cloudron-box/src/backuptask.js

'use strict';

exports = module.exports = {
    fullBackup,

    restore,

    backupApp,
    downloadApp,

    backupMail,
    downloadMail,

    upload,

    _restoreFsMetadata: restoreFsMetadata,
    _saveFsMetadata: saveFsMetadata,
};

const apps = require('./apps.js'),
    assert = require('assert'),
    async = require('async'),
    backups = require('./backups.js'),
    BoxError = require('./boxerror.js'),
    constants = require('./constants.js'),
    crypto = require('crypto'),
    DataLayout = require('./datalayout.js'),
    database = require('./database.js'),
    debug = require('debug')('box:backuptask'),
    fs = require('fs'),
    once = require('once'),
    path = require('path'),
    paths = require('./paths.js'),
    progressStream = require('progress-stream'),
    safe = require('safetydance'),
    services = require('./services.js'),
    settings = require('./settings.js'),
    shell = require('./shell.js'),
    storage = require('./storage.js'),
    syncer = require('./syncer.js'),
    tar = require('tar-fs'),
    TransformStream = require('stream').Transform,
    zlib = require('zlib'),
    util = require('util');

const BACKUP_UPLOAD_CMD = path.join(__dirname, 'scripts/backupupload.js');
const getBackupConfig = util.callbackify(settings.getBackupConfig);

function canBackupApp(app) {
    // only backup apps that are installed or specific pending states

    // stopped apps cannot be backed up because addons might be down (redis)
    if (app.runState === apps.RSTATE_STOPPED) return false;

    // we used to check the health here but that doesn't work for stopped apps. it's better to just fail
    // and inform the user if the backup fails and the app addons have not been setup yet.
    return  app.installationState === apps.ISTATE_INSTALLED ||
            app.installationState === apps.ISTATE_PENDING_CONFIGURE ||
            app.installationState === apps.ISTATE_PENDING_BACKUP ||  // called from apptask
            app.installationState === apps.ISTATE_PENDING_UPDATE; // called from apptask
}

function encryptFilePath(filePath, encryption) {
    assert.strictEqual(typeof filePath, 'string');
    assert.strictEqual(typeof encryption, 'object');

    var encryptedParts = filePath.split('/').map(function (part) {
        let hmac = crypto.createHmac('sha256', Buffer.from(encryption.filenameHmacKey, 'hex'));
        const iv = hmac.update(part).digest().slice(0, 16); // iv has to be deterministic, for our sync (copy) logic to work
        const cipher = crypto.createCipheriv('aes-256-cbc', Buffer.from(encryption.filenameKey, 'hex'), iv);
        let crypt = cipher.update(part);
        crypt = Buffer.concat([ iv, crypt, cipher.final() ]);

        return crypt.toString('base64')     // ensures path is valid
            .replace(/\//g, '-')            // replace '/' of base64 since it conflicts with path separator
            .replace(/=/g,'');              // strip trailing = padding. this is only needed if we concat base64 strings, which we don't
    });

    return encryptedParts.join('/');
}

function decryptFilePath(filePath, encryption) {
    assert.strictEqual(typeof filePath, 'string');
    assert.strictEqual(typeof encryption, 'object');

    let decryptedParts = [];
    for (let part of filePath.split('/')) {
        part = part + Array(part.length % 4).join('='); // add back = padding
        part = part.replace(/-/g, '/');                 // replace with '/'

        try {
            const buffer = Buffer.from(part, 'base64');
            const iv = buffer.slice(0, 16);
            let decrypt = crypto.createDecipheriv('aes-256-cbc', Buffer.from(encryption.filenameKey, 'hex'), iv);
            const plainText = decrypt.update(buffer.slice(16));
            const plainTextString = Buffer.concat([ plainText, decrypt.final() ]).toString('utf8');
            const hmac = crypto.createHmac('sha256', Buffer.from(encryption.filenameHmacKey, 'hex'));
            if (!hmac.update(plainTextString).digest().slice(0, 16).equals(iv)) return { error: new BoxError(BoxError.CRYPTO_ERROR, `mac error decrypting part ${part} of path ${filePath}`) };

            decryptedParts.push(plainTextString);
        } catch (error) {
            debug(`Error decrypting part ${part} of path ${filePath}:`, error);
            return { error: new BoxError(BoxError.CRYPTO_ERROR, `Error decrypting part ${part} of path ${filePath}: ${error.message}`) };
        }
    }

    return { result: decryptedParts.join('/') };
}

class EncryptStream extends TransformStream {
    constructor(encryption) {
        super();
        this._headerPushed = false;
        this._iv = crypto.randomBytes(16);
        this._cipher = crypto.createCipheriv('aes-256-cbc', Buffer.from(encryption.dataKey, 'hex'), this._iv);
        this._hmac = crypto.createHmac('sha256', Buffer.from(encryption.dataHmacKey, 'hex'));
    }

    pushHeaderIfNeeded() {
        if (!this._headerPushed) {
            const magic = Buffer.from('CBV2');
            this.push(magic);
            this._hmac.update(magic);
            this.push(this._iv);
            this._hmac.update(this._iv);
            this._headerPushed = true;
        }
    }

