Support object storage in prune script

Also prune original files and user exports
This commit is contained in:
Chocobozzz 2024-06-03 16:37:44 +02:00
parent 568a1b1e85
commit 54c140c800
No known key found for this signature in database
GPG Key ID: 583A612D890159BE
12 changed files with 618 additions and 295 deletions

View File

@ -53,7 +53,7 @@
"create-move-video-storage-job": "node ./dist/scripts/create-move-video-storage-job.js",
"create-generate-storyboard-job": "node ./dist/scripts/create-generate-storyboard-job.js",
"parse-log": "node ./dist/scripts/parse-log.js",
"prune-storage": "node ./dist/scripts/prune-storage.js",
"prune-storage": "LOGGER_LEVEL=warn node ./dist/scripts/prune-storage.js",
"test": "bash ./scripts/test.sh",
"generate-cli-doc": "bash ./scripts/generate-cli-doc.sh",
"generate-types-package": "tsx --conditions=peertube:tsx ./packages/types-generator/generate-package.ts",

View File

@ -1,77 +1,49 @@
/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
import { expect } from 'chai'
import { createFile } from 'fs-extra/esm'
import { readdir } from 'fs/promises'
import { join } from 'path'
import { wait } from '@peertube/peertube-core-utils'
import { buildUUID } from '@peertube/peertube-node-utils'
import { HttpStatusCode, VideoPlaylistPrivacy, VideoPrivacy } from '@peertube/peertube-models'
import { getAllFiles, wait } from '@peertube/peertube-core-utils'
import { FileStorage, HttpStatusCode, HttpStatusCodeType, VideoPlaylistPrivacy, VideoPrivacy } from '@peertube/peertube-models'
import { areMockObjectStorageTestsDisabled, buildUUID } from '@peertube/peertube-node-utils'
import {
cleanupTests,
CLICommand,
ObjectStorageCommand,
PeerTubeServer,
cleanupTests,
createMultipleServers,
doubleFollow,
killallServers,
makeGetRequest,
PeerTubeServer,
makeRawRequest,
setAccessTokensToServers,
setDefaultVideoChannel,
waitJobs
} from '@peertube/peertube-server-commands'
async function assertNotExists (server: PeerTubeServer, directory: string, substring: string) {
const files = await readdir(server.servers.buildDirectory(directory))
for (const f of files) {
expect(f).to.not.contain(substring)
}
}
async function assertCountAreOkay (servers: PeerTubeServer[]) {
for (const server of servers) {
const videosCount = await server.servers.countFiles('web-videos')
expect(videosCount).to.equal(9) // 2 videos with 4 resolutions + private directory
const privateVideosCount = await server.servers.countFiles('web-videos/private')
expect(privateVideosCount).to.equal(4)
const torrentsCount = await server.servers.countFiles('torrents')
expect(torrentsCount).to.equal(24)
const previewsCount = await server.servers.countFiles('previews')
expect(previewsCount).to.equal(3)
const thumbnailsCount = await server.servers.countFiles('thumbnails')
expect(thumbnailsCount).to.equal(5) // 3 local videos, 1 local playlist, 2 remotes videos (lazy downloaded) and 1 remote playlist
const avatarsCount = await server.servers.countFiles('avatars')
expect(avatarsCount).to.equal(8)
const hlsRootCount = await server.servers.countFiles(join('streaming-playlists', 'hls'))
expect(hlsRootCount).to.equal(3) // 2 videos + private directory
const hlsPrivateRootCount = await server.servers.countFiles(join('streaming-playlists', 'hls', 'private'))
expect(hlsPrivateRootCount).to.equal(1)
}
}
import { SQLCommand } from '@tests/shared/sql-command.js'
import { expect } from 'chai'
import { createFile } from 'fs-extra/esm'
import { readdir } from 'fs/promises'
import { join } from 'path'
describe('Test prune storage scripts', function () {
let servers: PeerTubeServer[]
const badNames: { [directory: string]: string[] } = {}
before(async function () {
this.timeout(120000)
servers = await createMultipleServers(2, { transcoding: { enabled: true } })
servers = await createMultipleServers(2)
await setAccessTokensToServers(servers)
await setDefaultVideoChannel(servers)
for (const server of servers) {
await server.videos.upload({ attributes: { name: 'video 1', privacy: VideoPrivacy.PUBLIC } })
await server.videos.upload({ attributes: { name: 'video 2', privacy: VideoPrivacy.PUBLIC } })
await server.config.enableMinimumTranscoding({ keepOriginal: true })
await server.config.enableUserExport()
}
await server.videos.upload({ attributes: { name: 'video 3', privacy: VideoPrivacy.PRIVATE } })
for (const server of servers) {
await server.videos.quickUpload({ name: 'video 1', privacy: VideoPrivacy.PUBLIC })
await server.videos.quickUpload({ name: 'video 2', privacy: VideoPrivacy.PUBLIC })
await server.videos.quickUpload({ name: 'video 3', privacy: VideoPrivacy.PRIVATE })
await server.users.updateMyAvatar({ fixture: 'avatar.png' })
@ -85,6 +57,12 @@ describe('Test prune storage scripts', function () {
})
}
for (const server of servers) {
const user = await server.users.getMyInfo()
await server.userExports.request({ userId: user.id, withVideoFiles: false })
}
await doubleFollow(servers[0], servers[1])
// Lazy load the remote avatars
@ -119,8 +97,53 @@ describe('Test prune storage scripts', function () {
await wait(1000)
})
describe('On filesystem', function () {
const badNames: { [directory: string]: string[] } = {}
async function assertNotExists (server: PeerTubeServer, directory: string, substring: string) {
const files = await readdir(server.servers.buildDirectory(directory))
for (const f of files) {
