[Saved object migrations] Collect all documents that fail to transform before stopping the migration

src/core/server/saved_objects/migrations/core/document_migrator.test.ts

@@ -11,6 +11,7 @@ import { set } from '@elastic/safer-lodash-set';
 import _ from 'lodash';
 import { SavedObjectUnsanitizedDoc } from '../../serialization';
 import { DocumentMigrator } from './document_migrator';
+import { TransformSavedObjectDocumentError } from './transform_saved_object_document_error';
 import { loggingSystemMock } from '../../../logging/logging_system.mock';
 import { SavedObjectsType } from '../../types';
 import { SavedObjectTypeRegistry } from '../../saved_objects_type_registry';
@@ -724,6 +725,12 @@ describe('DocumentMigrator', () => {
 
     it('logs the original error and throws a transform error if a document transform fails', () => {
       const log = mockLogger;
+      const failedDoc = {
+        id: 'smelly',
+        type: 'dog',
+        attributes: {},
+        migrationVersion: {},
+      };
       const migrator = new DocumentMigrator({
         ...testOpts(),
         typeRegistry: createRegistry({
@@ -737,12 +744,6 @@ describe('DocumentMigrator', () => {
         log,
       });
       migrator.prepareMigrations();
-      const failedDoc = {
-        id: 'smelly',
-        type: 'dog',
-        attributes: {},
-        migrationVersion: {},
-      };
       try {
         migrator.migrate(_.cloneDeep(failedDoc));
         expect('Did not throw').toEqual('But it should have!');
@@ -751,6 +752,7 @@ describe('DocumentMigrator', () => {
           "Failed to transform document smelly. Transform: dog:1.2.3
           Doc: {\\"id\\":\\"smelly\\",\\"type\\":\\"dog\\",\\"attributes\\":{},\\"migrationVersion\\":{}}"
         `);
+        expect(error).toBeInstanceOf(TransformSavedObjectDocumentError);
         expect(loggingSystemMock.collect(mockLoggerFactory).error[0][0]).toMatchInlineSnapshot(
           `[Error: Dang diggity!]`
         );

src/core/server/saved_objects/migrations/core/document_migrator.ts

@@ -62,6 +62,7 @@ import {
   SavedObjectsType,
 } from '../../types';
 import { MigrationLogger } from './migration_logger';
+import { TransformSavedObjectDocumentError } from '.';
 import { ISavedObjectTypeRegistry } from '../../saved_objects_type_registry';
 import { SavedObjectMigrationFn, SavedObjectMigrationMap } from '../types';
 import { DEFAULT_NAMESPACE_STRING } from '../../service/lib/utils';
@@ -679,9 +680,15 @@ function wrapWithTry(
       const failedTransform = `${type.name}:${version}`;
       const failedDoc = JSON.stringify(doc);
       log.error(error);
-
-      throw new Error(
-        `Failed to transform document ${doc?.id}. Transform: ${failedTransform}\nDoc: ${failedDoc}`
+      // To make debugging failed migrations easier, we add items needed to convert the
+      // saved object id to the full raw id (the id only contains the uuid part) and the full error itself
+      throw new TransformSavedObjectDocumentError(
+        doc.id,
+        doc.type,
+        doc.namespace ?? undefined,
+        failedTransform,
+        failedDoc,
+        error
       );
     }
   };

src/core/server/saved_objects/migrations/core/index.ts

@@ -15,3 +15,9 @@ export type { MigrationResult, MigrationStatus } from './migration_coordinator';
 export { createMigrationEsClient } from './migration_es_client';
 export type { MigrationEsClient } from './migration_es_client';
 export { excludeUnusedTypesQuery } from './elastic_index';
+export { TransformSavedObjectDocumentError } from './transform_saved_object_document_error';
+export type {
+  DocumentsTransformFailed,
+  DocumentsTransformSuccess,
+  TransformErrorObjects,
+} from './migrate_raw_docs';

src/core/server/saved_objects/migrations/core/migrate_raw_docs.test.ts

@@ -10,7 +10,8 @@ import { set } from '@elastic/safer-lodash-set';
 import _ from 'lodash';
 import { SavedObjectTypeRegistry } from '../../saved_objects_type_registry';
 import { SavedObjectsSerializer } from '../../serialization';
-import { migrateRawDocs } from './migrate_raw_docs';
+import { migrateRawDocs, migrateRawDocsSafely } from './migrate_raw_docs';
+import { TransformSavedObjectDocumentError } from './transform_saved_object_document_error';
 