    _transform(chunk, ignoredEncoding, callback) {
        this.pushHeaderIfNeeded();

        try {
            const crypt = this._cipher.update(chunk);
            this._hmac.update(crypt);
            callback(null, crypt);
        } catch (error) {
            callback(error);
        }
    }

    _flush(callback) {
        try {
            this.pushHeaderIfNeeded(); // for 0-length files
            const crypt = this._cipher.final();
            this.push(crypt);
            this._hmac.update(crypt);
            callback(null, this._hmac.digest()); // +32 bytes
        } catch (error) {
            callback(error);
        }
    }
}

class DecryptStream extends TransformStream {
    constructor(encryption) {
        super();
        this._key = Buffer.from(encryption.dataKey, 'hex');
        this._header = Buffer.alloc(0);
        this._decipher = null;
        this._hmac = crypto.createHmac('sha256', Buffer.from(encryption.dataHmacKey, 'hex'));
        this._buffer = Buffer.alloc(0);
    }

    _transform(chunk, ignoredEncoding, callback) {
        const needed = 20 - this._header.length; // 4 for magic, 16 for iv

        if (this._header.length !== 20) { // not gotten header yet
            this._header = Buffer.concat([this._header, chunk.slice(0, needed)]);
            if (this._header.length !== 20) return callback();

            if (!this._header.slice(0, 4).equals(new Buffer.from('CBV2'))) return callback(new BoxError(BoxError.CRYPTO_ERROR, 'Invalid magic in header'));

            const iv = this._header.slice(4);
            this._decipher = crypto.createDecipheriv('aes-256-cbc', this._key, iv);
            this._hmac.update(this._header);
        }

        this._buffer = Buffer.concat([ this._buffer, chunk.slice(needed) ]);
        if (this._buffer.length < 32) return callback(); // hmac trailer length is 32

        try {
            const cipherText = this._buffer.slice(0, -32);
            this._hmac.update(cipherText);
            const plainText = this._decipher.update(cipherText);
            this._buffer = this._buffer.slice(-32);
            callback(null, plainText);
        } catch (error) {
            callback(error);
        }
    }

    _flush (callback) {
        if (this._buffer.length !== 32) return callback(new BoxError(BoxError.CRYPTO_ERROR, 'Invalid password or tampered file (not enough data)'));

        try {
            if (!this._hmac.digest().equals(this._buffer)) return callback(new BoxError(BoxError.CRYPTO_ERROR, 'Invalid password or tampered file (mac mismatch)'));

            const plainText = this._decipher.final();
            callback(null, plainText);
        } catch (error) {
            callback(error);
        }
    }
}

function createReadStream(sourceFile, encryption) {
    assert.strictEqual(typeof sourceFile, 'string');
    assert.strictEqual(typeof encryption, 'object');

    var stream = fs.createReadStream(sourceFile);
    var ps = progressStream({ time: 10000 }); // display a progress every 10 seconds

    stream.on('error', function (error) {
        debug(`createReadStream: read stream error at ${sourceFile}`, error);
        ps.emit('error', new BoxError(BoxError.FS_ERROR, `Error reading ${sourceFile}: ${error.message} ${error.code}`));
    });

    stream.on('open', () => ps.emit('open'));

    if (encryption) {
        let encryptStream = new EncryptStream(encryption);

        encryptStream.on('error', function (error) {
            debug(`createReadStream: encrypt stream error ${sourceFile}`, error);
            ps.emit('error', new BoxError(BoxError.CRYPTO_ERROR, `Encryption error at ${sourceFile}: ${error.message}`));
        });

        return stream.pipe(encryptStream).pipe(ps);
    } else {
        return stream.pipe(ps);
    }
}

function createWriteStream(destFile, encryption) {
    assert.strictEqual(typeof destFile, 'string');
    assert.strictEqual(typeof encryption, 'object');

    var stream = fs.createWriteStream(destFile);
    var ps = progressStream({ time: 10000 }); // display a progress every 10 seconds

    stream.on('error', function (error) {
        debug(`createWriteStream: write stream error ${destFile}`, error);
        ps.emit('error', new BoxError(BoxError.FS_ERROR, `Write error ${destFile}: ${error.message}`));
    });

    stream.on('finish', function () {
        debug('createWriteStream: done.');
        // we use a separate event because ps is a through2 stream which emits 'finish' event indicating end of inStream and not write
        ps.emit('done');
    });

    if (encryption) {
        let decrypt = new DecryptStream(encryption);
        decrypt.on('error', function (error) {
            debug(`createWriteStream: decrypt stream error ${destFile}`, error);
            ps.emit('error', new BoxError(BoxError.CRYPTO_ERROR, `Decryption error at ${destFile}: ${error.message}`));
        });

        ps.pipe(decrypt).pipe(stream);
    } else {
        ps.pipe(stream);
    }

    return ps;
}

function tarPack(dataLayout, encryption, callback) {
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof encryption, 'object');
    assert.strictEqual(typeof callback, 'function');