expect(f).to.not.contain(substring)
}
}
async function assertCountAreOkay () {
for (const server of servers) {
const videosCount = await server.servers.countFiles('web-videos')
expect(videosCount).to.equal(5) // 2 videos with 2 resolutions + private directory
const privateVideosCount = await server.servers.countFiles('web-videos/private')
expect(privateVideosCount).to.equal(2)
const torrentsCount = await server.servers.countFiles('torrents')
expect(torrentsCount).to.equal(12)
const previewsCount = await server.servers.countFiles('previews')
expect(previewsCount).to.equal(3)
const thumbnailsCount = await server.servers.countFiles('thumbnails')
expect(thumbnailsCount).to.equal(5) // 3 local videos, 1 local playlist, 2 remotes videos (lazy downloaded) and 1 remote playlist
const avatarsCount = await server.servers.countFiles('avatars')
expect(avatarsCount).to.equal(8)
const hlsRootCount = await server.servers.countFiles(join('streaming-playlists', 'hls'))
expect(hlsRootCount).to.equal(3) // 2 videos + private directory
const hlsPrivateRootCount = await server.servers.countFiles(join('streaming-playlists', 'hls', 'private'))
expect(hlsPrivateRootCount).to.equal(1)
const originalVideoFilesCount = await server.servers.countFiles(join('original-video-files'))
expect(originalVideoFilesCount).to.equal(3)
const userExportFilesCount = await server.servers.countFiles(join('tmp-persistent'))
expect(userExportFilesCount).to.equal(1)
}
}
it('Should have the files on the disk', async function () {
await assertCountAreOkay(servers)
await assertCountAreOkay()
})
it('Should create some dirty files', async function () {
@ -198,6 +221,27 @@ describe('Test prune storage scripts', function () {
await createFile(join(basePrivate, n1))
badNames[directory] = [ n1 ]
}
{
const base = servers[0].servers.buildDirectory('original-video-files')
const n1 = buildUUID() + '.mp4'
await createFile(join(base, n1))
badNames['original-video-files'] = [ n1 ]
}
{
const base = servers[0].servers.buildDirectory('tmp-persistent')
const n1 = 'user-export-1.zip'
const n2 = 'user-export-2.zip'
await createFile(join(base, n1))
await createFile(join(base, n2))
badNames['tmp-persistent'] = [ n1, n2 ]
}
}
})
@ -209,7 +253,7 @@ describe('Test prune storage scripts', function () {
})
it('Should have removed files', async function () {
await assertCountAreOkay(servers)
await assertCountAreOkay()
for (const directory of Object.keys(badNames)) {
for (const name of badNames[directory]) {
@ -217,6 +261,96 @@ describe('Test prune storage scripts', function () {
}
}
})
})
describe('On object storage', function () {
if (areMockObjectStorageTestsDisabled()) return
const videos: string[] = []
const objectStorage = new ObjectStorageCommand()
let sqlCommand: SQLCommand
let rootId: number
async function checkVideosFiles (uuids: string[], expectedStatus: HttpStatusCodeType) {
for (const uuid of uuids) {
const video = await servers[0].videos.getWithToken({ id: uuid })
for (const file of getAllFiles(video)) {
await makeRawRequest({ url: file.fileUrl, token: servers[0].accessToken, expectedStatus })
}
const source = await servers[0].videos.getSource({ id: uuid })
await makeRawRequest({ url: source.fileDownloadUrl, redirects: 1, token: servers[0].accessToken, expectedStatus })
}
}
async function checkUserExport (expectedStatus: HttpStatusCodeType) {
const { data } = await servers[0].userExports.list({ userId: rootId })
await makeRawRequest({ url: data[0].privateDownloadUrl, redirects: 1, expectedStatus })
}
before(async function () {
this.timeout(120000)
sqlCommand = new SQLCommand(servers[0])
await objectStorage.prepareDefaultMockBuckets()
await servers[0].run(objectStorage.getDefaultMockConfig())
{
const { uuid } = await servers[0].videos.quickUpload({ name: 's3 video 1', privacy: VideoPrivacy.PUBLIC })
videos.push(uuid)
}
{
const { uuid } = await servers[0].videos.quickUpload({ name: 's3 video 2', privacy: VideoPrivacy.PUBLIC })
videos.push(uuid)
}
{
const { uuid } = await servers[0].videos.quickUpload({ name: 's3 video 3', privacy: VideoPrivacy.PRIVATE })
videos.push(uuid)
}
const user = await servers[0].users.getMyInfo()
rootId = user.id
await servers[0].userExports.deleteAllArchives({ userId: rootId })
await servers[0].userExports.request({ userId: rootId, withVideoFiles: false })
await waitJobs([ servers[0] ])
})
it('Should have the files on object storage', async function () {
await checkVideosFiles(videos, HttpStatusCode.OK_200)
await checkUserExport(HttpStatusCode.OK_200)
})
it('Should run prune-storage script on videos', async function () {
await sqlCommand.setVideoFileStorageOf(videos[1], FileStorage.FILE_SYSTEM)
await sqlCommand.setVideoFileStorageOf(videos[2], FileStorage.FILE_SYSTEM)
await checkVideosFiles([ videos[1], videos[2] ], HttpStatusCode.NOT_FOUND_404)
await checkVideosFiles([ videos[0] ], HttpStatusCode.OK_200)
await checkUserExport(HttpStatusCode.OK_200)
})
it('Should run prune-storage script on exports', async function () {
await sqlCommand.setUserExportStorageOf(rootId, FileStorage.FILE_SYSTEM)
await checkVideosFiles([ videos[1], videos[2] ], HttpStatusCode.NOT_FOUND_404)
await checkVideosFiles([ videos[0] ], HttpStatusCode.OK_200)
await checkUserExport(HttpStatusCode.NOT_FOUND_404)
})
after(async function () {
await sqlCommand.cleanup()
})
})
after(async function () {
await cleanupTests(servers)

