public static async Task Test0Async(string[] args)
{
const string V2FeedCountQuery = "/Packages/$count";
Console.WriteLine("Simple count test for distinct package ids and version between v2 feed and catalog");
if (args.Length != 2)
{
Console.WriteLine("Please enter only 2 arguments. First v2gallery feed url, and second catalog index.json url");
return;
}
else
{
string v2FeedUrl = args[0].TrimEnd('/');
string v2FeedCountUrl = v2FeedUrl.TrimEnd('/') + V2FeedCountQuery;
int v2FeedCount = 0;
using (var client = new HttpClient())
{
var response = await client.GetAsync(v2FeedCountUrl);
string v2FeedCountString = await response.Content.ReadAsStringAsync();
v2FeedCount = Int32.Parse(v2FeedCountString);
}
string catalog = args[1];
Uri catalogIndex = new Uri(catalog);
CatalogIndexReader reader = new CatalogIndexReader(catalogIndex);
var task = reader.GetEntries();
task.Wait();
var entries = task.Result;
Console.WriteLine("Total packages count from catalog is " + entries.Count());
var distinctCatalogPackages = entries.Distinct(new CatalogIndexEntryIdVersionComparer());
int v3CatalogPackagesCount = distinctCatalogPackages.Count();
Console.WriteLine("Distinct packages count from catalog is " + v3CatalogPackagesCount);
Console.WriteLine("Distinct packages count from " + v2FeedUrl + " is " + v2FeedCount);
Console.WriteLine("Current difference between v2Feed and v3 catalog is " + (v2FeedCount - v3CatalogPackagesCount));
}
}
private async Task PrepareAsync()
{
_log.WriteLine("Making sure folder {0} exists.", _outputFolder);
if (!Directory.Exists(_outputFolder))
{
Directory.CreateDirectory(_outputFolder);
}
// Create reindex file
_log.WriteLine("Start preparing lightning reindex file...");
var latestCommit = DateTime.MinValue;
int numberOfEntries = 0;
string indexFile = Path.Combine(_outputFolder, "index.txt");
using (var streamWriter = new StreamWriter(indexFile, false))
{
var collectorHttpClient = new CollectorHttpClient();
var catalogIndexReader = new CatalogIndexReader(new Uri(_catalogIndex), collectorHttpClient);
var catalogIndexEntries = await catalogIndexReader.GetEntries();
foreach (var packageRegistrationGroup in catalogIndexEntries
.OrderBy(x => x.CommitTimeStamp)
.ThenBy(x => x.Id)
.ThenBy(x => x.Version)
.GroupBy(x => x.Id))
{
streamWriter.WriteLine("Element@{0}. {1}", numberOfEntries++, packageRegistrationGroup.Key);
var latestCatalogPages = new Dictionary<string, Uri>();
foreach (CatalogIndexEntry catalogIndexEntry in packageRegistrationGroup)
{
string key = catalogIndexEntry.Version.ToNormalizedString();
if (latestCatalogPages.ContainsKey(key))
{
latestCatalogPages[key] = catalogIndexEntry.Uri;
}
else
{
latestCatalogPages.Add(key, catalogIndexEntry.Uri);
}
if (latestCommit < catalogIndexEntry.CommitTimeStamp)
{
latestCommit = catalogIndexEntry.CommitTimeStamp;
}
}
foreach (var latestCatalogPage in latestCatalogPages)
{
streamWriter.WriteLine("{0}", latestCatalogPage.Value);
}
}
}
_log.WriteLine("Finished preparing lightning reindex file. Output file: {0}", indexFile);
// Write cursor to storage
_log.WriteLine("Start writing new cursor...");
var account = CloudStorageAccount.Parse(_storageAccount);
var storageFactory = (StorageFactory)new AzureStorageFactory(account, _storageContainer);
var storage = storageFactory.Create();
var cursor = new DurableCursor(storage.ResolveUri("cursor.json"), storage, latestCommit);
cursor.Value = latestCommit;
await cursor.Save(CancellationToken.None);
_log.WriteLine("Finished writing new cursor.");
// Write command files
_log.WriteLine("Start preparing lightning reindex command files...");
string templateFileContents;
using (var templateStreamReader = new StreamReader(_templateFile))
{
templateFileContents = await templateStreamReader.ReadToEndAsync();
}
int batchNumber = 0;
int batchSizeValue = int.Parse(_batchSize);
for (int batchStart = 0; batchStart < numberOfEntries; batchStart += batchSizeValue)
{
var batchEnd = (batchStart + batchSizeValue - 1);
if (batchEnd >= numberOfEntries)
{
batchEnd = numberOfEntries - 1;
}
var cursorCommandFileName = "cursor" + batchNumber + ".cmd";