    var pack = tar.pack('/', {
        dereference: false, // pack the symlink and not what it points to
        entries: dataLayout.localPaths(),
        ignoreStatError: (path, err) => {
            debug(`tarPack: error stat'ing ${path} - ${err.code}`);
            return err.code === 'ENOENT'; // ignore if file or dir got removed (probably some temporary file)
        },
        map: function(header) {
            header.name = dataLayout.toRemotePath(header.name);
            // the tar pax format allows us to encode filenames > 100 and size > 8GB (see #640)
            // https://www.systutorials.com/docs/linux/man/5-star/
            if (header.size > 8589934590 || header.name > 99) header.pax = { size: header.size };
            return header;
        },
        strict: false // do not error for unknown types (skip fifo, char/block devices)
    });

    var gzip = zlib.createGzip({});
    var ps = progressStream({ time: 10000 }); // emit 'progress' every 10 seconds

    pack.on('error', function (error) {
        debug('tarPack: tar stream error.', error);
        ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    gzip.on('error', function (error) {
        debug('tarPack: gzip stream error.', error);
        ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    if (encryption) {
        const encryptStream = new EncryptStream(encryption);
        encryptStream.on('error', function (error) {
            debug('tarPack: encrypt stream error.', error);
            ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
        });

        pack.pipe(gzip).pipe(encryptStream).pipe(ps);
    } else {
        pack.pipe(gzip).pipe(ps);
    }

    return callback(null, ps);
}

function sync(backupConfig, backupId, dataLayout, progressCallback, callback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof backupId, 'string');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof progressCallback, 'function');
    assert.strictEqual(typeof callback, 'function');

    // the number here has to take into account the s3.upload partSize (which is 10MB). So 20=200MB
    const concurrency = backupConfig.syncConcurrency || (backupConfig.provider === 's3' ? 20 : 10);

    syncer.sync(dataLayout, function processTask(task, iteratorCallback) {
        debug('sync: processing task: %j', task);
        // the empty task.path is special to signify the directory
        const destPath = task.path && backupConfig.encryption ? encryptFilePath(task.path, backupConfig.encryption) : task.path;
        const backupFilePath = path.join(storage.getBackupFilePath(backupConfig, backupId, backupConfig.format), destPath);

        if (task.operation === 'removedir') {
            debug(`Removing directory ${backupFilePath}`);
            return storage.api(backupConfig.provider).removeDir(backupConfig, backupFilePath)
                .on('progress', (message) => progressCallback({ message }))
                .on('done', iteratorCallback);
        } else if (task.operation === 'remove') {
            debug(`Removing ${backupFilePath}`);
            return storage.api(backupConfig.provider).remove(backupConfig, backupFilePath, iteratorCallback);
        }

        var retryCount = 0;
        async.retry({ times: 5, interval: 20000 }, function (retryCallback) {
            retryCallback = once(retryCallback); // protect again upload() erroring much later after read stream error

            ++retryCount;
            if (task.operation === 'add') {
                progressCallback({ message: `Adding ${task.path}` + (retryCount > 1 ?  ` (Try ${retryCount})` : '') });
                debug(`Adding ${task.path} position ${task.position} try ${retryCount}`);
                var stream = createReadStream(dataLayout.toLocalPath('./' + task.path), backupConfig.encryption);
                stream.on('error', (error) => retryCallback(error.message.includes('ENOENT') ? null : error)); // ignore error if file disappears
                stream.on('progress', function (progress) {
                    const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
                    if (!transferred && !speed) return progressCallback({ message: `Uploading ${task.path}` }); // 0M@0MBps looks wrong
                    progressCallback({ message: `Uploading ${task.path}: ${transferred}M@${speed}MBps` }); // 0M@0MBps looks wrong
                });
                // only create the destination path when we have confirmation that the source is available. otherwise, we end up with
                // files owned as 'root' and the cp later will fail
                stream.on('open', function () {
                    storage.api(backupConfig.provider).upload(backupConfig, backupFilePath, stream, function (error) {
                        debug(error ? `Error uploading ${task.path} try ${retryCount}: ${error.message}` : `Uploaded ${task.path}`);
                        retryCallback(error);
                    });
                });
            }
        }, iteratorCallback);
    }, concurrency, function (error) {
        if (error) return callback(new BoxError(BoxError.EXTERNAL_ERROR, error.message));

        callback();
    });
}

// this is not part of 'snapshotting' because we need root access to traverse
async function saveFsMetadata(dataLayout, metadataFile) {
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof metadataFile, 'string');

    // contains paths prefixed with './'
    let metadata = {
        emptyDirs: [],
        execFiles: [],
        symlinks: []
    };

    // we assume small number of files. spawnSync will raise a ENOBUFS error after maxBuffer
    for (let lp of dataLayout.localPaths()) {
        const emptyDirs = safe.child_process.execSync(`find ${lp} -type d -empty`, { encoding: 'utf8', maxBuffer: 1024 * 1024 * 30 });
        if (emptyDirs === null) throw new BoxError(BoxError.FS_ERROR, `Error finding empty dirs: ${safe.error.message}`);
        if (emptyDirs.length) metadata.emptyDirs = metadata.emptyDirs.concat(emptyDirs.trim().split('\n').map((ed) => dataLayout.toRemotePath(ed)));

