UBERF-5061 Rewrite queries in index consistency check (#4430)

Signed-off-by: Alexander Onnikov <alexander.onnikov@xored.com>

models/core/src/core.ts

@@ -318,6 +318,11 @@ export class TDocIndexState extends TDoc implements DocIndexState {
   @Index(IndexKind.Indexed)
   @Hidden()
     stages!: Record<string, boolean | string>
+
+  @Prop(TypeString(), getEmbeddedLabel('Generation'))
+  @Index(IndexKind.Indexed)
+  @Hidden()
+    generationId?: string
 }
 
 @Model(core.class.IndexStageState, core.class.Doc, DOMAIN_DOC_INDEX_STATE)

packages/core/src/classes.ts

@@ -432,6 +432,8 @@ export interface DocIndexState extends Doc {
   attachedTo?: Ref<Doc>
   attachedToClass?: Ref<Class<Doc>>
 
+  generationId?: string
+
   // States for stages
   stages: Record<string, boolean | string>
 

server/core/src/indexer/indexer.ts

@@ -17,6 +17,7 @@ import core, {
   AttachedDoc,
   Class,
   DOMAIN_DOC_INDEX_STATE,
+  DOMAIN_FULLTEXT_BLOB,
   Doc,
   DocIndexState,
   DocumentQuery,
@@ -32,7 +33,8 @@ import core, {
   setObjectValue,
   toFindResult,
   versionToString,
-  docKey
+  docKey,
+  generateId
 } from '@hcengineering/core'
 import { DbAdapter } from '../adapter'
 import { RateLimitter } from '../limitter'
@@ -600,44 +602,83 @@ export class FullTextIndexPipeline implements FullTextPipeline {
 
         console.log(this.workspace.name, 'checking index', c)
 
-        // All saved state documents
+        const generationId = generateId()
-        const states = (
+
-          await this.storage.findAll(core.class.DocIndexState, { objectClass: c }, { projection: { _id: 1 } })
+        let lastId = ''
-        ).map((it) => it._id)
 
         while (true) {
           if (this.cancelling) {
             return
           }
+
           let newDocs: DocIndexState[] = []
+          let updates = new Map<Ref<DocIndexState>, DocumentUpdate<DocIndexState>>()
+
           try {
-            newDocs = (
+            const docs = await dbStorage.findAll<Doc>(
-              await dbStorage.findAll<Doc>(
+              this.metrics,
-                this.metrics,
+              c,
-                c,
+              { _class: c, _id: { $gt: lastId as any } },
-                { _class: c, _id: { $nin: states } },
+              {
-                { limit: 500, projection: { _id: 1, attachedTo: 1, attachedToClass: 1 } as any }
+                limit: 10000,
+                sort: { _id: 1 },
+                projection: { _id: 1, attachedTo: 1, attachedToClass: 1 } as any
+              }
+            )
+
+            if (docs.length === 0) {
+              // All updated for this class
+              break
+            }
+
+            lastId = docs[docs.length - 1]._id
+
+            const states = (
+              await this.storage.findAll(
+                core.class.DocIndexState,
+                {
+                  objectClass: c,
+                  _id: {
+                    $gte: docs[0]._id as any,
+                    $lte: docs[docs.length - 1]._id as any
+                  }
+                },
+                { projection: { _id: 1 } }
               )
-            ).map((it) => {
+            ).map((it) => it._id)
-              return createStateDoc(it._id, c, {
+            const statesSet = new Set(states)
-                stages: {},
+
-                attributes: {},
+            // create missing index states
-                removed: false,
+            newDocs = docs
-                space: it.space,
+              .filter((it) => !statesSet.has(it._id as Ref<DocIndexState>))
-                attachedTo: (it as AttachedDoc)?.attachedTo ?? undefined,
+              .map((it) => {
-                attachedToClass: (it as AttachedDoc)?.attachedToClass ?? undefined
+                return createStateDoc(it._id, c, {
+                  generationId,
+                  stages: {},
+                  attributes: {},
+                  removed: false,
+                  space: it.space,
+                  attachedTo: (it as AttachedDoc)?.attachedTo ?? undefined,
+                  attachedToClass: (it as AttachedDoc)?.attachedToClass ?? undefined
+                })
+              })
+
+            // update generationId for existing index states
+            updates = new Map()
+            docs
+              .filter((it) => statesSet.has(it._id as Ref<DocIndexState>))
+              .forEach((it) => {
+                updates.set(it._id as Ref<DocIndexState>, { generationId })
               })
-            })
           } catch (e) {
             console.error(e)
             break
           }
 
-          states.push(...newDocs.map((it) => it._id))
+          try {
-
+            await this.storage.update(DOMAIN_DOC_INDEX_STATE, updates)
-          if (newDocs.length === 0) {
+          } catch (err: any) {
-            // All updated for this class
+            console.error(err)
-            break
           }
 
           try {
@@ -646,11 +687,20 @@ export class FullTextIndexPipeline implements FullTextPipeline {
             console.error(err)
           }
         }
-        const statesSet = new Set(states)
+
-        const docIds = (await dbStorage.findAll<Doc>(this.metrics, c, { _class: c }, { projection: { _id: 1 } }))
+        // remove index states for documents that do not exist
-          .filter((it) => !statesSet.has(it._id as Ref<DocIndexState>))
+        const toRemove = (
-          .map((it) => it._id)
+          await this.storage.findAll(
-        await this.storage.clean(DOMAIN_DOC_INDEX_STATE, docIds)
+            core.class.DocIndexState,
+            { objectClass: c, generationId: { $ne: generationId } },
+            { projection: { _id: 1 } }
+          )
+        ).map((it) => it._id)
+
+        if (toRemove.length > 0) {
+          await this.storage.clean(DOMAIN_DOC_INDEX_STATE, toRemove)
+          await this.storage.clean(DOMAIN_FULLTEXT_BLOB, toRemove)
+        }
       }
 
       // Clean for non existing classes