View File

@ -1,6 +1,7 @@
import { QueryTypes, Sequelize } from 'sequelize'
import { forceNumber } from '@peertube/peertube-core-utils'
import { PeerTubeServer } from '@peertube/peertube-server-commands'
import { FileStorageType } from '@peertube/peertube-models'
export class SQLCommand {
private sequelize: Sequelize
@ -58,6 +59,30 @@ export class SQLCommand {
// ---------------------------------------------------------------------------
async setVideoFileStorageOf (uuid: string, storage: FileStorageType) {
await this.updateQuery(
`UPDATE "videoFile" SET storage = :storage ` +
`WHERE "videoId" IN (SELECT id FROM "video" WHERE uuid = :uuid) OR ` +
// eslint-disable-next-line max-len
`"videoStreamingPlaylistId" IN (` +
`SELECT "videoStreamingPlaylist".id FROM "videoStreamingPlaylist" ` +
`INNER JOIN video ON video.id = "videoStreamingPlaylist"."videoId" AND "video".uuid = :uuid` +
`)`,
{ storage, uuid }
)
await this.updateQuery(
`UPDATE "videoSource" SET storage = :storage WHERE "videoId" IN (SELECT id FROM "video" WHERE uuid = :uuid)`,
{ storage, uuid }
)
}
async setUserExportStorageOf (userId: number, storage: FileStorageType) {
await this.updateQuery(`UPDATE "userExport" SET storage = :storage WHERE "userId" = :userId`, { storage, userId })
}
// ---------------------------------------------------------------------------
setPluginVersion (pluginName: string, newVersion: string) {
return this.setPluginField(pluginName, 'version', newVersion)
}

View File

@ -1,11 +1,11 @@
import { context } from '@opentelemetry/api'
import { getSpanContext } from '@opentelemetry/api/build/src/trace/context-utils.js'
import { omit } from '@peertube/peertube-core-utils'
import { stat } from 'fs/promises'
import { join } from 'path'
import { format as sqlFormat } from 'sql-formatter'
import { createLogger, format, transports } from 'winston'
import { FileTransportOptions } from 'winston/lib/winston/transports'
import { context } from '@opentelemetry/api'
import { getSpanContext } from '@opentelemetry/api/build/src/trace/context-utils.js'
import { omit } from '@peertube/peertube-core-utils'
import { CONFIG } from '../initializers/config.js'
import { LOG_FILENAME } from '../initializers/constants.js'
@ -60,7 +60,7 @@ if (CONFIG.LOG.ROTATION.ENABLED) {
function buildLogger (labelSuffix?: string) {
return createLogger({
level: CONFIG.LOG.LEVEL,
level: process.env.LOGGER_LEVEL ?? CONFIG.LOG.LEVEL,
defaultMeta: {
get traceId () { return getSpanContext(context.active())?.traceId },
get spanId () { return getSpanContext(context.active())?.spanId },
@ -154,18 +154,10 @@ async function mtimeSortFilesDesc (files: string[], basePath: string) {
// ---------------------------------------------------------------------------
export {
type LoggerTagsFn,
type LoggerTags,
buildLogger,
timestampFormatter,
labelFormatter,
consoleLoggerFormat,
jsonLoggerFormat,
mtimeSortFilesDesc,
logger,
loggerTagsFactory,
bunyanLogger
buildLogger, bunyanLogger, consoleLoggerFormat,
jsonLoggerFormat, labelFormatter, logger,
loggerTagsFactory, mtimeSortFilesDesc, timestampFormatter, type LoggerTags, type LoggerTagsFn
}
// ---------------------------------------------------------------------------