        const execFiles = safe.child_process.execSync(`find ${lp} -type f -executable`, { encoding: 'utf8', maxBuffer: 1024 * 1024 * 30 });
        if (execFiles === null) throw new BoxError(BoxError.FS_ERROR, `Error finding executables: ${safe.error.message}`);
        if (execFiles.length) metadata.execFiles = metadata.execFiles.concat(execFiles.trim().split('\n').map((ef) => dataLayout.toRemotePath(ef)));

        const symlinks = safe.child_process.execSync(`find ${lp} -type l`, { encoding: 'utf8', maxBuffer: 1024 * 1024 * 30 });
        if (symlinks === null) throw new BoxError(BoxError.FS_ERROR, `Error finding symlinks: ${safe.error.message}`);
        if (symlinks.length) metadata.symlinks = metadata.symlinks.concat(symlinks.trim().split('\n').map((sl) => {
            const target = safe.fs.readlinkSync(sl);
            return { path: dataLayout.toRemotePath(sl), target };
        }));
    }

    if (!safe.fs.writeFileSync(metadataFile, JSON.stringify(metadata, null, 4))) throw new BoxError(BoxError.FS_ERROR, `Error writing fs metadata: ${safe.error.message}`);
}

// this function is called via backupupload (since it needs root to traverse app's directory)
function upload(backupId, format, dataLayoutString, progressCallback, callback) {
    assert.strictEqual(typeof backupId, 'string');
    assert.strictEqual(typeof format, 'string');
    assert.strictEqual(typeof dataLayoutString, 'string');
    assert.strictEqual(typeof progressCallback, 'function');
    assert.strictEqual(typeof callback, 'function');

    debug(`upload: id ${backupId} format ${format} dataLayout ${dataLayoutString}`);

    const dataLayout = DataLayout.fromString(dataLayoutString);

    getBackupConfig(function (error, backupConfig) {
        if (error) return callback(error);

        storage.api(backupConfig.provider).checkPreconditions(backupConfig, dataLayout, function (error) {
            if (error) return callback(error);

            if (format === 'tgz') {
                async.retry({ times: 5, interval: 20000 }, function (retryCallback) {
                    retryCallback = once(retryCallback); // protect again upload() erroring much later after tar stream error

                    tarPack(dataLayout, backupConfig.encryption, function (error, tarStream) {
                        if (error) return retryCallback(error);

                        tarStream.on('progress', function (progress) {
                            const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
                            if (!transferred && !speed) return progressCallback({ message: 'Uploading backup' }); // 0M@0MBps looks wrong
                            progressCallback({ message: `Uploading backup ${transferred}M@${speed}MBps` });
                        });
                        tarStream.on('error', retryCallback); // already returns BoxError

                        storage.api(backupConfig.provider).upload(backupConfig, storage.getBackupFilePath(backupConfig, backupId, format), tarStream, retryCallback);
                    });
                }, callback);
            } else {
                async.series([
                    saveFsMetadata.bind(null, dataLayout, `${dataLayout.localRoot()}/fsmetadata.json`),
                    sync.bind(null, backupConfig, backupId, dataLayout, progressCallback)
                ], callback);
            }
        });
    });
}

function tarExtract(inStream, dataLayout, encryption, callback) {
    assert.strictEqual(typeof inStream, 'object');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof encryption, 'object');
    assert.strictEqual(typeof callback, 'function');

    var gunzip = zlib.createGunzip({});
    var ps = progressStream({ time: 10000 }); // display a progress every 10 seconds
    var extract = tar.extract('/', {
        map: function (header) {
            header.name = dataLayout.toLocalPath(header.name);
            return header;
        },
        dmode: 500 // ensure directory is writable
    });

    const emitError = once((error) => {
        inStream.destroy();
        ps.emit('error', error);
    });

    inStream.on('error', function (error) {
        debug('tarExtract: input stream error.', error);
        emitError(new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    gunzip.on('error', function (error) {
        debug('tarExtract: gunzip stream error.', error);
        emitError(new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    extract.on('error', function (error) {
        debug('tarExtract: extract stream error.', error);
        emitError(new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    extract.on('finish', function () {
        debug('tarExtract: done.');
        // we use a separate event because ps is a through2 stream which emits 'finish' event indicating end of inStream and not extract
        ps.emit('done');
    });

    if (encryption) {
        let decrypt = new DecryptStream(encryption);
        decrypt.on('error', function (error) {
            debug('tarExtract: decrypt stream error.', error);
            emitError(new BoxError(BoxError.EXTERNAL_ERROR, `Failed to decrypt: ${error.message}`));
        });
        inStream.pipe(ps).pipe(decrypt).pipe(gunzip).pipe(extract);
    } else {
        inStream.pipe(ps).pipe(gunzip).pipe(extract);
    }

    callback(null, ps);
}

async function restoreFsMetadata(dataLayout, metadataFile) {
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof metadataFile, 'string');

    debug(`Recreating empty directories in ${dataLayout.toString()}`);

    var metadataJson = safe.fs.readFileSync(metadataFile, 'utf8');
    if (metadataJson === null) throw new BoxError(BoxError.EXTERNAL_ERROR, 'Error loading fsmetadata.json:' + safe.error.message);
    var metadata = safe.JSON.parse(metadataJson);
    if (metadata === null) throw new BoxError(BoxError.EXTERNAL_ERROR, 'Error parsing fsmetadata.json:' + safe.error.message);

