cleanup-docker-registry-images.js

#!/usr/bin/env node

/**
 * Crude script to find images that are lingering and/or missing in the docker registry 2.0
 *
 * Finds all unreferences images in all repos, given the path to the registry-data 
 * directory (the one that contains 'storage'). Also lists all images that are referenced,
 * but are missing on disk.
 *
 * License: GPL 2.0
 * Author: Mattias Nilsson <[email protected]>
 *
 * Changelog: 
 *   - 2016-04-15: Initial revision
 *
 */

var fs = require('fs');
var path = require('path');

var regDataDir = process.argv[2];
if(!regDataDir) {
    console.error('Specify registry-data directory as argument');
    process.exit(1);
}

var storageDir = path.join(regDataDir, 'storage');
var reposDir = path.join(storageDir, 'repositories');
var imagesDir = path.join(storageDir, 'images');

var numRepos = 0;
var _imageIndex = {};

function formatImageList(imageList) {
    return imageList.map(function(img_id) {
        return img_id + (_imageIndex[img_id] ? ' ( ' + _imageIndex[img_id].join(',') + ' )' : '');
    }).join('\n');
}

fs.readdir(reposDir, function(err, files) {
    if(err) {
        console.error('failed to scan repositories directory %s: %s', reposDir, err);
        process.exit(1);
    }

    function imageExists(image_id) {
        var image_dir = path.join(imagesDir, image_id);
        var json_file = path.join(image_dir, 'json');
        try {
            var stat = fs.statSync(image_dir);
            if(!stat || !stat.isDirectory()) {
                return false;
            }
            stat = fs.statSync(json_file);
            return stat && stat.isFile();
        } catch(e) {
            return false;
        }
    }
    
    Promise.all(
        files.filter(function(f) {
            var stat = fs.statSync(path.join(reposDir, f));
            return stat && stat.isDirectory();
        }).map(function(subName) {
            var subDir = path.join(reposDir, subName);
            return new Promise(function(resolve, reject) {
                fs.readdir(subDir, function(err, repoList) {
                    if(err) {
                        return reject(err);
                    }
                    numRepos++;
                    resolve({path: subDir,
                             name: subName,
                             repos: repoList.map(function(repoName) {
                                 var imgsFile = path.join(subDir, repoName, '_index_images');
                                 var _index_images = JSON.parse(fs.readFileSync(imgsFile).toString());
                                 return {name: repoName,
                                         path: path.join(subDir, repoName),
                                         index_images: _index_images.map(function(img) {
                                             if(!_imageIndex[img.id]) { _imageIndex[img.id] = [];}
                                             _imageIndex[img.id].push(repoName);
                                             return {id: img.id, exists: imageExists(img.id)};
                                         })
                                        };
                             })
                            });
                });

            }).then(function(x) {
                return x;
            });
        })
    ).then(function(result) {
        var _usedImages = {};
        var allUsedImages = result.map(function(sub) {
            return sub.repos.map(function(repo) {
                return repo.index_images.map(function(im) {
                    _usedImages[im.id] = true;
                    return im.id;
                });
            });
        }).reduce(function(a, b) {
            return a.concat(b);
        }, []).reduce(function(a, b) {
            return a.concat(b);
        }, []);

        fs.readdir(imagesDir, function(err, image_dir_list) {
            var unusedImages = image_dir_list.filter(function(img_dir) {
                return !_usedImages[img_dir];
            });
            var missingImages = Object.keys(_usedImages).filter(function(img_id) {
                return image_dir_list.indexOf(img_id) === -1;
            });
            
            console.log('Scanned %d repositories', numRepos);
            console.log('- total # of images:   %d', image_dir_list.length);
            console.log('- referenced images:   %d', Object.keys(_usedImages).length);
            console.log('- unreferenced images: %d', unusedImages.length);
            if(unusedImages.length > 0) {
                console.log(formatImageList(unusedImages));
            }

            console.log('- missing images:      %d', missingImages.length);
            if(missingImages.length > 0) {
                console.log(formatImageList(missingImages));
            }

            process.exit(0);
        });

    }).catch(function(err) {
        console.error('Error occurred: %s', err);
        process.exit(1);
    });
});