View File

@ -845,6 +845,7 @@ const NSFW_POLICY_TYPES: { [ id: string ]: NSFWPolicyType } = {
// ---------------------------------------------------------------------------
const USER_EXPORT_MAX_ITEMS = 1000
const USER_EXPORT_FILE_PREFIX = 'user-export-'
// ---------------------------------------------------------------------------
@ -1268,6 +1269,7 @@ export {
CONSTRAINTS_FIELDS,
EMBED_SIZE,
REDUNDANCY,
USER_EXPORT_FILE_PREFIX,
JOB_CONCURRENCY,
JOB_ATTEMPTS,
AP_CLEANER,

View File

@ -7,9 +7,9 @@ import { createReadStream, createWriteStream } from 'fs'
import { ensureDir } from 'fs-extra/esm'
import { dirname } from 'path'
import { Readable } from 'stream'
import { getInternalUrl } from '../urls.js'
import { getClient } from './client.js'
import { lTags } from './logger.js'
import { getInternalUrl } from './urls.js'
import { getClient } from './shared/client.js'
import { lTags } from './shared/logger.js'
import type { _Object, ObjectCannedACL, PutObjectCommandInput, S3Client } from '@aws-sdk/client-s3'
@ -18,7 +18,7 @@ type BucketInfo = {
PREFIX?: string
}
async function listKeysOfPrefix (prefix: string, bucketInfo: BucketInfo) {
async function listKeysOfPrefix (prefix: string, bucketInfo: BucketInfo, continuationToken?: string) {
const s3Client = await getClient()
const { ListObjectsV2Command } = await import('@aws-sdk/client-s3')
@ -26,14 +26,21 @@ async function listKeysOfPrefix (prefix: string, bucketInfo: BucketInfo) {
const commandPrefix = bucketInfo.PREFIX + prefix
const listCommand = new ListObjectsV2Command({
Bucket: bucketInfo.BUCKET_NAME,
Prefix: commandPrefix
Prefix: commandPrefix,
ContinuationToken: continuationToken
})
const listedObjects = await s3Client.send(listCommand)
if (isArray(listedObjects.Contents) !== true) return []
return listedObjects.Contents.map(c => c.Key)
let keys = listedObjects.Contents.map(c => c.Key)
if (listedObjects.IsTruncated) {
keys = keys.concat(await listKeysOfPrefix(prefix, bucketInfo, listedObjects.NextContinuationToken))
}
return keys
}
// ---------------------------------------------------------------------------
@ -145,7 +152,7 @@ function removeObject (objectStorageKey: string, bucketInfo: BucketInfo) {
return removeObjectByFullKey(key, bucketInfo)
}
async function removeObjectByFullKey (fullKey: string, bucketInfo: BucketInfo) {
async function removeObjectByFullKey (fullKey: string, bucketInfo: Pick<BucketInfo, 'BUCKET_NAME'>) {
logger.debug('Removing file %s in bucket %s', fullKey, bucketInfo.BUCKET_NAME, lTags())
const { DeleteObjectCommand } = await import('@aws-sdk/client-s3')

View File

@ -1,3 +1,3 @@
export * from './client.js'
export * from './logger.js'
export * from './object-storage-helpers.js'
export * from '../object-storage-helpers.js'