    for (const emptyDir of metadata.emptyDirs) {
        const [mkdirError] = await safe(fs.promises.mkdir(dataLayout.toLocalPath(emptyDir), { recursive: true }));
        if (mkdirError) throw new BoxError(BoxError.FS_ERROR, `unable to create path: ${mkdirError.message}`);
    }

    for (const execFile of metadata.execFiles) {
        const [chmodError] = await safe(fs.promises.chmod(dataLayout.toLocalPath(execFile), parseInt('0755', 8)));
        if (chmodError) throw new BoxError(BoxError.FS_ERROR, `unable to chmod: ${chmodError.message}`);
    }

    for (const symlink of (metadata.symlinks || [])) {
        if (!symlink.target) continue;
        // the path may not exist if we had a directory full of symlinks
        const [mkdirError] = await safe(fs.promises.mkdir(path.dirname(dataLayout.toLocalPath(symlink.path)), { recursive: true }));
        if (mkdirError) throw new BoxError(BoxError.FS_ERROR, `unable to symlink (mkdir): ${mkdirError.message}`);
        const [symlinkError] = await safe(fs.promises.symlink(symlink.target, dataLayout.toLocalPath(symlink.path), 'file'));
        if (symlinkError) throw new BoxError(BoxError.FS_ERROR, `unable to symlink: ${symlinkError.message}`);
    }
}

function downloadDir(backupConfig, backupFilePath, dataLayout, progressCallback, callback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof backupFilePath, 'string');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof progressCallback, 'function');
    assert.strictEqual(typeof callback, 'function');

    debug(`downloadDir: ${backupFilePath} to ${dataLayout.toString()}`);

    function downloadFile(entry, done) {
        let relativePath = path.relative(backupFilePath, entry.fullPath);
        if (backupConfig.encryption) {
            const { error, result } = decryptFilePath(relativePath, backupConfig.encryption);
            if (error) return done(new BoxError(BoxError.CRYPTO_ERROR, 'Unable to decrypt file'));
            relativePath = result;
        }
        const destFilePath = dataLayout.toLocalPath('./' + relativePath);

        fs.mkdir(path.dirname(destFilePath), { recursive: true }, function (error) {
            if (error) return done(new BoxError(BoxError.FS_ERROR, error.message));

            async.retry({ times: 5, interval: 20000 }, function (retryCallback) {
                storage.api(backupConfig.provider).download(backupConfig, entry.fullPath, function (error, sourceStream) {
                    if (error) {
                        progressCallback({ message: `Download ${entry.fullPath} to ${destFilePath} errored: ${error.message}` });
                        return retryCallback(error);
                    }

                    let destStream = createWriteStream(destFilePath, backupConfig.encryption);

                    // protect against multiple errors. must destroy the write stream so that a previous retry does not write
                    let closeAndRetry = once((error) => {
                        if (error) progressCallback({ message: `Download ${entry.fullPath} to ${destFilePath} errored: ${error.message}` });
                        else progressCallback({ message: `Download ${entry.fullPath} to ${destFilePath} finished` });
                        sourceStream.destroy();
                        destStream.destroy();
                        retryCallback(error);
                    });

                    destStream.on('progress', function (progress) {
                        const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
                        if (!transferred && !speed) return progressCallback({ message: `Downloading ${entry.fullPath}` }); // 0M@0MBps looks wrong
                        progressCallback({ message: `Downloading ${entry.fullPath}: ${transferred}M@${speed}MBps` });
                    });
                    destStream.on('error', closeAndRetry);

                    sourceStream.on('error', closeAndRetry);

                    progressCallback({ message: `Downloading ${entry.fullPath} to ${destFilePath}` });

                    sourceStream.pipe(destStream, { end: true }).on('done', closeAndRetry);
                });
            }, done);
        });
    }

    storage.api(backupConfig.provider).listDir(backupConfig, backupFilePath, 1000, function (entries, iteratorDone) {
        // https://www.digitalocean.com/community/questions/rate-limiting-on-spaces?answer=40441
        const concurrency = backupConfig.downloadConcurrency || (backupConfig.provider === 's3' ? 30 : 10);

        async.eachLimit(entries, concurrency, downloadFile, iteratorDone);
    }, callback);
}

function download(backupConfig, backupId, format, dataLayout, progressCallback, callback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof backupId, 'string');
    assert.strictEqual(typeof format, 'string');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof progressCallback, 'function');
    assert.strictEqual(typeof callback, 'function');

    debug(`download: Downloading ${backupId} of format ${format} to ${dataLayout.toString()}`);

    const backupFilePath = storage.getBackupFilePath(backupConfig, backupId, format);

    if (format === 'tgz') {
        async.retry({ times: 5, interval: 20000 }, function (retryCallback) {
            progressCallback({ message: `Downloading backup ${backupId}` });

            storage.api(backupConfig.provider).download(backupConfig, backupFilePath, function (error, sourceStream) {
                if (error) return retryCallback(error);

                tarExtract(sourceStream, dataLayout, backupConfig.encryption, function (error, ps) {
                    if (error) return retryCallback(error);

                    ps.on('progress', function (progress) {
                        const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
                        if (!transferred && !speed) return progressCallback({ message: 'Downloading backup' }); // 0M@0MBps looks wrong
                        progressCallback({ message: `Downloading ${transferred}M@${speed}MBps` });
                    });
                    ps.on('error', retryCallback);
                    ps.on('done', retryCallback);
                });
            });
        }, callback);
    } else {
        downloadDir(backupConfig, backupFilePath, dataLayout, progressCallback, async function (error) {
            if (error) return callback(error);

