datalake/coordinator: add reset topic state escape hatch by nvartolomei · Pull Request #29596 · redpanda-data/redpanda

nvartolomei · 2026-02-12T17:34:22Z

Backports Required

Release Notes

none

Copilot

Pull request overview

Adds an admin “escape hatch” to reset datalake coordinator per-topic state, including optional per-partition last_committed overrides, and validates it with unit + e2e coverage.

Changes:

Add CoordinatorResetState admin RPC + generated client bindings and protobuf types.
Implement coordinator/frontend/state-machine support for reset_topic_state updates and overrides.
Add coordinator reset unit tests and a new datalake e2e test exercising reset behavior.

Reviewed changes

Copilot reviewed 25 out of 25 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
tests/rptest/tests/datalake/datalake_e2e_test.py	Adds an end-to-end test for coordinator reset semantics and overrides.
tests/rptest/clients/types.py	Adds a TopicSpec config key for iceberg partition spec used by tests.
tests/rptest/clients/admin/proto/redpanda/core/admin/internal/datalake/v1/datalake_pb2_connect.py	Adds client stubs for the new `CoordinatorResetState` RPC.
tests/rptest/clients/admin/proto/redpanda/core/admin/internal/datalake/v1/datalake_pb2.pyi	Adds typing stubs for new protobuf messages.
tests/rptest/clients/admin/proto/redpanda/core/admin/internal/datalake/v1/datalake_pb2.py	Updates generated protobuf module with new messages + service method.
src/v/redpanda/admin/services/datalake/datalake.h	Exposes `coordinator_reset_state` handler in admin service impl.
src/v/redpanda/admin/services/datalake/datalake.cc	Implements admin RPC handler calling coordinator frontend reset.
src/v/datalake/coordinator/types.h	Adds RPC request/reply types for reset topic state + overrides.
src/v/datalake/coordinator/types.cc	Adds logging formatter for reset reply.
src/v/datalake/coordinator/tests/state_update_test.cc	Adds unit tests for reset update behavior.
src/v/datalake/coordinator/tests/BUILD	Adds deps for new reset/override types in tests.
src/v/datalake/coordinator/state_update.h	Adds `reset_topic_state` update key + update struct.
src/v/datalake/coordinator/state_update.cc	Implements apply/can_apply + logging for reset update.
src/v/datalake/coordinator/state_machine.cc	Applies the new reset update key in the STM.
src/v/datalake/coordinator/service.h	Adds RPC endpoint for reset topic state.
src/v/datalake/coordinator/service.cc	Implements RPC dispatch into frontend reset handler.
src/v/datalake/coordinator/rpc.json	Declares new RPC method for codegen.
src/v/datalake/coordinator/partition_state_override.h	Introduces the override struct for per-partition fields.
src/v/datalake/coordinator/partition_state_override.cc	Implements logging formatter for overrides.
src/v/datalake/coordinator/frontend.h	Adds frontend reset API + local execution helper.
src/v/datalake/coordinator/frontend.cc	Implements frontend reset path and coordinator-manager invocation.
src/v/datalake/coordinator/coordinator.h	Adds coordinator sync API to replicate reset updates.
src/v/datalake/coordinator/coordinator.cc	Implements STM replication of the reset update.
src/v/datalake/coordinator/BUILD	Adds build targets/deps for new override type.
proto/redpanda/core/admin/internal/datalake/v1/datalake.proto	Adds protobuf messages + admin RPC definition.

src/v/datalake/coordinator/frontend.cc

src/v/datalake/coordinator/coordinator.cc

src/v/datalake/coordinator/partition_state_override.h

tests/rptest/tests/datalake/datalake_e2e_test.py

Operators may need to clear pending files from the datalake coordinator when Iceberg catalog state becomes inconsistent (e.g. after manual catalog modifications) which can result in stuck coordinator. The new CoordinatorResetState RPC is exposed via the admin API with SUPERUSER authorization and plumbed through the coordinator frontend, RPC service, and state machine.

andrwng

Nice, LGTM! One nit/question about the test, but doesn't need to block

tests/rptest/tests/datalake/datalake_e2e_test.py

proto/redpanda/core/admin/internal/datalake/v1/datalake.proto

nvartolomei · 2026-02-18T16:11:07Z

/ci-repeat 1
skip-redpanda-build
skip-rebase

oleiman

lgtm. i have one question about the override logic.

oleiman · 2026-02-18T18:18:05Z

src/v/datalake/coordinator/coordinator.cc

+    auto check_res = update.can_apply(stm_->state());
+    if (check_res.has_error()) {
+        vlog(
+          datalake_log.debug,


nitpick: INFO?