View File

@ -1,23 +1,25 @@
import { FindOptions, Op } from 'sequelize'
import { AllowNull, BeforeDestroy, BelongsTo, Column, CreatedAt, DataType, ForeignKey, Table, UpdatedAt } from 'sequelize-typescript'
import { MUserAccountId, MUserExport } from '@server/types/models/index.js'
import { UserModel } from './user.js'
import { getSort } from '../shared/sort.js'
import { UserExportState, type UserExport, type UserExportStateType, type FileStorageType, FileStorage } from '@peertube/peertube-models'
import { FileStorage, UserExportState, type FileStorageType, type UserExport, type UserExportStateType } from '@peertube/peertube-models'
import { logger } from '@server/helpers/logger.js'
import { remove } from 'fs-extra/esm'
import { getFSUserExportFilePath } from '@server/lib/paths.js'
import { CONFIG } from '@server/initializers/config.js'
import {
JWT_TOKEN_USER_EXPORT_FILE_LIFETIME,
STATIC_DOWNLOAD_PATHS,
USER_EXPORT_FILE_PREFIX,
USER_EXPORT_STATES,
WEBSERVER
} from '@server/initializers/constants.js'
import { join } from 'path'
import jwt from 'jsonwebtoken'
import { CONFIG } from '@server/initializers/config.js'
import { removeUserExportObjectStorage } from '@server/lib/object-storage/user-export.js'
import { getFSUserExportFilePath } from '@server/lib/paths.js'
import { MUserAccountId, MUserExport } from '@server/types/models/index.js'
import { remove } from 'fs-extra/esm'
import jwt from 'jsonwebtoken'
import { join } from 'path'
import { FindOptions, Op } from 'sequelize'
import { AllowNull, BeforeDestroy, BelongsTo, Column, CreatedAt, DataType, ForeignKey, Table, UpdatedAt } from 'sequelize-typescript'
import { doesExist } from '../shared/query.js'
import { SequelizeModel } from '../shared/sequelize-type.js'
import { getSort } from '../shared/sort.js'
import { UserModel } from './user.js'
@Table({
tableName: 'userExport',
@ -147,11 +149,20 @@ export class UserExportModel extends SequelizeModel<UserExportModel> {
// ---------------------------------------------------------------------------
static async doesOwnedFileExist (filename: string, storage: FileStorageType) {
const query = 'SELECT 1 FROM "userExport" ' +
`WHERE "filename" = $filename AND "storage" = $storage LIMIT 1`
return doesExist({ sequelize: this.sequelize, query, bind: { filename, storage } })
}
// ---------------------------------------------------------------------------
generateAndSetFilename () {
if (!this.userId) throw new Error('Cannot generate filename without userId')
if (!this.createdAt) throw new Error('Cannot generate filename without createdAt')
this.filename = `user-export-${this.userId}-${this.createdAt.toISOString()}.zip`
this.filename = `${USER_EXPORT_FILE_PREFIX}${this.userId}-${this.createdAt.toISOString()}.zip`
}
canBeSafelyRemoved () {

View File

@ -289,11 +289,11 @@ export class VideoFileModel extends SequelizeModel<VideoFileModel> {
return doesExist({ sequelize: this.sequelize, query, bind: { filename } })
}
static async doesOwnedWebVideoFileExist (filename: string) {
static async doesOwnedFileExist (filename: string, storage: FileStorageType) {
const query = 'SELECT 1 FROM "videoFile" INNER JOIN "video" ON "video"."id" = "videoFile"."videoId" AND "video"."remote" IS FALSE ' +
`WHERE "filename" = $filename AND "storage" = ${FileStorage.FILE_SYSTEM} LIMIT 1`
`WHERE "filename" = $filename AND "storage" = $storage LIMIT 1`
return doesExist({ sequelize: this.sequelize, query, bind: { filename } })
return doesExist({ sequelize: this.sequelize, query, bind: { filename, storage } })
}
static loadByFilename (filename: string) {

View File

@ -1,12 +1,12 @@
import type { FileStorageType, VideoSource } from '@peertube/peertube-models'
import { type FileStorageType, type VideoSource } from '@peertube/peertube-models'
import { STATIC_DOWNLOAD_PATHS, WEBSERVER } from '@server/initializers/constants.js'
import { MVideoSource } from '@server/types/models/video/video-source.js'
import { join } from 'path'
import { Transaction } from 'sequelize'
import { AllowNull, BelongsTo, Column, CreatedAt, DataType, ForeignKey, Table, UpdatedAt } from 'sequelize-typescript'
import { SequelizeModel, getSort } from '../shared/index.js'
import { SequelizeModel, doesExist, getSort } from '../shared/index.js'
import { getResolutionLabel } from './formatter/video-api-format.js'
import { VideoModel } from './video.js'
import { MVideoSource } from '@server/types/models/video/video-source.js'
@Table({
tableName: 'videoSource',
@ -103,6 +103,18 @@ export class VideoSourceModel extends SequelizeModel<VideoSourceModel> {
})
}
// ---------------------------------------------------------------------------
static async doesOwnedFileExist (filename: string, storage: FileStorageType) {
const query = 'SELECT 1 FROM "videoSource" ' +
'INNER JOIN "video" ON "video"."id" = "videoSource"."videoId" AND "video"."remote" IS FALSE ' +
`WHERE "keptOriginalFilename" = $filename AND "storage" = $storage LIMIT 1`
return doesExist({ sequelize: this.sequelize, query, bind: { filename, storage } })
}
// ---------------------------------------------------------------------------
getFileDownloadUrl () {
if (!this.keptOriginalFilename) return null

