|
| 1 | +'use strict'; |
| 2 | + |
| 3 | +const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms)); |
| 4 | +const MAX_ATTEMPTS = 10; |
| 5 | +const BATCH_SIZE = 100; |
| 6 | +const SLEEP_TIME_MS = 5000; |
| 7 | + |
| 8 | +/** @type {import('sequelize-cli').Migration} */ |
| 9 | +module.exports = { |
| 10 | + async up(queryInterface) { |
| 11 | + let totalDeleted = 0; |
| 12 | + let batchCount = 0; |
| 13 | + let attempts = 0; |
| 14 | + |
| 15 | + console.info(`Batch size: ${BATCH_SIZE} duplicate groups per batch`); |
| 16 | + |
| 17 | + console.info('Starting cleanup of duplicate backup folders...'); |
| 18 | + |
| 19 | + const deleteQuery = ` |
| 20 | + WITH duplicate_groups AS ( |
| 21 | + SELECT |
| 22 | + plain_name, |
| 23 | + bucket, |
| 24 | + user_id, |
| 25 | + MIN(id) as id_to_keep |
| 26 | + FROM folders |
| 27 | + WHERE |
| 28 | + created_at >= '2025-12-17 14:16:00' |
| 29 | + AND created_at <= '2026-01-05 21:50:00' |
| 30 | + AND parent_id IS NULL |
| 31 | + AND parent_uuid IS NULL |
| 32 | + AND deleted = false |
| 33 | + AND removed = false |
| 34 | + AND plain_name IS NOT NULL |
| 35 | + GROUP BY plain_name, bucket, user_id |
| 36 | + HAVING COUNT(*) > 1 |
| 37 | + LIMIT ${BATCH_SIZE} |
| 38 | + ), |
| 39 | + folders_to_delete AS ( |
| 40 | + SELECT f.id |
| 41 | + FROM folders f |
| 42 | + INNER JOIN duplicate_groups dg |
| 43 | + ON f.plain_name = dg.plain_name |
| 44 | + AND f.bucket = dg.bucket |
| 45 | + AND f.user_id = dg.user_id |
| 46 | + WHERE |
| 47 | + f.id != dg.id_to_keep |
| 48 | + AND NOT EXISTS ( |
| 49 | + SELECT 1 |
| 50 | + FROM files |
| 51 | + WHERE folder_id = f.id |
| 52 | + AND deleted = false |
| 53 | + ) |
| 54 | + ) |
| 55 | + UPDATE folders |
| 56 | + SET |
| 57 | + deleted = true, |
| 58 | + deleted_at = NOW(), |
| 59 | + removed = true, |
| 60 | + removed_at = NOW() |
| 61 | + FROM folders_to_delete |
| 62 | + WHERE folders.id = folders_to_delete.id |
| 63 | + AND folders.deleted = false |
| 64 | + RETURNING folders.id; |
| 65 | + `; |
| 66 | + |
| 67 | + let hasMore = true; |
| 68 | + |
| 69 | + while (hasMore) { |
| 70 | + try { |
| 71 | + const [results] = await queryInterface.sequelize.query(deleteQuery); |
| 72 | + const deletedInBatch = results.length; |
| 73 | + batchCount++; |
| 74 | + totalDeleted += deletedInBatch; |
| 75 | + attempts = 0; |
| 76 | + |
| 77 | + console.info( |
| 78 | + `Batch ${batchCount}: Deleted ${deletedInBatch} folders (Total: ${totalDeleted})`, |
| 79 | + ); |
| 80 | + |
| 81 | + hasMore = deletedInBatch > 0; |
| 82 | + |
| 83 | + if (hasMore) { |
| 84 | + await sleep(SLEEP_TIME_MS); |
| 85 | + } |
| 86 | + } catch (err) { |
| 87 | + attempts++; |
| 88 | + console.error( |
| 89 | + `[ERROR]: Error in batch ${batchCount} (attempt ${attempts}/${MAX_ATTEMPTS}): ${err.message}`, |
| 90 | + ); |
| 91 | + |
| 92 | + if (attempts >= MAX_ATTEMPTS) { |
| 93 | + console.error( |
| 94 | + '[ERROR]: Maximum retry attempts reached, exiting migration.', |
| 95 | + ); |
| 96 | + break; |
| 97 | + } |
| 98 | + |
| 99 | + await sleep(SLEEP_TIME_MS); |
| 100 | + } |
| 101 | + } |
| 102 | + |
| 103 | + console.info('\n=== Cleanup Complete ==='); |
| 104 | + console.info(`Total batches processed: ${batchCount}`); |
| 105 | + console.info(`Total folders deleted: ${totalDeleted}`); |
| 106 | + }, |
| 107 | + async down() {}, |
| 108 | +}; |
0 commit comments