successful 

https://jenkins.impala.io/job/gerrit-code-review-checks/19935/ : Initial code review checks passed. Use gerrit-verify-dryrun-external or gerrit-verify-dryrun to run full precommit tests.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 2
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Comment-Date: Tue, 22 Jul 2025 00:04:04 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/7/tests/custom_cluster/test_admission_controller.py@2259
PS7, Line 2259:     client2 = coord2.service.create_hs2_client()
              :     handle2 = client2.execute_async(short_query)
              :     # Wait until the second query is confirmed to be queued.
              :     self._wait_for_change_to_profile(handle2, "Admission result: Queued", client=client2)
> Have you try run this test again after rebase?
Umm, I don't see any issue running the test locally or in the build with the PS7 https://master-03.jenkins.cloudera.com/job/impala-private-parameterized/4204/



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Fri, 26 Sep 2025 18:10:31 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#3). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure caused by a version mismatch when
the admissiond cluster membership is recovering after a statestore
restart. During recovery, the new cluster membership version could
be lower than the version associated with a queued query, violating
the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 91 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/3
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 3
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 3:

Build Successful 

https://jenkins.impala.io/job/gerrit-code-review-checks/19992/ : Initial code review checks passed. Use gerrit-verify-dryrun-external or gerrit-verify-dryrun to run full precommit tests.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 3
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 30 Jul 2025 18:21:12 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has uploaded a new patch set (#4). ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................

IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

This patch fixes a DCHECK failure caused by a version mismatch when
the admissiond cluster membership is recovering after a statestore
restart. During recovery, the new cluster membership version could
be lower than the version associated with a queued query, violating
the DCHECK condition.

To address this, the patch sets the current version to the recovering
new cluster membership, ensuring it is not lower than any version
stored in queued queries.

Tests:
Added a regression test in
test_dcheck_issue_coord_statestore_kill_order.

Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
---
M be/src/scheduling/cluster-membership-mgr.cc
M tests/custom_cluster/test_admission_controller.py
2 files changed, 98 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/23204/4
-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 4
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Impala Public Jenkins (Code Review)" <ge...@cloudera.org>.
Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

Build Successful 

https://jenkins.impala.io/job/gerrit-code-review-checks/20568/ : Initial code review checks passed. Use gerrit-verify-dryrun-external or gerrit-verify-dryrun to run full precommit tests.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Wed, 24 Sep 2025 00:10:23 +0000
Gerrit-HasComments: No

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 6:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/23204/5//COMMIT_MSG@9
PS5, Line 9: This patch fixes a DCHECK failure
> Mention where the DCHEK is. Is it AdmissionController::ComputeGroupSchedule
Yes, the DCHECK is in AdmissionController::ComputeGroupScheduleState. Because it seems an old code, not sure the original purpose, but I guess it is something common sense that the version of cluster membership just keeps increasing, so the smaller current version compared to the old version is treated as an abnormal case. However in the release version, if the current membership version is lower, it's treated as being equal to the query's version, meaning no rescheduling is needed and no retry is needed in https://github.com/apache/impala/blob/16c350ce5aa5dbf1a9ea8b481c64780d445e9a2a/be/src/scheduling/admission-controller.cc#L2235. If the current cluster membership version is larger, it will try to find out the latest match from the cluster.
Updated the commit message.


http://gerrit.cloudera.org:8080/#/c/23204/2/tests/custom_cluster/test_admission_controller.py
File tests/custom_cluster/test_admission_controller.py:

http://gerrit.cloudera.org:8080/#/c/23204/2/tests/custom_cluster/test_admission_controller.py@2296
PS2, Line 2296:         continue
> Probably also update the test to actually change cluster membership snapsho
This is a good suggestion for a more comprehensive test. Since this test is narrowly focused on reproducing the DCHECK regression, I'd prefer to add more complex tests like that in a separate follow up patch. Does that sound reasonable?



-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 6
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 23 Sep 2025 22:47:14 +0000
Gerrit-HasComments: Yes

[Impala-ASF-CR] IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering

Posted by "Yida Wu (Code Review)" <ge...@cloudera.org>.
Yida Wu has posted comments on this change. ( http://gerrit.cloudera.org:8080/23204 )

Change subject: IMPALA-14234: Fix a version mismatch DCHECK hit when admissiond cluster membership recovering
......................................................................


Patch Set 7:

Rebased in PS7.


-- 
To view, visit http://gerrit.cloudera.org:8080/23204
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Iea77347bd4775abd7866817146e326c7c5042f5e
Gerrit-Change-Number: 23204
Gerrit-PatchSet: 7
Gerrit-Owner: Yida Wu <wy...@gmail.com>
Gerrit-Reviewer: Abhishek Rawat <ar...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <im...@cloudera.com>
Gerrit-Reviewer: Jason Fehr <jf...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <ri...@cloudera.com>
Gerrit-Reviewer: Wenzhe Zhou <wz...@cloudera.com>
Gerrit-Reviewer: Yida Wu <wy...@gmail.com>
Gerrit-Comment-Date: Tue, 23 Sep 2025 23:47:02 +0000
Gerrit-HasComments: No