            [error] = await safe(restoreFsMetadata(dataLayout, `${dataLayout.localRoot()}/fsmetadata.json`));
            callback(error);
        });
    }
}

async function restore(backupConfig, backupId, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof backupId, 'string');
    assert.strictEqual(typeof progressCallback, 'function');

    const boxDataDir = safe.fs.realpathSync(paths.BOX_DATA_DIR);
    if (!boxDataDir) throw new BoxError(BoxError.FS_ERROR, `Error resolving boxdata: ${safe.error.message}`);
    const dataLayout = new DataLayout(boxDataDir, []);

    await util.promisify(download)(backupConfig, backupId, backupConfig.format, dataLayout, progressCallback);

    debug('restore: download completed, importing database');

    await database.importFromFile(`${dataLayout.localRoot()}/box.mysqldump`);

    debug('restore: database imported');

    await settings.initCache();
}

async function downloadApp(app, restoreConfig, progressCallback) {
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof restoreConfig, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const appDataDir = safe.fs.realpathSync(path.join(paths.APPS_DATA_DIR, app.id));
    if (!appDataDir) throw new BoxError(BoxError.FS_ERROR, safe.error.message);
    const dataLayout = new DataLayout(appDataDir, app.dataDir ? [{ localDir: app.dataDir, remoteDir: 'data' }] : []);

    const startTime = new Date();
    const backupConfig = restoreConfig.backupConfig || await settings.getBackupConfig();

    const downloadAsync = util.promisify(download);
    await downloadAsync(backupConfig, restoreConfig.backupId, restoreConfig.backupFormat, dataLayout, progressCallback);
    debug('downloadApp: time: %s', (new Date() - startTime)/1000);
}

function runBackupUpload(uploadConfig, progressCallback, callback) {
    assert.strictEqual(typeof uploadConfig, 'object');
    assert.strictEqual(typeof progressCallback, 'function');
    assert.strictEqual(typeof callback, 'function');

    const { backupId, backupConfig, dataLayout, progressTag } = uploadConfig;
    assert.strictEqual(typeof backupId, 'string');
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof progressTag, 'string');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');

    let result = ''; // the script communicates error result as a string

    // https://stackoverflow.com/questions/48387040/node-js-recommended-max-old-space-size
    const envCopy = Object.assign({}, process.env);
    if (backupConfig.memoryLimit && backupConfig.memoryLimit >= 2*1024*1024*1024) {
        const heapSize = Math.min((backupConfig.memoryLimit/1024/1024) - 256, 8192);
        debug(`runBackupUpload: adjusting heap size to ${heapSize}M`);
        envCopy.NODE_OPTIONS = `--max-old-space-size=${heapSize}`;
    }

    shell.sudo(`backup-${backupId}`, [ BACKUP_UPLOAD_CMD, backupId, backupConfig.format, dataLayout.toString() ], { env: envCopy, preserveEnv: true, ipc: true }, function (error) {
        if (error && (error.code === null /* signal */ || (error.code !== 0 && error.code !== 50))) { // backuptask crashed
            return callback(new BoxError(BoxError.INTERNAL_ERROR, 'Backuptask crashed'));
        } else if (error && error.code === 50) { // exited with error
            return callback(new BoxError(BoxError.EXTERNAL_ERROR, result));
        }

        callback();
    }).on('message', function (progress) { // this is { message } or { result }
        if ('message' in progress) return progressCallback({ message: `${progress.message} (${progressTag})` });
        debug(`runBackupUpload: result - ${JSON.stringify(progress)}`);
        result = progress.result;
    });
}

async function snapshotBox(progressCallback) {
    assert.strictEqual(typeof progressCallback, 'function');

    progressCallback({ message: 'Snapshotting box' });

    const startTime = new Date();

    await database.exportToFile(`${paths.BOX_DATA_DIR}/box.mysqldump`);
    debug(`snapshotBox: took ${(new Date() - startTime)/1000} seconds`);
}

async function uploadBoxSnapshot(backupConfig, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    await snapshotBox(progressCallback);

    const boxDataDir = safe.fs.realpathSync(paths.BOX_DATA_DIR);
    if (!boxDataDir) throw new BoxError(BoxError.FS_ERROR, `Error resolving boxdata: ${safe.error.message}`);

    const uploadConfig = {
        backupId: 'snapshot/box',
        backupConfig,
        dataLayout: new DataLayout(boxDataDir, []),
        progressTag: 'box'
    };

    progressCallback({ message: 'Uploading box snapshot' });

    const startTime = new Date();

    await util.promisify(runBackupUpload)(uploadConfig, progressCallback);

    debug(`uploadBoxSnapshot: took ${(new Date() - startTime)/1000} seconds`);

    await backups.setSnapshotInfo('box', { timestamp: new Date().toISOString(), format: backupConfig.format });
}

