[history-v1] add readOnly lookup for raw chunks (#23318)

* [history-v1] add readOnly lookup for raw chunks

Co-authored-by: Eric Mc Sween <eric.mcsween@overleaf.com>

* [history-v1] reduce min poolsize for readOnly pool to 0

Co-authored-by: Brian Gough <brian.gough@overleaf.com>

---------

Co-authored-by: Eric Mc Sween <eric.mcsween@overleaf.com>
Co-authored-by: Brian Gough <brian.gough@overleaf.com>
GitOrigin-RevId: a711c4ee4f3ea3775bd090e620d1ef52689fa1f4

server-ce/config/custom-environment-variables.json

@@ -1,5 +1,6 @@
 {
   "databaseUrl": "HISTORY_CONNECTION_STRING",
+  "databaseUrlReadOnly": "HISTORY_FOLLOWER_CONNECTION_STRING",
   "herokuDatabaseUrl": "DATABASE_URL",
   "databasePoolMin": "DATABASE_POOL_MIN",
   "databasePoolMax": "DATABASE_POOL_MAX",

services/history-v1/api/controllers/projects.js

@@ -88,9 +88,10 @@ async function getLatestHistory(req, res, next) {
 
 async function getLatestHistoryRaw(req, res, next) {
   const projectId = req.swagger.params.project_id.value
+  const readOnly = req.swagger.params.readOnly.value
   try {
     const { startVersion, endVersion, endTimestamp } =
-      await chunkStore.loadLatestRaw(projectId)
+      await chunkStore.loadLatestRaw(projectId, { readOnly })
     res.json({
       startVersion,
       endVersion,

services/history-v1/api/swagger/projects.js

@@ -335,6 +335,13 @@ exports.paths = {
           required: true,
           type: 'string',
         },
+        {
+          name: 'readOnly',
+          in: 'query',
+          description: 'use read only database connection',
+          required: false,
+          type: 'boolean',
+        },
       ],
       responses: {
         200: {

services/history-v1/config/custom-environment-variables.json

@@ -1,5 +1,6 @@
 {
   "databaseUrl": "HISTORY_CONNECTION_STRING",
+  "databaseUrlReadOnly": "HISTORY_FOLLOWER_CONNECTION_STRING",
   "herokuDatabaseUrl": "DATABASE_URL",
   "databasePoolMin": "DATABASE_POOL_MIN",
   "databasePoolMax": "DATABASE_POOL_MAX",

services/history-v1/config/test.json

@@ -1,5 +1,6 @@
 {
   "databaseUrl": "postgres://overleaf:overleaf@postgres/overleaf-history-v1-test",
+  "databaseUrlReadOnly": "postgres://read_only:password@postgres/overleaf-history-v1-test",
   "persistor": {
     "backend": "gcs",
     "gcs": {

services/history-v1/docker-compose.ci.yml

@@ -72,6 +72,8 @@ services:
       POSTGRES_USER: overleaf
       POSTGRES_PASSWORD: overleaf
       POSTGRES_DB: overleaf-history-v1-test
+    volumes:
+      - ./test/acceptance/pg-init/:/docker-entrypoint-initdb.d/
     healthcheck:
       test: pg_isready --quiet
       interval: 1s

services/history-v1/docker-compose.yml

@@ -80,6 +80,8 @@ services:
       POSTGRES_USER: overleaf
       POSTGRES_PASSWORD: overleaf
       POSTGRES_DB: overleaf-history-v1-test
+    volumes:
+      - ./test/acceptance/pg-init/:/docker-entrypoint-initdb.d/
     healthcheck:
       test: pg_isready --host=localhost --quiet
       interval: 1s

services/history-v1/storage/lib/chunk_store/index.js

@@ -82,13 +82,15 @@ async function lazyLoadHistoryFiles(history, batchBlobStore) {
  * Load the latest Chunk stored for a project, including blob metadata.
  *
  * @param {string} projectId
+ * @param {Object} [opts]
+ * @param {boolean} [opts.readOnly]
  * @return {Promise<{id: string, startVersion: number, endVersion: number, endTimestamp: Date}>}
  */
-async function loadLatestRaw(projectId) {
+async function loadLatestRaw(projectId, opts) {
   assert.projectId(projectId, 'bad projectId')
 
   const backend = getBackend(projectId)
-  const chunkRecord = await backend.getLatestChunk(projectId)
+  const chunkRecord = await backend.getLatestChunk(projectId, opts)
   if (chunkRecord == null) {
     throw new Chunk.NotFoundError(projectId)
   }

services/history-v1/storage/lib/chunk_store/mongo.js

@@ -1,4 +1,4 @@
-const { ObjectId } = require('mongodb')
+const { ObjectId, ReadPreference } = require('mongodb')
 const { Chunk } = require('overleaf-editor-core')
 const OError = require('@overleaf/o-error')
 const assert = require('../assert')
@@ -9,13 +9,22 @@ const DUPLICATE_KEY_ERROR_CODE = 11000
 