View File

@ -229,13 +229,13 @@ export class VideoStreamingPlaylistModel extends SequelizeModel<VideoStreamingPl
return Object.assign(playlist, { Video: video })
}
static doesOwnedHLSPlaylistExist (videoUUID: string) {
static doesOwnedVideoUUIDExist (videoUUID: string, storage: FileStorageType) {
const query = `SELECT 1 FROM "videoStreamingPlaylist" ` +
`INNER JOIN "video" ON "video"."id" = "videoStreamingPlaylist"."videoId" ` +
`AND "video"."remote" IS FALSE AND "video"."uuid" = $videoUUID ` +
`AND "storage" = ${FileStorage.FILE_SYSTEM} LIMIT 1`
`AND "storage" = $storage LIMIT 1`
return doesExist({ sequelize: this.sequelize, query, bind: { videoUUID } })
return doesExist({ sequelize: this.sequelize, query, bind: { videoUUID, storage } })
}
assignP2PMediaLoaderInfoHashes (video: MVideo, files: unknown[]) {

View File

@ -1,13 +1,16 @@
import { uniqify } from '@peertube/peertube-core-utils'
import { FileStorage, ThumbnailType, ThumbnailType_Type } from '@peertube/peertube-models'
import { DIRECTORIES, USER_EXPORT_FILE_PREFIX } from '@server/initializers/constants.js'
import { listKeysOfPrefix, removeObjectByFullKey } from '@server/lib/object-storage/object-storage-helpers.js'
import { UserExportModel } from '@server/models/user/user-export.js'
import { VideoFileModel } from '@server/models/video/video-file.js'
import { VideoSourceModel } from '@server/models/video/video-source.js'
import { VideoStreamingPlaylistModel } from '@server/models/video/video-streaming-playlist.js'
import Bluebird from 'bluebird'
import { remove } from 'fs-extra/esm'
import { readdir, stat } from 'fs/promises'
import { basename, join } from 'path'
import { basename, dirname, join } from 'path'
import prompt from 'prompt'
import { uniqify } from '@peertube/peertube-core-utils'
import { ThumbnailType, ThumbnailType_Type } from '@peertube/peertube-models'
import { DIRECTORIES } from '@server/initializers/constants.js'
import { VideoFileModel } from '@server/models/video/video-file.js'
import { VideoStreamingPlaylistModel } from '@server/models/video/video-streaming-playlist.js'
import { getUUIDFromFilename } from '../core/helpers/utils.js'
import { CONFIG } from '../core/initializers/config.js'
import { initDatabaseModels } from '../core/initializers/database.js'
@ -24,6 +27,121 @@ run()
})
async function run () {
await initDatabaseModels(true)
await new FSPruner().prune()
console.log('\n')
await new ObjectStoragePruner().prune()
}
// ---------------------------------------------------------------------------
// Object storage
// ---------------------------------------------------------------------------
class ObjectStoragePruner {
private readonly keysToDelete: { bucket: string, key: string }[] = []
async prune () {
if (!CONFIG.OBJECT_STORAGE.ENABLED) return
console.log('Pruning object storage.')
await this.findFilesToDelete(CONFIG.OBJECT_STORAGE.WEB_VIDEOS, this.doesWebVideoFileExistFactory())
await this.findFilesToDelete(CONFIG.OBJECT_STORAGE.STREAMING_PLAYLISTS, this.doesStreamingPlaylistFileExistFactory())
await this.findFilesToDelete(CONFIG.OBJECT_STORAGE.ORIGINAL_VIDEO_FILES, this.doesOriginalFileExistFactory())
await this.findFilesToDelete(CONFIG.OBJECT_STORAGE.USER_EXPORTS, this.doesUserExportFileExistFactory())
if (this.keysToDelete.length === 0) {
console.log('No unknown object storage files to delete.')
return
}
const formattedKeysToDelete = this.keysToDelete.map(({ bucket, key }) => ` In bucket ${bucket}: ${key}`).join('\n')
console.log(`${this.keysToDelete.length} unknown files from object storage can be deleted:\n${formattedKeysToDelete}\n`)
const res = await askConfirmation()
if (res !== true) {
console.log('Exiting without deleting object storage files.')
return
}
console.log('Deleting object storage files...\n')
for (const { bucket, key } of this.keysToDelete) {
await removeObjectByFullKey(key, { BUCKET_NAME: bucket })
}
console.log(`${this.keysToDelete.length} object storage files deleted.`)
}
private async findFilesToDelete (
config: { BUCKET_NAME: string, PREFIX?: string },
existFun: (file: string) => Promise<boolean> | boolean
) {
try {
const keys = await listKeysOfPrefix('', config)
await Bluebird.map(keys, async key => {
if (await existFun(key) !== true) {
this.keysToDelete.push({ bucket: config.BUCKET_NAME, key })
