Cleaning up orphaned chunks in GridFS

Program.cs

using MongoDB.Bson;
using MongoDB.Driver;
using Serilog;
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Threading;
using System.Threading.Tasks;

namespace Mongo
{
    internal class Program
    {
        private static async Task<int> Main(string[] args)
        {
            Log.Logger = new LoggerConfiguration()
             .MinimumLevel.Debug()
             .WriteTo.Console()
             .WriteTo.File("mongo.txt")
             .CreateLogger();

            var mongoSettings = MongoClientSettings.FromConnectionString("");
            var mongoClient = new MongoClient(mongoSettings);
            var mongoDatabase = mongoClient.GetDatabase("packagesDb");
            var chunksCollection = mongoDatabase.GetCollection<BsonDocument>("packages.chunks");
            var filesCollection = mongoDatabase.GetCollection<BsonDocument>("packages.files");

            using var cts = new CancellationTokenSource();
            Console.CancelKeyPress += (sender, e) =>
            {
                e.Cancel = true;
                cts.Cancel();
            };

            var sw = new Stopwatch();
            var numberOfChunksDeleted = 0UL;

            try
            {
                var chunkFilter = Builders<BsonDocument>.Filter.Empty;
                
                using var cursor = await chunksCollection.FindAsync(chunkFilter, new FindOptions<BsonDocument, BsonDocument>()
                {
                    BatchSize = 1000,
                    Skip = 0,
                    Limit = 100_000,
                    //Sort = Builders<BsonDocument>.Sort.Descending("_id"),
                    Projection = Builders<BsonDocument>.Projection.Exclude("data")
                }, cts.Token);

                sw.Start();

                var uniqueFiles = new HashSet<string>();
                while (await cursor.MoveNextAsync(cts.Token))
                {
                    var batch = cursor.Current;

                    foreach (var gridFSChunk in batch)
                    {
                        var objecdId = gridFSChunk.GetValue("_id").AsObjectId;
                        uniqueFiles.Add(gridFSChunk.GetValue("files_id").AsObjectId.ToString());
                    }
                }
                
                foreach (var files_id in uniqueFiles)
                {
                    var filter = Builders<BsonDocument>.Filter.Eq("_id", new ObjectId(files_id));
                    var count = await filesCollection.CountDocumentsAsync(filter, null, cts.Token);
                    if (count > 0)
                    {
                        continue;
                    }

                    var deleteFilter = Builders<BsonDocument>.Filter.Eq("files_id", new ObjectId(files_id));
                    var deleteResult = await chunksCollection.DeleteManyAsync(deleteFilter, null,cts.Token);
                    numberOfChunksDeleted += (ulong)deleteResult.DeletedCount;
                    Log.Information("{0},{1}", deleteResult, files_id);
                }
            }
            catch (Exception ex)
            {
                Log.Error(ex, "Something went wrong.");
            }
            finally
            {
                sw.Stop();
                Log.Information("Number of chunks deleted: {0}.", numberOfChunksDeleted);
                Log.Information("Elapsed time: {0} minutes.", sw.Elapsed.TotalMinutes);
                Log.CloseAndFlush();
            }

            return 0;
        }
    }
}

That means it is taking longer than 10 minutes between calls to MoveNextAsync() and the cursor is getting cleaned up. If you add NoCursorTimeout = true to the FindOptions<BsonDocument>, it will prevent the mongod from cleaning up the cursor, but that means unless the client cleans it up, it won't be cleaned up until a server restart. So just make sure all error paths call Dispose on the chunksCursor. The using directive should solve that for you automatically, but an explicit try...finally wouldn't hurt either.

See https://github.com/Dualog-students/GridFSCleaner