From fbdf245517feed9e31a957bcc95e75795692cf57 Mon Sep 17 00:00:00 2001
From: Eric Mc Sween <5454374+emcsween@users.noreply.github.com>
Date: Fri, 5 Jul 2024 08:16:31 -0400
Subject: [PATCH] Merge pull request #19273 from
 overleaf/em-history-migration-concurrency

Add concurrency option to history ranges support migration script

GitOrigin-RevId: 8707abc9b76116090332b6abb11030adb17ceb4e
---
 .../History/HistoryRangesSupportMigration.js  | 114 ++++++++++++------
 .../scripts/history/migrate_ranges_support.js |   5 +
 2 files changed, 82 insertions(+), 37 deletions(-)

diff --git a/services/web/app/src/Features/History/HistoryRangesSupportMigration.js b/services/web/app/src/Features/History/HistoryRangesSupportMigration.js
index 48de0a1b8d..8c7a38141d 100644
--- a/services/web/app/src/Features/History/HistoryRangesSupportMigration.js
+++ b/services/web/app/src/Features/History/HistoryRangesSupportMigration.js
@@ -26,6 +26,7 @@ const {
  * @param {boolean} [opts.force]
  * @param {boolean} [opts.stopOnError]
  * @param {boolean} [opts.quickOnly]
+ * @param {number} [opts.concurrency]
  */
 async function migrateProjects(opts = {}) {
   const {
@@ -38,6 +39,7 @@ async function migrateProjects(opts = {}) {
     force = false,
     stopOnError = false,
     quickOnly = false,
+    concurrency = 1,
   } = opts
 
   const clauses = []
@@ -70,11 +72,31 @@ async function migrateProjects(opts = {}) {
     })
     .sort({ _id: -1 })
 
+  let terminating = false
+  const handleSignal = signal => {
+    logger.info({ signal }, 'History ranges support migration received signal')
+    terminating = true
+  }
+  process.on('SIGINT', handleSignal)
+  process.on('SIGTERM', handleSignal)
+
   let projectsProcessed = 0
+  const jobsByProjectId = new Map()
+  let errors = 0
+
   for await (const project of projects) {
     if (projectsProcessed >= maxCount) {
       break
     }
+
+    if (errors > 0 && stopOnError) {
+      break
+    }
+
+    if (terminating) {
+      break
+    }
+
     const projectId = project._id.toString()
 
     if (!force) {
@@ -89,45 +111,63 @@ async function migrateProjects(opts = {}) {
       }
     }
 
-    const startTimeMs = Date.now()
-    let quickMigrationSuccess
-    try {
-      quickMigrationSuccess = await quickMigration(projectId, direction)
-      if (!quickMigrationSuccess) {
-        if (quickOnly) {
-          logger.info(
-            { projectId, direction },
-            'Quick migration failed, skipping project'
-          )
-        } else {
-          await migrateProject(projectId, direction)
-        }
-      }
-    } catch (err) {
-      logger.error(
-        { err, projectId, direction, projectsProcessed },
-        'Failed to migrate history ranges support'
-      )
-      projectsProcessed += 1
-      if (stopOnError) {
-        break
-      } else {
-        continue
-      }
+    if (jobsByProjectId.size >= concurrency) {
+      // Wait until the next job finishes
+      await Promise.race(jobsByProjectId.values())
     }
-    const elapsedMs = Date.now() - startTimeMs
-    projectsProcessed += 1
-    logger.info(
-      {
-        projectId,
-        direction,
-        projectsProcessed,
-        elapsedMs,
-        quick: quickMigrationSuccess,
-      },
-      'Migrated history ranges support'
-    )
+
+    const job = processProject(projectId, direction, quickOnly)
+      .then(info => {
+        jobsByProjectId.delete(projectId)
+        projectsProcessed += 1
+        logger.info(
+          {
+            projectId,
+            direction,
+            projectsProcessed,
+            errors,
+            ...info,
+          },
+          'History ranges support migration'
+        )
+      })
+      .catch(err => {
+        jobsByProjectId.delete(projectId)
+        errors += 1
+        logger.error(
+          { err, projectId, direction, projectsProcessed, errors },
+          'Failed to migrate history ranges support'
+        )
+      })
+
+    jobsByProjectId.set(projectId, job)
   }
+
+  // Let the last jobs finish
+  await Promise.all(jobsByProjectId.values())
+}
+
+/**
+ * Migrate a single project
+ *
+ * @param {string} projectId
+ * @param {"forwards" | "backwards"} direction
+ * @param {boolean} quickOnly
+ */
+async function processProject(projectId, direction, quickOnly) {
+  const startTimeMs = Date.now()
+  const quickMigrationSuccess = await quickMigration(projectId, direction)
+  let migrationType
+  if (quickMigrationSuccess) {
+    migrationType = 'quick'
+  } else if (quickOnly) {
+    migrationType = 'skipped'
+  } else {
+    await migrateProject(projectId, direction)
+    migrationType = 'resync'
+  }
+  const elapsedMs = Date.now() - startTimeMs
+  return { migrationType, elapsedMs }
 }
 
 /**
diff --git a/services/web/scripts/history/migrate_ranges_support.js b/services/web/scripts/history/migrate_ranges_support.js
index 5d9d98acd7..616c31c808 100644
--- a/services/web/scripts/history/migrate_ranges_support.js
+++ b/services/web/scripts/history/migrate_ranges_support.js
@@ -14,6 +14,7 @@ async function main() {
     force,
     stopOnError,
     quickOnly,
+    concurrency,
   } = parseArgs()
   await HistoryRangesSupportMigration.promises.migrateProjects({
     projectIds,
@@ -25,6 +26,7 @@ async function main() {
     force,
     stopOnError,
     quickOnly,
+    concurrency,
   })
 }
 
@@ -44,6 +46,7 @@ Options:
     --force          Migrate projects even if they were already migrated
     --stop-on-error  Stop after first migration error
     --quick-only     Do not try a resync migration if quick migration fails
+    --concurrency    How many jobs to run in parallel
 `)
 }
 
@@ -67,6 +70,7 @@ function parseArgs() {
   const force = args.force
   const stopOnError = args['stop-on-error']
   const quickOnly = args['quick-only']
+  const concurrency = args.concurrency ?? 1
   const all = args.all
 
   if (
@@ -94,6 +98,7 @@ function parseArgs() {
     force,
     stopOnError,
     quickOnly,
+    concurrency,
   }
 }