Add options for controlling hint & retries to history upgrade scripts (#5967)

* Add USE_QUERY_HINT to allow configuring hint usage in history scripts

* Add RETRY_FAILED option to re-attempt previously failed history upgrades

* Add HistoryConversionFailed count to counts script

* Increase default batch size/concurrency on count script

GitOrigin-RevId: a498294719a504b224c04c95363bf098bb80b574

services/web/scripts/history/count_project_history_categories.js

@@ -1,7 +1,8 @@
 const VERBOSE_LOGGING = process.env.VERBOSE_LOGGING === 'true'
 const VERBOSE_PROJECT_NAMES = process.env.VERBOSE_PROJECT_NAMES === 'true'
-const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 5
-const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 100
+const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 50
+const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 500
+const USE_QUERY_HINT = process.env.USE_QUERY_HINT !== 'false'
 // persist fallback in order to keep batchedUpdate in-sync
 process.env.BATCH_SIZE = BATCH_SIZE
 // raise mongo timeout to 1hr if otherwise unspecified
@@ -21,6 +22,7 @@ const COUNT = {
   NoneWithConversion: 0,
   NoneWithTemporaryHistory: 0,
   HistoryUpgradeFailed: 0,
+  HistoryConversionFailed: 0,
 }
 
 // Timestamp of when 'Enable history for SL in background' release
@@ -59,6 +61,15 @@ async function processProject(project) {
           } has a history upgrade failure recorded`
         )
       }
+    } else if (project.overleaf.history.conversionFailed) {
+      COUNT.HistoryConversionFailed += 1
+      if (VERBOSE_LOGGING) {
+        console.log(
+          `project ${
+            project[VERBOSE_PROJECT_NAMES ? 'name' : '_id']
+          } has a history upgrade failure recorded`
+        )
+      }
     } else if (project.overleaf.history.display) {
       // v2: full project history, do nothing, (query shoudln't include any, but we should stlll check?)
       COUNT.v2 += 1
@@ -231,8 +242,9 @@ async function main() {
     _id: 1,
     overleaf: 1,
   }
-  const options = {
-    hint: { _id: 1 },
+  const options = {}
+  if (USE_QUERY_HINT) {
+    options.hint = { _id: 1 }
   }
   if (VERBOSE_PROJECT_NAMES) {
     projection.name = 1

services/web/scripts/history/upgrade_none_with_conversion_if_sl_history.js

@@ -1,8 +1,10 @@
-const SCRIPT_VERSION = 1
+const SCRIPT_VERSION = 2
 const VERBOSE_LOGGING = process.env.VERBOSE_LOGGING === 'true'
 const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
 const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 100
 const DRY_RUN = process.env.DRY_RUN !== 'false'
+const USE_QUERY_HINT = process.env.USE_QUERY_HINT !== 'false'
+const RETRY_FAILED = process.env.RETRY_FAILED === 'true'
 const MAX_UPGRADES_TO_ATTEMPT =
   parseInt(process.env.MAX_UPGRADES_TO_ATTEMPT, 10) || false
 const MAX_FAILURES = parseInt(process.env.MAX_FAILURES, 10) || 50
@@ -25,6 +27,8 @@ console.log({
   BATCH_SIZE,
   MAX_UPGRADES_TO_ATTEMPT,
   MAX_FAILURES,
+  USE_QUERY_HINT,
+  RETRY_FAILED,
 })
 
 const RESULT = {
@@ -64,15 +68,17 @@ async function processProject(project) {
   if (INTERRUPT) {
     return
   }
-  // safety check
-  if (project.overleaf && project.overleaf.history) {
-    if (
-      project.overleaf.history.conversionFailed ||
-      project.overleaf.history.upgradeFailed
-    ) {
-      // we don't want to attempt upgrade on projects
-      // that have been previously attempted and failed
-      return
+  // skip safety check if we want to retry failed upgrades
+  if (!RETRY_FAILED) {
+    if (project.overleaf && project.overleaf.history) {
+      if (
+        project.overleaf.history.conversionFailed ||
+        project.overleaf.history.upgradeFailed
+      ) {
+        // we don't want to attempt upgrade on projects
+        // that have been previously attempted and failed
+        return
+      }
     }
   }
   const anyDocHistory = await anyDocHistoryExists(project)
@@ -122,6 +128,10 @@ async function doUpgradeForNoneWithConversion(project) {
         $set: {
           'overleaf.history.upgradeReason': `none-with-conversion/${SCRIPT_VERSION}`,
         },
+        $unset: {
+          'overleaf.history.upgradeFailed': true,
+          'overleaf.history.conversionFailed': true,
+        },
       }
     )
   }
@@ -158,8 +168,9 @@ async function main() {
     _id: 1,
     overleaf: 1,
   }
-  const options = {
-    hint: { _id: 1 },
+  const options = {}
+  if (USE_QUERY_HINT) {
+    options.hint = { _id: 1 }
   }
   await batchedUpdate(
     'projects',

services/web/scripts/history/upgrade_none_without_conversion_if_no_sl_history.js

@@ -1,8 +1,10 @@
-const SCRIPT_VERSION = 1
+const SCRIPT_VERSION = 2
 const VERBOSE_LOGGING = process.env.VERBOSE_LOGGING === 'true'
 const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
 const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 100
 const DRY_RUN = process.env.DRY_RUN !== 'false'
+const USE_QUERY_HINT = process.env.USE_QUERY_HINT !== 'false'
+const RETRY_FAILED = process.env.RETRY_FAILED === 'true'
 const MAX_UPGRADES_TO_ATTEMPT =
   parseInt(process.env.MAX_UPGRADES_TO_ATTEMPT, 10) || false
 const MAX_FAILURES = parseInt(process.env.MAX_FAILURES, 10) || 50
@@ -26,6 +28,8 @@ console.log({
   BATCH_SIZE,
   MAX_UPGRADES_TO_ATTEMPT,
   MAX_FAILURES,
+  USE_QUERY_HINT,
+  RETRY_FAILED,
 })
 