async function copy(backupConfig, sourceBackupId, destBackupId, options, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof sourceBackupId, 'string');
    assert.strictEqual(typeof destBackupId, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const format = backupConfig.format;

    return new Promise((resolve, reject) => {
        const startTime = new Date();

        const copyEvents = storage.api(backupConfig.provider).copy(backupConfig, storage.getBackupFilePath(backupConfig, sourceBackupId, format), storage.getBackupFilePath(backupConfig, destBackupId, format));
        copyEvents.on('progress', (message) => progressCallback({ message }));
        copyEvents.on('done', async function (copyBackupError) {
            const state = copyBackupError ? backups.BACKUP_STATE_ERROR : backups.BACKUP_STATE_NORMAL;

            const [error] = await safe(backups.update(destBackupId, { preserveSecs: options.preserveSecs || 0, state }));
            if (copyBackupError) return reject(copyBackupError);
            if (error) return reject(error);

            debug(`copy: copied successfully to id ${destBackupId}. Took ${(new Date() - startTime)/1000} seconds`);

            resolve();
        });
    });
}

async function rotateBoxBackup(backupConfig, tag, options, dependsOn, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert(Array.isArray(dependsOn));
    assert.strictEqual(typeof progressCallback, 'function');

    const backupId = `${tag}/box_v${constants.VERSION}`;
    const format = backupConfig.format;

    debug(`rotateBoxBackup: rotating to id ${backupId}`);

    const data = {
        encryptionVersion: backupConfig.encryption ? 2 : null,
        packageVersion: constants.VERSION,
        type: backups.BACKUP_TYPE_BOX,
        state: backups.BACKUP_STATE_CREATING,
        identifier: backups.BACKUP_IDENTIFIER_BOX,
        dependsOn,
        manifest: null,
        format
    };

    await backups.add(backupId, data);
    await copy(backupConfig, 'snapshot/box', backupId, options, progressCallback);
    return backupId;
}

async function backupBox(dependsOn, tag, options, progressCallback) {
    assert(Array.isArray(dependsOn));
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const backupConfig = await settings.getBackupConfig();

    await uploadBoxSnapshot(backupConfig, progressCallback);

    const backupId = await rotateBoxBackup(backupConfig, tag, options, dependsOn, progressCallback);
    return backupId;
}

async function rotateAppBackup(backupConfig, app, tag, options, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const snapshotInfo = backups.getSnapshotInfo(app.id);

    const manifest = snapshotInfo.restoreConfig ? snapshotInfo.restoreConfig.manifest : snapshotInfo.manifest; // compat
    const backupId = `${tag}/app_${app.fqdn}_v${manifest.version}`;
    const format = backupConfig.format;

    debug(`rotateAppBackup: rotating ${app.fqdn} to id ${backupId}`);

    const data = {
        encryptionVersion: backupConfig.encryption ? 2 : null,
        packageVersion: manifest.version,
        type: backups.BACKUP_TYPE_APP,
        state: backups.BACKUP_STATE_CREATING,
        identifier: app.id,
        dependsOn: [ ],
        manifest,
        format: format
    };

    await backups.add(backupId, data);
    await copy(backupConfig,  `snapshot/app_${app.id}`, backupId, options, progressCallback);
    return backupId;
}

async function backupApp(app, options, progressCallback) {
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    if (options.snapshotOnly) return await snapshotApp(app, progressCallback);

    const tag = (new Date()).toISOString().replace(/[T.]/g, '-').replace(/[:Z]/g,'');

    debug(`backupApp: backing up ${app.fqdn} with tag ${tag}`);

    await backupAppWithTag(app, tag, options, progressCallback);
}

async function snapshotApp(app, progressCallback) {
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const startTime = new Date();
    progressCallback({ message: `Snapshotting app ${app.fqdn}` });

    await apps.backupConfig(app);
    await services.backupAddons(app, app.manifest.addons);

    debug(`snapshotApp: ${app.fqdn} took ${(new Date() - startTime)/1000} seconds`);
}

async function uploadAppSnapshot(backupConfig, app, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    await snapshotApp(app, progressCallback);

    const backupId = util.format('snapshot/app_%s', app.id);
    const appDataDir = safe.fs.realpathSync(path.join(paths.APPS_DATA_DIR, app.id));
    if (!appDataDir) throw new BoxError(BoxError.FS_ERROR, `Error resolving appsdata: ${safe.error.message}`);

    const dataLayout = new DataLayout(appDataDir, app.dataDir ? [{ localDir: app.dataDir, remoteDir: 'data' }] : []);

    progressCallback({ message: `Uploading app snapshot ${app.fqdn}`});

    const uploadConfig = {
        backupId,
        backupConfig,
        dataLayout,
        progressTag: app.fqdn
    };

    const startTime = new Date();

    await util.promisify(runBackupUpload)(uploadConfig, progressCallback);

    debug(`uploadAppSnapshot: ${app.fqdn} upload with id ${backupId}. ${(new Date() - startTime)/1000} seconds`);

    await backups.setSnapshotInfo(app.id, { timestamp: new Date().toISOString(), manifest: app.manifest, format: backupConfig.format });
}

