branch-4.0: [improve](cloud) cloud reduce get_tablet_stats rpc to meta_service (#60543) by mymeiyi · Pull Request #61945 · apache/doris

mymeiyi · 2026-03-31T10:11:49Z

pick cloud reduce get_tablet_stats rpc to meta_service (#60543)

…pache#60543)

hello-stephen · 2026-03-31T10:12:00Z

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

What problem was fixed (it's best to include specific error reporting information). How it was fixed.
Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
What features were added. Why was this function added?
Which code was refactored and why was this part of the code refactored?
Which functions were optimized and what is the difference before and after the optimization?

mymeiyi · 2026-03-31T10:28:12Z

run buildall

Copilot

Pull request overview

This PR optimizes cloud-mode tablet statistics collection by reducing MetaService get_tablet_stats RPC volume, shifting to an “active tablets + interval ladder” fetch model, and introducing a master-to-follower stats sync RPC.

Changes:

Extend commit/compaction reporting to include tablet IDs so the FE can mark tablets “active” for stats refresh.
Add a new FE thrift RPC (syncCloudTabletStats) and implement master push + follower receive of tablet stats.
Introduce a versioned tablet-stats collection strategy (cloud_get_tablet_stats_version) with an interval ladder to reduce fetch frequency for stable tablets.

Reviewed changes

Copilot reviewed 15 out of 15 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
gensrc/thrift/FrontendService.thrift	Adds `tabletIds` to commit report request and introduces `syncCloudTabletStats` RPC + request struct.
fe/fe-core/src/main/java/org/apache/doris/transaction/GlobalTransactionMgrIface.java	Extends `afterCommitTxnResp` to accept `tabletIds`.
fe/fe-core/src/main/java/org/apache/doris/transaction/GlobalTransactionMgr.java	Updates interface implementation signature (no-op in non-cloud impl).
fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java	Handles tablet IDs in commit/compaction reports and implements `syncCloudTabletStats`.
fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java	Adds session variable `cloud_force_sync_tablet_stats`.
fe/fe-core/src/main/java/org/apache/doris/common/proc/TabletsProcDir.java	Optional “force sync” path to mark table tablets active when listing tablets (cloud-only, session-gated).
fe/fe-core/src/main/java/org/apache/doris/common/ClientPool.java	Adds `frontendStatsPool` for the new sync RPC.
fe/fe-core/src/main/java/org/apache/doris/cloud/transaction/CloudGlobalTransactionMgr.java	Tracks committed tablet IDs and marks them active after commit.
fe/fe-core/src/main/java/org/apache/doris/cloud/catalog/CloudReplica.java	Adds per-replica stats polling state (`lastGetTabletStatsTime`, `statsIntervalIndex`).
fe/fe-core/src/main/java/org/apache/doris/catalog/Replica.java	Adjusts default `rowsetCount` initialization (used by stats).
fe/fe-core/src/main/java/org/apache/doris/catalog/CloudTabletStatMgr.java	Implements versioned stats collection, active-tablet refresh, interval ladder, and master push to followers/observers.
fe/fe-core/src/main/java/org/apache/doris/alter/CloudSchemaChangeJobV2.java	Marks affected tablets active after schema change commit.
fe/fe-core/src/main/java/org/apache/doris/alter/CloudRollupJobV2.java	Marks affected tablets active after rollup creation.
fe/fe-common/src/main/java/org/apache/doris/common/Config.java	Adds `cloud_get_tablet_stats_version` and sync thread pool config.
be/src/cloud/cloud_meta_mgr.cpp	Sends tablet IDs to FE in async commit report; uses txnId=-1 path for compaction-triggered stats refresh.

Comments suppressed due to low confidence (1)

fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java:4574

On protobuf parse failure, this code prints a stack trace and still returns OK, which can silently drop stats updates. Please replace printStackTrace() with structured logging and return a non-OK TStatus (e.g., INVALID_ARGUMENT with an error message) so callers can detect/report the failure.

 } catch (InvalidProtocolBufferException e) {
// Handle the exception, log it, or take appropriate action
e.printStackTrace();
}

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-31T10:36:51Z

+ TSyncCloudTabletStatsRequest request = new TSyncCloudTabletStatsRequest();
+ request.setTabletStatsPb(ByteBuffer.wrap(response.toByteArray()));
+ for (Frontend fe : frontends) {


pushTabletStats sets tabletStatsPb using ByteBuffer.wrap(...), but the receiving side (FrontendServiceImpl.syncCloudTabletStats) reads this field as a byte[] (request.getTabletStatsPb()). This looks like a type mismatch that will either not compile or will serialize incorrectly. Set the field using the expected byte[] payload (e.g., response.toByteArray()) and drop the ByteBuffer usage/import here.