 /**
  * Get the latest chunk's metadata from the database
+ * @param {string} projectId
+ * @param {Object} [opts]
+ * @param {boolean} [opts.readOnly]
  */
-async function getLatestChunk(projectId) {
+async function getLatestChunk(projectId, opts = {}) {
   assert.mongoId(projectId, 'bad projectId')
+  const { readOnly = false } = opts
 
   const record = await mongodb.chunks.findOne(
     { projectId: new ObjectId(projectId), state: 'active' },
-    { sort: { startVersion: -1 } }
+    {
+      sort: { startVersion: -1 },
+      readPreference: readOnly
+        ? ReadPreference.secondaryPreferred
+        : ReadPreference.primary,
+    }
   )
   if (record == null) {
     return null

services/history-v1/storage/lib/chunk_store/postgres.js

@@ -1,18 +1,23 @@
 const { Chunk } = require('overleaf-editor-core')
 const assert = require('../assert')
 const knex = require('../knex')
+const knexReadOnly = require('../knex_read_only')
 const { ChunkVersionConflictError } = require('./errors')
 
 const DUPLICATE_KEY_ERROR_CODE = '23505'
 
 /**
  * Get the latest chunk's metadata from the database
+ * @param {string} projectId
+ * @param {Object} [opts]
+ * @param {boolean} [opts.readOnly]
  */
-async function getLatestChunk(projectId) {
+async function getLatestChunk(projectId, opts = {}) {
   projectId = parseInt(projectId, 10)
   assert.integer(projectId, 'bad projectId')
+  const { readOnly = false } = opts
 
-  const record = await knex('chunks')
+  const record = await (readOnly ? knexReadOnly : knex)('chunks')
     .where('doc_id', projectId)
     .orderBy('end_version', 'desc')
     .first()

services/history-v1/storage/lib/knex_read_only.js

@@ -0,0 +1,19 @@
+'use strict'
+
+const config = require('config')
+const knexfile = require('../../knexfile')
+
+const env = process.env.NODE_ENV || 'development'
+
+if (config.databaseUrlReadOnly) {
+  module.exports = require('knex')({
+    ...knexfile[env],
+    pool: {
+      ...knexfile[env].pool,
+      min: 0,
+    },
+    connection: config.databaseUrlReadOnly,
+  })
+} else {
+  module.exports = require('./knex')
+}

services/history-v1/test/acceptance/js/api/projects.test.js

@@ -138,6 +138,23 @@ describe('project controller', function () {
         testFiles.STRING_AB_HASH
       )
     })
+    describe('getLatestHistoryRaw', function () {
+      it('should handles read', async function () {
+        const projectId = fixtures.docs.initializedProject.id
+        const response =
+          await testServer.pseudoJwtBasicAuthClient.apis.Project.getLatestHistoryRaw(
+            {
+              project_id: projectId,
+              readOnly: 'true',
+            }
+          )
+        expect(response.body).to.deep.equal({
+          startVersion: 0,
+          endVersion: 1,
+          endTimestamp: '2032-01-01T00:00:00.000Z',
+        })
+      })
+    })
   })
 
   describe('deleteProject', function () {

services/history-v1/test/acceptance/js/storage/chunk_store.test.js

@@ -69,7 +69,7 @@ describe('chunkStore', function () {
           await chunkStore.update(projectId, oldEndVersion, chunk)
         })
 
-        it('records the correct metadata in db', async function () {
+        it('records the correct metadata in db readOnly=false', async function () {
           const raw = await chunkStore.loadLatestRaw(projectId)
           expect(raw).to.deep.include({
             startVersion: 0,
@@ -78,6 +78,17 @@ describe('chunkStore', function () {
           })
         })
 
+        it('records the correct metadata in db readOnly=true', async function () {
+          const raw = await chunkStore.loadLatestRaw(projectId, {
+            readOnly: true,
+          })
+          expect(raw).to.deep.include({
+            startVersion: 0,
+            endVersion: 2,
+            endTimestamp: lastChangeTimestamp,
+          })
+        })
+
         it('records the correct timestamp', async function () {
           const chunk = await chunkStore.loadLatest(projectId)
           expect(chunk.getEndTimestamp()).to.deep.equal(lastChangeTimestamp)

services/history-v1/test/acceptance/pg-init/set-up-readOnly-user.sql

@@ -0,0 +1,2 @@
+CREATE USER read_only PASSWORD 'password';
+ALTER DEFAULT PRIVILEGES FOR USER overleaf IN SCHEMA public GRANT SELECT ON TABLES TO read_only;

services/project-history/app/js/HistoryStoreManager.js

@@ -90,15 +90,18 @@ export function getMostRecentVersion(projectId, historyId, callback) {
 /**
  * @param {string} projectId
  * @param {string} historyId
+ * @param {Object} opts
+ * @param {boolean} [opts.readOnly]
  * @param {(error: Error, rawChunk?: { startVersion: number, endVersion: number, endTimestamp: Date}) => void} callback
  */
-export function getMostRecentVersionRaw(projectId, historyId, callback) {
+export function getMostRecentVersionRaw(projectId, historyId, opts, callback) {
   const path = `projects/${historyId}/latest/history/raw`
   logger.debug(
     { projectId, historyId },
     'getting raw chunk from history service'
   )
-  _requestHistoryService({ path, json: true }, (err, body) => {
+  const qs = opts.readOnly ? { readOnly: true } : {}
+  _requestHistoryService({ path, json: true, qs }, (err, body) => {
     if (err) return callback(OError.tag(err))
     const { startVersion, endVersion, endTimestamp } = body
     callback(null, {

services/project-history/scripts/bulk_resync_file_fix_up.mjs

@@ -117,7 +117,8 @@ function checkFileTreeNeedsResync(folder) {
 async function getLastEndTimestamp(projectId, historyId) {
   const raw = await HistoryStoreManager.promises.getMostRecentVersionRaw(
     projectId,
-    historyId
+    historyId,
+    { readOnly: true }
   )
   if (!raw) throw new Error('bug: history not initialized')
   return raw.endTimestamp