Copy pasted. Can address in follow up.

oleiman · 2026-02-18T18:40:35Z

src/v/datalake/coordinator/partition_state_override.cc

+    if (p.last_committed.has_value()) {
+        fmt::print(o, "{{last_committed: {}}}", p.last_committed.value());
+    } else {
+        fmt::print(o, "{{last_committed: nullopt}}");
+    }


nitpick/fyi: I think just formatting the optional will do the right thing, no?

It doesn't work without additional includes here and I refused to add them because it is a hack and we'll have to get rid of it soon-ish.

This violates C++ standard guidelines https://nvartolomei.com/cpp-code-hygiene/#specializing-standard-library-know-your-boundaries.

In newer fmt, ostream formatters aren't used afaik. Preferred correct code rather than yoloing.

missed that it was an ostream operator... why are we even writing one, exactly?

src/v/datalake/coordinator/state_update.cc

nvartolomei · 2026-02-19T08:42:39Z

/ci-repeat 1
skip-redpanda-build
skip-rebase

vbotbuildovich · 2026-02-19T09:42:52Z

Retry command for Build#80732

please wait until all jobs are finished before running the slash command

/ci-repeat 1
skip-redpanda-build
skip-units
skip-rebase
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsCompactionTest.test_compact@{"cloud_topics_compaction_key_map_memory_kb":10}
tests/rptest/tests/describe_topics_test.py::DescribeTopicsTest.test_describe_topics_with_documentation_and_types
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsCompactionTest.test_compact@{"cloud_topics_compaction_key_map_memory_kb":3}
tests/rptest/tests/cloud_topics/l0_gc_test.py::CloudTopicsL0GCTest.test_l0_gc@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsTest.test_write
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsCompactionTest.test_compact@{"cloud_topics_compaction_key_map_memory_kb":131072}
tests/rptest/tests/cluster_linking_e2e_test.py::ShadowLinkBasicTests.test_disallowed_topic_properties
tests/rptest/tests/cloud_topics/iceberg_test.py::EndToEndCloudTopicsIcebergCompactionTest.test_compaction_preserves_all_offsets_in_iceberg@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/retention_test.py::CloudTopicsRetentionTest.test_size_based_retention@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsTest.test_get_size
tests/rptest/tests/cloud_topics/l0_gc_test.py::CloudTopicsL0GCTest.test_l0_gc@{"cloud_storage_type":2}
tests/rptest/tests/data_migrations_api_test.py::DataMigrationsApiTest.test_conflicting_group_migrations
tests/rptest/tests/cloud_topics/l0_gc_test.py::CloudTopicsL0GCAdminTest.test_basic_pause_unpause@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/retention_test.py::CloudTopicsRetentionTest.test_time_based_retention@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsTxTest.test_write
tests/rptest/tests/cloud_topics/retention_test.py::CloudTopicsRetentionTest.test_size_based_retention@{"cloud_storage_type":2}
tests/rptest/tests/cluster_config_test.py::ClusterConfigTest.test_valid_settings
tests/rptest/tests/cloud_topics/l0_gc_test.py::CloudTopicsL0GCAdminTest.test_single_node_pause_unpause@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/e2e_test.py::EndToEndCloudTopicsTest.test_delete_records
tests/rptest/tests/data_migrations_api_test.py::DataMigrationsApiTest.test_cloud_topic_unmount_rejected
tests/rptest/tests/cloud_topics/iceberg_test.py::EndToEndCloudTopicsIcebergDeletionTest.test_deletion_blocked_until_translated@{"cloud_storage_type":1}
tests/rptest/tests/cloud_topics/retention_test.py::CloudTopicsRetentionTest.test_time_based_retention@{"cloud_storage_type":2}