 const RESULT = {
@@ -65,6 +69,15 @@ async function processProject(project) {
   if (INTERRUPT) {
     return
   }
+  if (!RETRY_FAILED) {
+    if (
+      project.overleaf &&
+      project.overleaf.history &&
+      project.overleaf.history.upgradeFailed
+    ) {
+      return
+    }
+  }
   const anyDocHistory = await anyDocHistoryExists(project)
   if (anyDocHistory) {
     return
@@ -96,21 +109,20 @@ async function doUpgradeForNoneWithoutConversion(project) {
       const historyId = await ProjectHistoryHandler.promises.getHistoryId(
         projectId
       )
-      if (historyId != null) {
-        return
-      }
-      const history = await HistoryManager.promises.initializeProject()
-      if (history && history.overleaf_id) {
-        await ProjectHistoryHandler.promises.setHistoryId(
-          projectId,
-          history.overleaf_id
-        )
-        await HistoryManager.promises.resyncProject(projectId, {
-          force: true,
-          origin: { kind: 'history-migration' },
-        })
-        await HistoryManager.promises.flushProject(projectId)
+      if (!historyId) {
+        const history = await HistoryManager.promises.initializeProject()
+        if (history && history.overleaf_id) {
+          await ProjectHistoryHandler.promises.setHistoryId(
+            projectId,
+            history.overleaf_id
+          )
+        }
       }
+      await HistoryManager.promises.resyncProject(projectId, {
+        force: true,
+        origin: { kind: 'history-migration' },
+      })
+      await HistoryManager.promises.flushProject(projectId)
     } catch (err) {
       RESULT.failed += 1
       console.error(`project ${project._id} FAILED with error: `, err)
@@ -132,6 +144,9 @@ async function doUpgradeForNoneWithoutConversion(project) {
           'overleaf.history.upgradedAt': new Date(),
           'overleaf.history.upgradeReason': `none-without-sl-history/${SCRIPT_VERSION}`,
         },
+        $unset: {
+          'overleaf.history.upgradeFailed': true,
+        },
       }
     )
   }
@@ -166,8 +181,9 @@ async function main() {
     _id: 1,
     overleaf: 1,
   }
-  const options = {
-    hint: { _id: 1 },
+  const options = {}
+  if (USE_QUERY_HINT) {
+    options.hint = { _id: 1 }
   }
   await batchedUpdate(
     'projects',

services/web/scripts/history/upgrade_v1_without_conversion_if_created_after_fph_enabled.js

@@ -3,6 +3,7 @@ const VERBOSE_LOGGING = process.env.VERBOSE_LOGGING === 'true'
 const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
 const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 100
 const DRY_RUN = process.env.DRY_RUN !== 'false'
+const USE_QUERY_HINT = process.env.USE_QUERY_HINT !== 'false'
 // persist fallback in order to keep batchedUpdate in-sync
 process.env.BATCH_SIZE = BATCH_SIZE
 // raise mongo timeout to 1hr if otherwise unspecified
@@ -14,7 +15,13 @@ const { db } = require('../../app/src/infrastructure/mongodb')
 const { promiseMapWithLimit } = require('../../app/src/util/promises')
 const { batchedUpdate } = require('../helpers/batchedUpdate')
 
-console.log({ DRY_RUN, VERBOSE_LOGGING, WRITE_CONCURRENCY, BATCH_SIZE })
+console.log({
+  DRY_RUN,
+  VERBOSE_LOGGING,
+  WRITE_CONCURRENCY,
+  BATCH_SIZE,
+  USE_QUERY_HINT,
+})
 
 const RESULT = {
   DRY_RUN,
@@ -123,8 +130,9 @@ async function main() {
     _id: 1,
     overleaf: 1,
   }
-  const options = {
-    hint: { _id: 1 },
+  const options = {}
+  if (USE_QUERY_HINT) {
+    options.hint = { _id: 1 }
   }
   await batchedUpdate(
     'projects',

services/web/scripts/history/upgrade_v1_without_conversion_if_no_sl_history.js

@@ -3,6 +3,7 @@ const VERBOSE_LOGGING = process.env.VERBOSE_LOGGING === 'true'
 const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
 const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 100
 const DRY_RUN = process.env.DRY_RUN !== 'false'
+const USE_QUERY_HINT = process.env.USE_QUERY_HINT !== 'false'
 // persist fallback in order to keep batchedUpdate in-sync
 process.env.BATCH_SIZE = BATCH_SIZE
 // raise mongo timeout to 1hr if otherwise unspecified
@@ -14,7 +15,13 @@ const { db } = require('../../app/src/infrastructure/mongodb')
 const { promiseMapWithLimit } = require('../../app/src/util/promises')
 const { batchedUpdate } = require('../helpers/batchedUpdate')
 
-console.log({ DRY_RUN, VERBOSE_LOGGING, WRITE_CONCURRENCY, BATCH_SIZE })
+console.log({
+  DRY_RUN,
+  VERBOSE_LOGGING,
+  WRITE_CONCURRENCY,
+  BATCH_SIZE,
+  USE_QUERY_HINT,
+})
 
 const RESULT = {
   DRY_RUN,
@@ -111,8 +118,9 @@ async function main() {
     _id: 1,
     overleaf: 1,
   }
-  const options = {
-    hint: { _id: 1 },
+  const options = {}
+  if (USE_QUERY_HINT) {
+    options.hint = { _id: 1 }
   }
   await batchedUpdate(
     'projects',