Copilot · 2026-03-31T10:36:51Z

+ activeTablets.removeAll(copiedTablets);
+ getActiveTabletStats(copiedTablets);


activeTablets is cleared (removeAll) before getActiveTabletStats(copiedTablets) runs. If fetching tablet stats fails (RPC error / task failure), those tablet IDs are still removed and won’t be retried until they become active again, which can leave stats stale. Consider only removing IDs after a successful fetch, or re-adding failed IDs back to activeTablets.
Suggested change
activeTablets.removeAll(copiedTablets);
getActiveTabletStats(copiedTablets);
getActiveTabletStats(copiedTablets);
activeTablets.removeAll(copiedTablets);

Copilot · 2026-03-31T10:36:51Z

- return new TStatus(TStatusCode.INVALID_ARGUMENT);
+ List<Long> tabletIds = request.isSetTabletIds() ? request.getTabletIds() : Collections.emptyList();
+ if (LOG.isDebugEnabled()) {
+ LOG.debug("force sync tablet stats for txnId: {}, tabletNum: {}, tabletIds: {}", request.txnId,


This debug log references request.txnId directly instead of using the Thrift accessor (getTxnId()), which is inconsistent with the rest of the method and can be misleading when the field is unset/defaulted. Prefer request.getTxnId() for consistency.
Suggested change
LOG.debug("force sync tablet stats for txnId: {}, tabletNum: {}, tabletIds: {}", request.txnId,
LOG.debug("force sync tablet stats for txnId: {}, tabletNum: {}, tabletIds: {}", request.getTxnId(),

Copilot · 2026-03-31T10:36:52Z

+ // last time to get tablet stats
+ @Getter
+ @Setter
+ long lastGetTabletStatsTime = 0;
+ /**
+ * The index of {@link org.apache.doris.catalog.CloudTabletStatMgr#DEFAULT_INTERVAL_LADDER_MS} array.
+ * Used to control the interval of getting tablet stats.
+ * When get tablet stats:
+ * if the stats is unchanged, will update this index to next value to get stats less frequently;
+ * if the stats is changed, will update this index to 0 to get stats more frequently.
+ */
+ @Getter
+ @Setter
+ int statsIntervalIndex = 0;


New fields lastGetTabletStatsTime and statsIntervalIndex are package-private and lack @SerializedName, while CloudReplica is Gson-persisted and other fields use short @SerializedName keys. Make these fields private and add appropriate @SerializedName (with alternates if needed) to keep metadata serialization stable/backward-compatible.

Copilot · 2026-03-31T10:36:52Z

+ /**
+ * The index of {@link org.apache.doris.catalog.CloudTabletStatMgr#DEFAULT_INTERVAL_LADDER_MS} array.
+ * Used to control the interval of getting tablet stats.
+ * When get tablet stats:
+ * if the stats is unchanged, will update this index to next value to get stats less frequently;
+ * if the stats is changed, will update this index to 0 to get stats more frequently.
+ */


The Javadoc links to CloudTabletStatMgr#DEFAULT_INTERVAL_LADDER_MS, but that constant is private in CloudTabletStatMgr, so the Javadoc reference will be broken. Either make the constant accessible (public/protected) or remove/adjust the link text.

Copilot · 2026-03-31T10:36:52Z

+ + "Version 1: get all tablets; Version 2: get active and interval expired tablets"})
+ public static int cloud_get_tablet_stats_version = 2;
+
+ @ConfField(description = {"Maximum concurrent number of get tablet stat jobs."})