nvartolomei · 2026-02-19T11:22:20Z

/ci-repeat 1

src/v/datalake/coordinator/state_update.cc

In some cases we want to manually choose from where translation should resume after a coordinator reset. Add per-partition last_committed overrides and a reset_all_partitions flag. By default reset_all_partitions is false and the request is a no-op unless partition_overrides are provided — only the listed partitions have their pending entries cleared and last_committed set. Setting reset_all_partitions to true clears all partitions first, then applies any overrides. This default-safe design minimizes surprise and makes it harder to accidentally wipe state.

Copilot AI review requested due to automatic review settings February 12, 2026 17:34

github-actions bot added area/build area/redpanda labels Feb 12, 2026

Copilot AI reviewed Feb 12, 2026

View reviewed changes

src/v/datalake/coordinator/frontend.cc Show resolved Hide resolved

src/v/datalake/coordinator/coordinator.cc Show resolved Hide resolved

src/v/datalake/coordinator/partition_state_override.h Show resolved Hide resolved

tests/rptest/tests/datalake/datalake_e2e_test.py Show resolved Hide resolved

This comment was marked as outdated.

Sign in to view

nvartolomei force-pushed the nv/iceberg-reset branch from ccc3337 to c8b0b7a Compare February 17, 2026 11:03

nvartolomei requested review from andrwng and oleiman February 17, 2026 11:43

andrwng previously approved these changes Feb 17, 2026

View reviewed changes

tests/rptest/tests/datalake/datalake_e2e_test.py Outdated Show resolved Hide resolved

dotnwat reviewed Feb 17, 2026

View reviewed changes

proto/redpanda/core/admin/internal/datalake/v1/datalake.proto Show resolved Hide resolved

nvartolomei dismissed andrwng’s stale review via 715890a February 18, 2026 15:33

nvartolomei force-pushed the nv/iceberg-reset branch from c8b0b7a to 715890a Compare February 18, 2026 15:33

nvartolomei enabled auto-merge February 18, 2026 15:41

nvartolomei requested a review from andrwng February 18, 2026 15:41

nvartolomei force-pushed the nv/iceberg-reset branch from 715890a to 82b20d2 Compare February 18, 2026 16:29

andrwng previously approved these changes Feb 18, 2026

View reviewed changes

oleiman reviewed Feb 18, 2026

View reviewed changes

nvartolomei disabled auto-merge February 19, 2026 11:37

oleiman previously approved these changes Feb 19, 2026

View reviewed changes

src/v/datalake/coordinator/state_update.cc Show resolved Hide resolved

nvartolomei dismissed stale reviews from oleiman and andrwng via f44ca9d February 20, 2026 11:44

nvartolomei force-pushed the nv/iceberg-reset branch from 82b20d2 to f44ca9d Compare February 20, 2026 11:44

nvartolomei requested a review from oleiman February 20, 2026 11:45

oleiman approved these changes Feb 20, 2026

View reviewed changes

nvartolomei merged commit f87258f into redpanda-data:dev Feb 20, 2026
30 checks passed

nvartolomei mentioned this pull request Mar 2, 2026

WIP: iceberg: reset coordinator pending files #28743

Closed

8 tasks

Conversation

nvartolomei commented Feb 12, 2026

Backports Required

Release Notes

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

This comment was marked as outdated.

This comment was marked as outdated.

andrwng left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

nvartolomei commented Feb 18, 2026

Uh oh!

oleiman left a comment

Choose a reason for hiding this comment

Uh oh!

oleiman Feb 18, 2026

Choose a reason for hiding this comment

Uh oh!

nvartolomei Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

oleiman Feb 18, 2026

Choose a reason for hiding this comment

Uh oh!

nvartolomei Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

oleiman Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

nvartolomei commented Feb 19, 2026

Uh oh!

vbotbuildovich commented Feb 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Retry command for Build#80732

Uh oh!

nvartolomei commented Feb 19, 2026

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

vbotbuildovich commented Feb 19, 2026 •

edited

Loading