var cursorTextFileName = "cursor" + batchNumber + ".txt";
using (var cursorCommandStreamWriter = new StreamWriter(Path.Combine(_outputFolder, cursorCommandFileName)))
{
using (var cursorTextStreamWriter = new StreamWriter(Path.Combine(_outputFolder, cursorTextFileName)))
{
var commandStreamContents = templateFileContents
.Replace("[index]", indexFile)
.Replace("[cursor]", cursorTextFileName)
.Replace("[contentbaseaddress]", _contentBaseAddress)
.Replace("[storageaccount]", _storageAccount)
.Replace("[storagecontainer]", _storageContainer)
.Replace("[storagebaseaddress]", _storageBaseAddress)
.Replace("[compress]", _compress.ToString().ToLowerInvariant());
await cursorCommandStreamWriter.WriteLineAsync(commandStreamContents);
await cursorTextStreamWriter.WriteLineAsync(batchStart + "," + batchEnd);
}
}
batchNumber++;
}
_log.WriteLine("Finished preparing lightning reindex command files.");
_log.WriteLine("You can now copy the {0} file and all cursor*.cmd, cursor*.txt", indexFile);
_log.WriteLine("to multiple machines and run the cursor*.cmd files in parallel.");
}
public override async Task RunCore(CancellationToken cancellationToken)
{
int nextMasterRegId = 0;
DateTime position = Cursor.Position;
JToken nextMasterRegIdToken = null;
if (Cursor.Metadata.TryGetValue("nextMasterRegId", out nextMasterRegIdToken))
{
nextMasterRegId = nextMasterRegIdToken.ToObject<int>();
}
// Get the catalog index
Uri catalogIndexUri = new Uri(Config.GetProperty("CatalogIndex"));
Log("Reading index entries");
var indexReader = new CatalogIndexReader(catalogIndexUri);
var indexEntries = await indexReader.GetRolledUpEntries();
var context = indexReader.GetContext();
Log("Finding new or editted entries");
var changedEntries = new HashSet<string>(indexEntries.Where(e => e.CommitTimeStamp.CompareTo(position) > 0)
.Select(e => e.Id.ToLowerInvariant()), StringComparer.OrdinalIgnoreCase);
DateTime newPosition = indexEntries.Select(e => e.CommitTimeStamp).OrderByDescending(e => e).FirstOrDefault();
ConcurrentDictionary<string, ConcurrentBag<Uri>> batches = new ConcurrentDictionary<string, ConcurrentBag<Uri>>(StringComparer.OrdinalIgnoreCase);
ParallelOptions options = new ParallelOptions();
options.MaxDegreeOfParallelism = 8;
Parallel.ForEach(indexEntries, options, entry =>
{
if (changedEntries.Contains(entry.Id))
{
batches.AddOrUpdate(entry.Id, new ConcurrentBag<Uri>() { entry.Uri }, (id, uris) =>
{
uris.Add(entry.Uri);
return uris;
});
}
});
Uri contentBaseAddress = new Uri(Config.GetProperty("ContentBaseAddress"));
if (batches.Count > 0)
{
Log("Building registrations from: " + position.ToString("O"));
options.MaxDegreeOfParallelism = 4;
for (int i = 0; i < 3 && batches.Count > 0; i++)
{
if (i != 0)
{
options.MaxDegreeOfParallelism = 1;
Console.WriteLine("Single batch run.");
}
var ids = batches.Keys.OrderBy(s => s).ToArray();
Stopwatch buildTimer = new Stopwatch();
buildTimer.Start();
int startingCount = ids.Length;
Parallel.ForEach(ids, options, id =>
{
try
{
BatchRegistrationCollector regCollector = new BatchRegistrationCollector(null, _factory);
regCollector.ContentBaseAddress = contentBaseAddress;
Stopwatch timer = new Stopwatch();
timer.Start();
var uriGroup = batches[id].ToArray();
regCollector.ProcessGraphs(_client, id, uriGroup, context, cancellationToken).Wait();
int rem = batches.Count;
timer.Stop();
string log = String.Format("Completed: {0} Duration: {1} Uris: {2} Remaining Ids: {3} Loop: {4}", id, timer.Elapsed, uriGroup.Length, rem, i);
Console.WriteLine(log);
// stats
double perPackage = buildTimer.Elapsed.TotalSeconds / (double)(startingCount - rem + 1);
DateTime finish = DateTime.Now.AddSeconds(Math.Ceiling(perPackage * rem));
Console.WriteLine("Estimated Finish: " + finish.ToString("O"));
ConcurrentBag<Uri> vals;
if (!batches.TryRemove(id, out vals))
{
Console.WriteLine("Unable to remove!");
}
}
catch (Exception ex)
{
LogError("Registration failed: " + id + " " + ex.ToString());
}
});
}
// mark this with the last commit we included
Cursor.Position = newPosition;
await Cursor.Save();
Log("Finished registrations: " + newPosition.ToString("O"));
}
}