Config description for cloud_sync_tablet_stats_task_threads_num says "get tablet stat jobs", but this setting controls the concurrency of syncing/pushing tablet stats between FEs. Please update the description to match the actual behavior to avoid operator confusion.
Suggested change
@ConfField(description = {"Maximum concurrent number of get tablet stat jobs."})
@ConfField(description = {"存算分离模式下 FE 之间同步 tablet 统计信息任务的最大并发数。",
"Maximum concurrent number of syncing tablet stats between FEs."})

doris-robot · 2026-03-31T11:06:19Z

Cloud UT Coverage Report

Increment line coverage 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	79.15% (1788/2259)
Line Coverage	64.42% (31940/49580)
Region Coverage	65.22% (15976/24494)
Branch Coverage	55.81% (8501/15232)

hello-stephen · 2026-03-31T11:41:43Z

FE UT Coverage Report

Increment line coverage 13.30% (31/233) 🎉
Increment coverage report
Complete coverage report

hello-stephen · 2026-03-31T11:55:12Z

BE UT Coverage Report

Increment line coverage 0.00% (0/17) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	52.94% (19209/36284)
Line Coverage	36.11% (178970/495572)
Region Coverage	32.73% (138692/423769)
Branch Coverage	33.70% (60268/178818)

hello-stephen · 2026-03-31T13:08:50Z

BE Regression && UT Coverage Report

Increment line coverage 80.00% (12/15) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.31% (25331/35521)
Line Coverage	54.00% (267119/494689)
Region Coverage	51.64% (221027/428045)
Branch Coverage	53.02% (95159/179461)

mymeiyi · 2026-04-01T03:44:29Z

run buildall

doris-robot · 2026-04-01T04:09:26Z

Cloud UT Coverage Report

Increment line coverage 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	79.35% (1795/2262)
Line Coverage	64.62% (32175/49794)
Region Coverage	65.45% (16094/24590)
Branch Coverage	56.03% (8579/15312)

doris-robot · 2026-04-01T07:25:41Z

BE UT Coverage Report

Increment line coverage 0.00% (0/17) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	52.90% (19233/36359)
Line Coverage	36.09% (179138/496389)
Region Coverage	32.71% (139034/425089)
Branch Coverage	33.65% (60304/179194)

hello-stephen · 2026-04-01T08:49:04Z

BE Regression && UT Coverage Report

Increment line coverage 58.82% (10/17) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.27% (25367/35593)
Line Coverage	54.03% (267703/495495)
Region Coverage	51.45% (220925/429362)
Branch Coverage	52.94% (95205/179837)

[improve](cloud) cloud reduce get_tablet_stats rpc to meta_service (a…
e59f22c
…pache#60543)

mymeiyi requested a review from yiguolei as a code owner March 31, 2026 10:11

CopilotAI review requested due to automatic review settings March 31, 2026 10:11

mymeiyi changed the title ~~[improve](cloud) cloud reduce get_tablet_stats rpc to meta_service (#60543)~~branch-4.0: [improve](cloud) cloud reduce get_tablet_stats rpc to meta_service (#60543)Mar 31, 2026

CopilotAI reviewed Mar 31, 2026
View reviewed changes

fix
d2d7b80

yiguolei approved these changes Apr 15, 2026
View reviewed changes

yiguolei merged commit 12ca10b into apache:branch-4.0Apr 15, 2026
24 of 28 checks passed

		activeTablets.removeAll(copiedTablets);
		getActiveTabletStats(copiedTablets);

	LOG.debug("force sync tablet stats for txnId: {}, tabletNum: {}, tabletIds: {}", request.txnId,
	LOG.debug("force sync tablet stats for txnId: {}, tabletNum: {}, tabletIds: {}", request.getTxnId(),

	@ConfField(description = {"Maximum concurrent number of get tablet stat jobs."})
	@ConfField(description = {"存算分离模式下 FE 之间同步 tablet 统计信息任务的最大并发数。",
	"Maximum concurrent number of syncing tablet stats between FEs."})

Uh oh!

Conversation

mymeiyi commented Mar 31, 2026

Uh oh!

hello-stephen commented Mar 31, 2026

Uh oh!

mymeiyi commented Mar 31, 2026

Uh oh!

CopilotAI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

CopilotAIMar 31, 2026

Choose a reason for hiding this comment

Uh oh!

doris-robot commented Mar 31, 2026

Cloud UT Coverage Report

Uh oh!

hello-stephen commented Mar 31, 2026

FE UT Coverage Report

Uh oh!

hello-stephen commented Mar 31, 2026

BE UT Coverage Report

Uh oh!

hello-stephen commented Mar 31, 2026

BE Regression && UT Coverage Report

Uh oh!

mymeiyi commented Apr 1, 2026

Uh oh!

doris-robot commented Apr 1, 2026

Cloud UT Coverage Report

Uh oh!

doris-robot commented Apr 1, 2026

BE UT Coverage Report

Uh oh!

hello-stephen commented Apr 1, 2026

BE Regression && UT Coverage Report

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants