refactor(1-1-restore): refactors 1-1-restore progress #4348

VAveryanov8 · 2025-04-14T14:25:44Z

This refactors 1-1-restore progress to address code review comments from #4296 by making following changes:

include nodetool refresh into table progress
only track progress of table/view restore
removes unnecessary stats (shard_count, host bandwidth, aggregation by
host, keyspaces). These stats are better to keep in metrics or
calculate on the client side.

Now getProgress api response will have following structure:

{
  "tables": [
    {
      "started_at": "2025-01-01T12:00:01Z",
      "completed_at": "2025-01-01T12:00:03Z",
      "size": 20,
      "restored": 20,
      "status": "done",
      "keyspace": "ks1",
      "table": "t2"
    },
    {
      "started_at": "2025-01-01T12:00:00Z",
      "completed_at": null,
      "size": 40,
      "restored": 40,
      "status": "in_progress",
      "keyspace": "ks1",
      "table": "t1"
    }
  ],
  "views": [
    {
      "started_at": "2025-01-01T12:00:01Z",
      "completed_at": null,
      "size": 0,
      "restored": 0,
      "status": "in_progress",
      "keyspace": "ks1",
      "table": "t1",
      "view": "v1",
      "type": "MaterializedView"
    },
    {
      "started_at": "2025-01-01T12:01:03Z",
      "completed_at": "2025-01-01T12:00:04Z",
      "size": 0,
      "restored": 0,
      "status": "done",
      "keyspace": "ks1",
      "table": "t2",
      "view": "v2",
      "type": "MaterializedView"
    }
  ]
}

status field is introduced to better cover cases when we can't say if restore operation of table/view is complete just by looking at size and restored, because for tables we need to wait until nodetool refresh is complete and for views - until view rebuild operation is complete.

Refs: #4205

Please make sure that:

Code is split to commits that address a single change
Commit messages are informative
Commit titles have module prefix
Commit titles have issue nr. suffix

karol-kokoszka · 2025-04-14T18:21:29Z

@VAveryanov8 please pull from the feature branch (or git stash your commits before, and git stash pop after pulling them)

Copilot

Copilot reviewed 39 out of 51 changed files in this pull request and generated no comments.

Files not reviewed (12)

pkg/service/one2onerestore/testdata/1.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/1.run_view_progress.json: Language not supported
pkg/service/one2onerestore/testdata/2.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/2.run_view_progress.json: Language not supported
pkg/service/one2onerestore/testdata/3.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/3.run_view_progress.json: Language not supported
pkg/service/one2onerestore/testdata/4.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/4.run_view_progress.json: Language not supported
pkg/service/one2onerestore/testdata/5.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/5.run_view_progress.json: Language not supported
pkg/service/one2onerestore/testdata/6.run_table_progress.json: Language not supported
pkg/service/one2onerestore/testdata/6.run_view_progress.json: Language not supported

Comments suppressed due to low confidence (3)

pkg/service/one2onerestore/progress.go:296

The mergeProgressStatus function returns 'in_progress' when merging differing statuses (except if one is 'failed'). Please verify that this behavior is intended for cases where one progress might already be 'done'.

func mergeProgressStatus(dst, src ProgressStatus) ProgressStatus {

pkg/service/one2onerestore/model.go:75

[nitpick] Consider renaming 'scyllaTableWithSize' to a more descriptive name such as 'tableToRestore' to clarify its role in tracking table sizes for restoration.

tablesToRestore []scyllaTableWithSize

pkg/service/one2onerestore/progress.go:9

[nitpick] Consider renaming the alias 'stderr' for the standard errors package to something more descriptive like 'stdErrors' to avoid potential confusion.

stderr "errors"

Michal-Leszczynski

Just some mostly subjective nits

Michal-Leszczynski · 2025-04-29T14:44:05Z

pkg/service/one2onerestore/model.go

+type scyllaTableWithSize struct {
+	scyllaTable


nit: I'm a fan of shorter names (e.g. just table or tableWithSize), as the default table meaning in the context of SM is the Scylla table.

Michal-Leszczynski · 2025-04-29T14:45:30Z

pkg/service/one2onerestore/model.go

+	TableSize           int64
+	Downloaded          int64
+	VersionedDownloaded int64
+	IsRefreshed         bool // indicates whether node tool refresh is completed for this table or not


node tool -> nodetool

Michal-Leszczynski · 2025-04-29T14:47:13Z

pkg/service/one2onerestore/model.go

+	KeyspaceName string
+	TableName    string


nit: you can use shorter names (e.g. keyspace or table) with db:"" tag in order to handle names reserved by CQL.

Michal-Leszczynski · 2025-04-29T14:50:42Z

schema/v3.6.0.cql

+CREATE TABLE one2onerestore_run_view_progress (
+    cluster_id        uuid,
+    task_id           uuid,
+    run_id            uuid,
+
+    started_at        timestamp,
+    completed_at      timestamp,
+
+    keyspace_name     text,
+    table_name        text,
+    view_type         text,


Since there is just table_name, it's not 100% obvious if it is the view table name or the base table name.
Maybe it would be better to rename it to or just add view_name?

Michal-Leszczynski · 2025-04-29T15:06:46Z

pkg/service/one2onerestore/model.go

+var (
+	// ProgressStatusNotStarted indicates that 1-1-restore of table/view is not yet started.
+	ProgressStatusNotStarted ProgressStatus = "not_started"
+	// ProgressStatusInProgress indicates that 1-1-restore of table/view is in progress.
+	ProgressStatusInProgress ProgressStatus = "in_progress"
+	// ProgressStatusDone indicates that 1-1-restore of table/view is done.
+	ProgressStatusDone ProgressStatus = "done"
+)


Shouldn't it be const?

Michal-Leszczynski · 2025-04-29T15:14:54Z

pkg/service/one2onerestore/progress.go

+	if rtp.StartedAt != nil && rtp.CompletedAt == nil {
+		return ProgressStatusInProgress
+	}
+	if rtp.StartedAt != nil && rtp.CompletedAt != nil && !rtp.IsRefreshed {
+		return ProgressStatusInProgress
 	}


nit: we can just remove those 2 if statements

This refactors 1-1-restore progress by making following changes: - include `nodetool refresh` into table progress - only track progress of table/view restore - removes unnecessary stats (shard_count, host bandwidth, aggregation by host, keyspaces). These stats is better to keep in metrics or calculate on the client side. Refs: #4205

This introduces "failed" status for the ProgressStatus, which is used only if table/view restore ended with error.

This makes various small changes like: - use shorter names for Keyspace, Table and View (instead of KeyspaceName and etc.) - add view_name to avoid confusions with table_name - change ProgressStatus from var to const - some smaller renaming and etc.

This refactors 1-1-restore progress by making following changes: - include `nodetool refresh` into table progress - only track progress of table/view restore - removes unnecessary stats (shard_count, host bandwidth, aggregation by host, keyspaces). These stats is better to keep in metrics or calculate on the client side. Refs: #4205

This refactors 1-1-restore progress by making following changes: - include `nodetool refresh` into table progress - only track progress of table/view restore - removes unnecessary stats (shard_count, host bandwidth, aggregation by host, keyspaces). These stats is better to keep in metrics or calculate on the client side. Refs: #4205 fix(1-1-restore): fix 1-1-restore progress aggregation (#4429) This fixes several small issues in progress aggregation: 1. Uniqueness of views was checked by keyspace + table name 2. Progress model had the wrong json tag for ViewType

This adds progress tracking of various 1-1-restore stage. Each progress update represented by RunProgress model saved into db. w.getProgress is used to aggregate current task progress by Keyspaces, Tables and Hosts. refactor(1-1-restore): refactors 1-1-restore progress (#4348) This refactors 1-1-restore progress by making following changes: - include `nodetool refresh` into table progress - only track progress of table/view restore - removes unnecessary stats (shard_count, host bandwidth, aggregation by host, keyspaces). These stats is better to keep in metrics or calculate on the client side. Refs: #4205 fix(1-1-restore): fix 1-1-restore progress aggregation (#4429) This fixes several small issues in progress aggregation: 1. Uniqueness of views was checked by keyspace + table name 2. Progress model had the wrong json tag for ViewType

This adds progress tracking of various 1-1-restore stage. Each progress update represented by RunProgress model saved into db. w.getProgress is used to aggregate current task progress by Keyspaces, Tables and Hosts. refactor(1-1-restore): refactors 1-1-restore progress (#4348) This refactors 1-1-restore progress by making following changes: - include `nodetool refresh` into table progress - only track progress of table/view restore - removes unnecessary stats (shard_count, host bandwidth, aggregation by host, keyspaces). These stats is better to keep in metrics or calculate on the client side. Refs: #4205 fix(1-1-restore): fix 1-1-restore progress aggregation (#4429) This fixes several small issues in progress aggregation: 1. Uniqueness of views was checked by keyspace + table name 2. Progress model had the wrong json tag for ViewType (cherry picked from commit a1c4314)

VAveryanov8 force-pushed the va/1-1-restore-progress-refactor branch from 8c15d92 to 04922ec Compare April 14, 2025 14:56

karol-kokoszka force-pushed the 1-1-restore-feature-branch branch from c5bf399 to f10f9f7 Compare April 14, 2025 18:00

karol-kokoszka force-pushed the 1-1-restore-feature-branch branch from f10f9f7 to d664493 Compare April 14, 2025 19:33

VAveryanov8 force-pushed the va/1-1-restore-progress-refactor branch 4 times, most recently from ca6e046 to c2f8135 Compare April 15, 2025 13:41

VAveryanov8 self-assigned this Apr 15, 2025

VAveryanov8 marked this pull request as ready for review April 15, 2025 14:36

VAveryanov8 requested review from karol-kokoszka and Michal-Leszczynski as code owners April 15, 2025 14:36

VAveryanov8 requested a review from Copilot April 17, 2025 13:24

Copilot AI reviewed Apr 17, 2025

View reviewed changes

Michal-Leszczynski approved these changes Apr 29, 2025

View reviewed changes

VAveryanov8 added 4 commits May 2, 2025 10:54

test(1-1-restore): adjusts tests for 1-1-restore progress

a5dcbbd

refactor: introduces "failed" status

e701bed

This introduces "failed" status for the ProgressStatus, which is used only if table/view restore ended with error.

fix: address code review comments

f981e5b

This makes various small changes like: - use shorter names for Keyspace, Table and View (instead of KeyspaceName and etc.) - add view_name to avoid confusions with table_name - change ProgressStatus from var to const - some smaller renaming and etc.

VAveryanov8 force-pushed the va/1-1-restore-progress-refactor branch from b3bc757 to f981e5b Compare May 2, 2025 08:55

VAveryanov8 merged commit a0c92d3 into 1-1-restore-feature-branch May 2, 2025
51 checks passed

VAveryanov8 deleted the va/1-1-restore-progress-refactor branch May 2, 2025 10:05

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

refactor(1-1-restore): refactors 1-1-restore progress #4348

refactor(1-1-restore): refactors 1-1-restore progress #4348

Uh oh!

VAveryanov8 commented Apr 14, 2025 •

edited

Loading

Uh oh!

karol-kokoszka commented Apr 14, 2025 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Michal-Leszczynski left a comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Michal-Leszczynski Apr 29, 2025

Uh oh!

Uh oh!

Uh oh!

refactor(1-1-restore): refactors 1-1-restore progress #4348

refactor(1-1-restore): refactors 1-1-restore progress #4348

Uh oh!

Conversation

VAveryanov8 commented Apr 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

karol-kokoszka commented Apr 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski left a comment

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Michal-Leszczynski Apr 29, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

VAveryanov8 commented Apr 14, 2025 •

edited

Loading

karol-kokoszka commented Apr 14, 2025 •

edited

Loading