async function backupAppWithTag(app, tag, options, progressCallback) {
    assert.strictEqual(typeof app, 'object');
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    if (!canBackupApp(app)) { // if we cannot backup, reuse it's most recent backup
        const results = await backups.getByIdentifierAndStatePaged(app.id, backups.BACKUP_STATE_NORMAL, 1, 1);
        if (results.length === 0) return null; // no backup to re-use

        return results[0].id;
    }

    const backupConfig = await settings.getBackupConfig();

    await uploadAppSnapshot(backupConfig, app, progressCallback);
    const backupId = await rotateAppBackup(backupConfig, app, tag, options, progressCallback);
    return backupId;
}

async function uploadMailSnapshot(backupConfig, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const mailDataDir = safe.fs.realpathSync(paths.MAIL_DATA_DIR);
    if (!mailDataDir) throw new BoxError(BoxError.FS_ERROR, `Error resolving maildata: ${safe.error.message}`);

    const uploadConfig = {
        backupId: 'snapshot/mail',
        backupConfig,
        dataLayout: new DataLayout(mailDataDir, []),
        progressTag: 'mail'
    };

    progressCallback({ message: 'Uploading mail snapshot' });

    const startTime = new Date();

    await util.promisify(runBackupUpload)(uploadConfig, progressCallback);

    debug(`uploadMailSnapshot: took ${(new Date() - startTime)/1000} seconds`);

    await backups.setSnapshotInfo('mail', { timestamp: new Date().toISOString(), format: backupConfig.format });
}

async function rotateMailBackup(backupConfig, tag, options, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const backupId = `${tag}/mail_v${constants.VERSION}`;
    const format = backupConfig.format;

    debug(`rotateMailBackup: rotating to id ${backupId}`);

    const data = {
        encryptionVersion: backupConfig.encryption ? 2 : null,
        packageVersion: constants.VERSION,
        type: backups.BACKUP_TYPE_MAIL,
        state: backups.BACKUP_STATE_CREATING,
        identifier: backups.BACKUP_IDENTIFIER_MAIL,
        dependsOn: [],
        manifest: null,
        format: format
    };

    await backups.add(backupId, data);
    await copy(backupConfig, 'snapshot/mail', backupId, options, progressCallback);
    return backupId;
}

async function backupMailWithTag(tag, options, progressCallback) {
    assert.strictEqual(typeof tag, 'string');
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    debug(`backupMailWithTag: backing up mail with tag ${tag}`);

    const backupConfig = await settings.getBackupConfig();
    await uploadMailSnapshot(backupConfig, progressCallback);
    const backupId = await rotateMailBackup(backupConfig, tag, options, progressCallback);
    return backupId;
}

async function backupMail(options, progressCallback) {
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const tag = (new Date()).toISOString().replace(/[T.]/g, '-').replace(/[:Z]/g,'');
    debug(`backupMail: backing up mail with tag ${tag}`);
    return await backupMailWithTag(tag, options, progressCallback);
}

async function downloadMail(restoreConfig, progressCallback) {
    assert.strictEqual(typeof restoreConfig, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const mailDataDir = safe.fs.realpathSync(paths.MAIL_DATA_DIR);
    if (!mailDataDir) throw new BoxError(BoxError.FS_ERROR, `Error resolving maildata: ${safe.error.message}`);
    const dataLayout = new DataLayout(mailDataDir, []);

    const startTime = new Date();
    const backupConfig = restoreConfig.backupConfig || await settings.getBackupConfig();

    const downloadAsync = util.promisify(download);
    await downloadAsync(backupConfig, restoreConfig.backupId, restoreConfig.backupFormat, dataLayout, progressCallback);
    debug('downloadMail: time: %s', (new Date() - startTime)/1000);
}

// this function is called from external process. calling process is expected to have a lock
async function fullBackup(options, progressCallback) {
    assert.strictEqual(typeof options, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const tag = (new Date()).toISOString().replace(/[T.]/g, '-').replace(/[:Z]/g,'');

    const allApps = await apps.list();

    let percent = 1;
    let step = 100/(allApps.length+3);

    const appBackupIds = [];
    for (const app of allApps) {
        progressCallback({ percent: percent, message: `Backing up ${app.fqdn}` });
        percent += step;

        if (!app.enableBackup) {
            debug(`fullBackup: skipped backup ${app.fqdn}`);
            return; // nothing to backup
        }

        const startTime = new Date();
        const appBackupId = await backupAppWithTag(app, tag, options, (progress) => progressCallback({ percent: percent, message: progress.message }));
        debug(`fullBackup: app ${app.fqdn} backup finished. Took ${(new Date() - startTime)/1000} seconds`);
        if (appBackupId) appBackupIds.push(appBackupId); // backupId can be null if in BAD_STATE and never backed up
    }

    progressCallback({ percent: percent, message: 'Backing up mail' });
    percent += step;
    const mailBackupId = await backupMailWithTag(tag, options, (progress) => progressCallback({ percent: percent, message: progress.message }));

    progressCallback({ percent: percent, message: 'Backing up system data' });
    percent += step;

    const dependsOn = appBackupIds.concat(mailBackupId);
    const backupId = await backupBox(dependsOn, tag, options, (progress) => progressCallback({ percent: percent, message: progress.message }));
    return backupId;
}