 describe('migrateRawDocs', () => {
   test('converts raw docs to saved objects', async () => {
@@ -120,3 +121,156 @@ describe('migrateRawDocs', () => {
     ).rejects.toThrowErrorMatchingInlineSnapshot(`"error during transform"`);
   });
 });
+
+describe('migrateRawDocsSafely', () => {
+  beforeEach(() => {
+    jest.clearAllMocks();
+  });
+
+  test('converts raw docs to saved objects', async () => {
+    let result: any;
+    const transform = jest.fn<any, any>((doc: any) => [
+      set(_.cloneDeep(doc), 'attributes.name', 'HOI!'),
+    ]);
+    const task = migrateRawDocsSafely(
+      new SavedObjectsSerializer(new SavedObjectTypeRegistry()),
+      transform,
+      [
+        { _id: 'a:b', _source: { type: 'a', a: { name: 'AAA' } } },
+        { _id: 'c:d', _source: { type: 'c', c: { name: 'DDD' } } },
+      ]
+    );
+    try {
+      result = await task();
+    } catch (e) {
+      /** ignore */
+    }
+    expect(result._tag).toEqual('Right');
+    expect(result.right.processedDocs).toEqual([
+      {
+        _id: 'a:b',
+        _source: { type: 'a', a: { name: 'HOI!' }, migrationVersion: {}, references: [] },
+      },
+      {
+        _id: 'c:d',
+        _source: { type: 'c', c: { name: 'HOI!' }, migrationVersion: {}, references: [] },
+      },
+    ]);
+
+    const obj1 = {
+      id: 'b',
+      type: 'a',
+      attributes: { name: 'AAA' },
+      migrationVersion: {},
+      references: [],
+    };
+    const obj2 = {
+      id: 'd',
+      type: 'c',
+      attributes: { name: 'DDD' },
+      migrationVersion: {},
+      references: [],
+    };
+    expect(transform).toHaveBeenCalledTimes(2);
+    expect(transform).toHaveBeenNthCalledWith(1, obj1);
+    expect(transform).toHaveBeenNthCalledWith(2, obj2);
+  });
+
+  test('returns a `left` tag when encountering a corrupt saved object document', async () => {
+    let result: any;
+    const transform = jest.fn<any, any>((doc: any) => [
+      set(_.cloneDeep(doc), 'attributes.name', 'TADA'),
+    ]);
+    const task = migrateRawDocsSafely(
+      new SavedObjectsSerializer(new SavedObjectTypeRegistry()),
+      transform,
+      [
+        { _id: 'foo:b', _source: { type: 'a', a: { name: 'AAA' } } },
+        { _id: 'c:d', _source: { type: 'c', c: { name: 'DDD' } } },
+      ]
+    );
+    try {
+      result = await task();
+    } catch (e) {
+      result = e;
+    }
+    expect(transform).toHaveBeenCalledTimes(1);
+    expect(result._tag).toEqual('Left');
+    expect(Object.keys(result.left)).toEqual(['type', 'corruptDocumentIds', 'transformErrors']);
+    expect(result.left.corruptDocumentIds.length).toEqual(1);
+    expect(result.left.transformErrors.length).toEqual(0);
+  });
+
+  test('handles when one document is transformed into multiple documents', async () => {
+    let result: any;
+    const transform = jest.fn<any, any>((doc: any) => [
+      set(_.cloneDeep(doc), 'attributes.name', 'HOI!'),
+      { id: 'bar', type: 'foo', attributes: { name: 'baz' } },
+    ]);
+    const task = migrateRawDocsSafely(
+      new SavedObjectsSerializer(new SavedObjectTypeRegistry()),
+      transform,
+      [{ _id: 'a:b', _source: { type: 'a', a: { name: 'AAA' } } }]
+    );
+    try {
+      result = await task();
+    } catch (err) {
+      /** ignore */
+    }
+    expect(result._tag).toEqual('Right');
+    expect(result.right.processedDocs).toEqual([
+      {
+        _id: 'a:b',
+        _source: { type: 'a', a: { name: 'HOI!' }, migrationVersion: {}, references: [] },
+      },
+      {
+        _id: 'foo:bar',
+        _source: { type: 'foo', foo: { name: 'baz' }, references: [] },
+      },
+    ]);
+
+    const obj = {
+      id: 'b',
+      type: 'a',
+      attributes: { name: 'AAA' },
+      migrationVersion: {},
+      references: [],
+    };
+    expect(transform).toHaveBeenCalledTimes(1);
+    expect(transform).toHaveBeenCalledWith(obj);
+  });
+
+  test('rejects when the transform function throws an error', async () => {
+    let result: any;
+    const transform = jest.fn<any, any>((doc: any) => {
+      throw new TransformSavedObjectDocumentError(
+        `${doc.id}`,
+        `${doc.type}`,
+        `${doc.namespace}`,
+        `${doc.type}1.2.3`,
+        JSON.stringify(doc),
+        new Error('error during transform')
+      );
+    });
+    const task = migrateRawDocsSafely(
+      new SavedObjectsSerializer(new SavedObjectTypeRegistry()),
+      transform,
+      [{ _id: 'a:b', _source: { type: 'a', a: { name: 'AAA' } } }] // this is the raw doc
+    );
+    try {
+      result = await task();
+    } catch (err) {
+      /* ignore */
+    }
+    expect(transform).toHaveBeenCalledTimes(1);
+    expect(result._tag).toEqual('Left');
+    expect(result.left.corruptDocumentIds.length).toEqual(0);
+    expect(result.left.transformErrors.length).toEqual(1);
+    expect(result.left.transformErrors[0].err.message).toMatchInlineSnapshot(
+      `
+      "Failed to transform document b. Transform: a1.2.3
+      Doc: {\\"type\\":\\"a\\",\\"id\\":\\"b\\",\\"attributes\\":{\\"name\\":\\"AAA\\"},\\"references\\":[],\\"migrationVersion\\":{}}"
+    `
+    );
+  });
+});