}
}, { concurrency: 20 })
} catch (err) {
const prefixMessage = config.PREFIX
? ` and prefix ${config.PREFIX}`
: ''
console.error('Cannot find files to delete in bucket ' + config.BUCKET_NAME + prefixMessage)
}
}
private doesWebVideoFileExistFactory () {
return (key: string) => {
const filename = this.sanitizeKey(key, CONFIG.OBJECT_STORAGE.WEB_VIDEOS)
return VideoFileModel.doesOwnedFileExist(filename, FileStorage.OBJECT_STORAGE)
}
}
private doesStreamingPlaylistFileExistFactory () {
return (key: string) => {
const uuid = basename(dirname(this.sanitizeKey(key, CONFIG.OBJECT_STORAGE.STREAMING_PLAYLISTS)))
return VideoStreamingPlaylistModel.doesOwnedVideoUUIDExist(uuid, FileStorage.OBJECT_STORAGE)
}
}
private doesOriginalFileExistFactory () {
return (key: string) => {
const filename = this.sanitizeKey(key, CONFIG.OBJECT_STORAGE.ORIGINAL_VIDEO_FILES)
return VideoSourceModel.doesOwnedFileExist(filename, FileStorage.OBJECT_STORAGE)
}
}
private doesUserExportFileExistFactory () {
return (key: string) => {
const filename = this.sanitizeKey(key, CONFIG.OBJECT_STORAGE.USER_EXPORTS)
return UserExportModel.doesOwnedFileExist(filename, FileStorage.OBJECT_STORAGE)
}
}
private sanitizeKey (key: string, config: { PREFIX: string }) {
return key.replace(new RegExp(`^${config.PREFIX}`), '')
}
}
// ---------------------------------------------------------------------------
// FS
// ---------------------------------------------------------------------------
class FSPruner {
private pathsToDelete: string[] = []
async prune () {
const dirs = Object.values(CONFIG.STORAGE)
if (uniqify(dirs).length !== dirs.length) {
@ -31,92 +149,96 @@ async function run () {
process.exit(0)
}
await initDatabaseModels(true)
console.log('Pruning filesystem storage.')
let toDelete: string[] = []
console.log('Detecting files to remove, it can take a while...')
console.log('Detecting files to remove, it could take a while...')
await this.findFilesToDelete(DIRECTORIES.WEB_VIDEOS.PUBLIC, this.doesWebVideoFileExistFactory())
await this.findFilesToDelete(DIRECTORIES.WEB_VIDEOS.PRIVATE, this.doesWebVideoFileExistFactory())
toDelete = toDelete.concat(
await pruneDirectory(DIRECTORIES.WEB_VIDEOS.PUBLIC, doesWebVideoFileExist()),
await pruneDirectory(DIRECTORIES.WEB_VIDEOS.PRIVATE, doesWebVideoFileExist()),
await this.findFilesToDelete(DIRECTORIES.HLS_STREAMING_PLAYLIST.PRIVATE, this.doesHLSPlaylistExistFactory())
await this.findFilesToDelete(DIRECTORIES.HLS_STREAMING_PLAYLIST.PUBLIC, this.doesHLSPlaylistExistFactory())
await pruneDirectory(DIRECTORIES.HLS_STREAMING_PLAYLIST.PRIVATE, doesHLSPlaylistExist()),
await pruneDirectory(DIRECTORIES.HLS_STREAMING_PLAYLIST.PUBLIC, doesHLSPlaylistExist()),
await this.findFilesToDelete(DIRECTORIES.ORIGINAL_VIDEOS, this.doesOriginalVideoExistFactory())
await pruneDirectory(CONFIG.STORAGE.TORRENTS_DIR, doesTorrentFileExist()),
await this.findFilesToDelete(CONFIG.STORAGE.TORRENTS_DIR, this.doesTorrentFileExistFactory())
await pruneDirectory(CONFIG.STORAGE.REDUNDANCY_DIR, doesRedundancyExist),
await this.findFilesToDelete(CONFIG.STORAGE.REDUNDANCY_DIR, this.doesRedundancyExistFactory())
await pruneDirectory(CONFIG.STORAGE.PREVIEWS_DIR, doesThumbnailExist(true, ThumbnailType.PREVIEW)),
await pruneDirectory(CONFIG.STORAGE.THUMBNAILS_DIR, doesThumbnailExist(false, ThumbnailType.MINIATURE)),
await this.findFilesToDelete(CONFIG.STORAGE.PREVIEWS_DIR, this.doesThumbnailExistFactory(true, ThumbnailType.PREVIEW))
await this.findFilesToDelete(CONFIG.STORAGE.THUMBNAILS_DIR, this.doesThumbnailExistFactory(false, ThumbnailType.MINIATURE))
await pruneDirectory(CONFIG.STORAGE.ACTOR_IMAGES_DIR, doesActorImageExist)
)
await this.findFilesToDelete(CONFIG.STORAGE.ACTOR_IMAGES_DIR, this.doesActorImageExistFactory())
await this.findFilesToDelete(CONFIG.STORAGE.TMP_PERSISTENT_DIR, this.doesUserExportExistFactory())
const tmpFiles = await readdir(CONFIG.STORAGE.TMP_DIR)
toDelete = toDelete.concat(tmpFiles.map(t => join(CONFIG.STORAGE.TMP_DIR, t)))
this.pathsToDelete = [ ...this.pathsToDelete, ...tmpFiles.map(t => join(CONFIG.STORAGE.TMP_DIR, t)) ]
if (toDelete.length === 0) {
console.log('No files to delete.')
if (this.pathsToDelete.length === 0) {
console.log('No unknown filesystem files to delete.')
return
}
console.log('Will delete %d files:\n\n%s\n\n', toDelete.length, toDelete.join('\n'))
const formattedKeysToDelete = this.pathsToDelete.map(p => ` ${p}`).join('\n')
console.log(`${this.pathsToDelete.length} unknown files from filesystem can be deleted:\n${formattedKeysToDelete}\n`)
const res = await askConfirmation()
if (res === true) {
console.log('Processing delete...\n')
if (res !== true) {
console.log('Exiting without deleting filesystem files.')
return
}
for (const path of toDelete) {
console.log('Deleting filesystem files...\n')
for (const path of this.pathsToDelete) {
await remove(path)
}
console.log('Done!')
} else {
console.log('Exiting without deleting files.')
console.log(`${this.pathsToDelete.length} filesystem files deleted.`)
}
}
type ExistFun = (file: string) => Promise<boolean> | boolean
async function pruneDirectory (directory: string, existFun: ExistFun) {
private async findFilesToDelete (directory: string, existFun: (file: string) => Promise<boolean> | boolean) {
const files = await readdir(directory)
const toDelete: string[] = []
await Bluebird.map(files, async file => {
const filePath = join(directory, file)
if (await existFun(filePath) !== true) {
toDelete.push(filePath)
this.pathsToDelete.push(filePath)
}
}, { concurrency: 20 })
}
return toDelete
}
function doesWebVideoFileExist () {
private doesWebVideoFileExistFactory () {
return (filePath: string) => {
// Don't delete private directory
if (filePath === DIRECTORIES.WEB_VIDEOS.PRIVATE) return true
return VideoFileModel.doesOwnedWebVideoFileExist(basename(filePath))
return VideoFileModel.doesOwnedFileExist(basename(filePath), FileStorage.FILE_SYSTEM)
}
}
}
function doesHLSPlaylistExist () {
private doesHLSPlaylistExistFactory () {
return (hlsPath: string) => {
// Don't delete private directory
if (hlsPath === DIRECTORIES.HLS_STREAMING_PLAYLIST.PRIVATE) return true
return VideoStreamingPlaylistModel.doesOwnedHLSPlaylistExist(basename(hlsPath))
return VideoStreamingPlaylistModel.doesOwnedVideoUUIDExist(basename(hlsPath), FileStorage.FILE_SYSTEM)
}
}
}
function doesTorrentFileExist () {
private doesOriginalVideoExistFactory () {
return (filePath: string) => {
return VideoSourceModel.doesOwnedFileExist(basename(filePath), FileStorage.FILE_SYSTEM)
}
}
private doesTorrentFileExistFactory () {
return (filePath: string) => VideoFileModel.doesOwnedTorrentFileExist(basename(filePath))
}
}
function doesThumbnailExist (keepOnlyOwned: boolean, type: ThumbnailType_Type) {
private doesThumbnailExistFactory (keepOnlyOwned: boolean, type: ThumbnailType_Type) {
return async (filePath: string) => {
const thumbnail = await ThumbnailModel.loadByFilename(basename(filePath), type)
if (!thumbnail) return false
@ -128,15 +250,18 @@ function doesThumbnailExist (keepOnlyOwned: boolean, type: ThumbnailType_Type) {
return true
}
}
}
async function doesActorImageExist (filePath: string) {
private doesActorImageExistFactory () {
return async (filePath: string) => {
const image = await ActorImageModel.loadByName(basename(filePath))
return !!image
}
}
}
async function doesRedundancyExist (filePath: string) {
private doesRedundancyExistFactory () {
return async (filePath: string) => {
const isPlaylist = (await stat(filePath)).isDirectory()
if (isPlaylist) {
@ -159,6 +284,19 @@ async function doesRedundancyExist (filePath: string) {
const redundancy = await VideoRedundancyModel.loadLocalByFileId(file.id)
return !!redundancy
}
}
private doesUserExportExistFactory () {
return (filePath: string) => {
const filename = basename(filePath)
// Only detect non-existing user export
if (!filename.startsWith(USER_EXPORT_FILE_PREFIX)) return true
return UserExportModel.doesOwnedFileExist(filename, FileStorage.FILE_SYSTEM)
}
}
}
async function askConfirmation () {
@ -169,10 +307,12 @@ async function askConfirmation () {
properties: {
confirm: {
type: 'string',
description: 'These following unused files can be deleted, but please check your backups first (bugs happen).' +
description: 'These unknown files can be deleted, but please check your backups first (bugs happen).' +
' Notice PeerTube must have been stopped when your ran this script.' +
' Can we delete these files?',
' Can we delete these files? (y/n)',
default: 'n',
validator: /y[es]*|n[o]?/,
warning: 'Must respond yes or no',
required: true
}
}