2 /* -----------------------------------------------------------------------
\r
3 * <copyright file="PollAgent.cs" company="GRNet">
\r
5 * Copyright 2011-2012 GRNET S.A. All rights reserved.
\r
7 * Redistribution and use in source and binary forms, with or
\r
8 * without modification, are permitted provided that the following
\r
9 * conditions are met:
\r
11 * 1. Redistributions of source code must retain the above
\r
12 * copyright notice, this list of conditions and the following
\r
15 * 2. Redistributions in binary form must reproduce the above
\r
16 * copyright notice, this list of conditions and the following
\r
17 * disclaimer in the documentation and/or other materials
\r
18 * provided with the distribution.
\r
21 * THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS
\r
22 * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
\r
23 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
\r
24 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR
\r
25 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
\r
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
\r
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
\r
28 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
\r
29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
\r
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
\r
31 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
\r
32 * POSSIBILITY OF SUCH DAMAGE.
\r
34 * The views and conclusions contained in the software and
\r
35 * documentation are those of the authors and should not be
\r
36 * interpreted as representing official policies, either expressed
\r
37 * or implied, of GRNET S.A.
\r
39 * -----------------------------------------------------------------------
\r
43 using System.Collections.Concurrent;
\r
44 using System.ComponentModel.Composition;
\r
45 using System.Diagnostics;
\r
46 using System.Diagnostics.Contracts;
\r
48 using System.Reflection;
\r
49 using System.Threading;
\r
50 using System.Threading.Tasks;
\r
51 using Castle.ActiveRecord;
\r
52 using Pithos.Interfaces;
\r
53 using Pithos.Network;
\r
56 namespace Pithos.Core.Agents
\r
59 using System.Collections.Generic;
\r
63 /// PollAgent periodically polls the server to detect object changes. The agent retrieves a listing of all
\r
64 /// objects and compares it with a previously cached version to detect differences.
\r
65 /// New files are downloaded, missing files are deleted from the local file system and common files are compared
\r
66 /// to determine the appropriate action
\r
69 public class PollAgent
\r
71 private static readonly ILog Log = LogManager.GetLogger(MethodBase.GetCurrentMethod().DeclaringType);
\r
73 [System.ComponentModel.Composition.Import]
\r
74 public IStatusKeeper StatusKeeper { get; set; }
\r
76 [System.ComponentModel.Composition.Import]
\r
77 public IPithosSettings Settings { get; set; }
\r
79 [System.ComponentModel.Composition.Import]
\r
80 public NetworkAgent NetworkAgent { get; set; }
\r
82 public IStatusNotification StatusNotification { get; set; }
\r
84 private bool _firstPoll = true;
\r
86 //The Sync Event signals a manual synchronisation
\r
87 private readonly AsyncManualResetEvent _syncEvent = new AsyncManualResetEvent();
\r
89 private readonly ConcurrentDictionary<string, DateTime> _lastSeen = new ConcurrentDictionary<string, DateTime>();
\r
90 private readonly ConcurrentDictionary<string, AccountInfo> _accounts = new ConcurrentDictionary<string,AccountInfo>();
\r
94 /// Start a manual synchronization
\r
96 public void SynchNow()
\r
102 /// Remote files are polled periodically. Any changes are processed
\r
104 /// <param name="since"></param>
\r
105 /// <returns></returns>
\r
106 public async Task PollRemoteFiles(DateTime? since = null)
\r
108 if (Log.IsDebugEnabled)
\r
109 Log.DebugFormat("Polling changes after [{0}]",since);
\r
111 Debug.Assert(Thread.CurrentThread.IsBackground, "Polling Ended up in the main thread!");
\r
114 using (ThreadContext.Stacks["Retrieve Remote"].Push("All accounts"))
\r
116 //If this poll fails, we will retry with the same since value
\r
117 var nextSince = since;
\r
120 UpdateStatus(PithosStatus.PollSyncing);
\r
122 var tasks = from accountInfo in _accounts.Values
\r
123 select ProcessAccountFiles(accountInfo, since);
\r
125 var nextTimes=await TaskEx.WhenAll(tasks.ToList());
\r
127 _firstPoll = false;
\r
128 //Reschedule the poll with the current timestamp as a "since" value
\r
130 if (nextTimes.Length>0)
\r
131 nextSince = nextTimes.Min();
\r
132 if (Log.IsDebugEnabled)
\r
133 Log.DebugFormat("Next Poll at [{0}]",nextSince);
\r
135 catch (Exception ex)
\r
137 Log.ErrorFormat("Error while processing accounts\r\n{0}", ex);
\r
138 //In case of failure retry with the same "since" value
\r
141 UpdateStatus(PithosStatus.PollComplete);
\r
142 //The multiple try blocks are required because we can't have an await call
\r
143 //inside a finally block
\r
144 //TODO: Find a more elegant solution for reschedulling in the event of an exception
\r
147 //Wait for the polling interval to pass or the Sync event to be signalled
\r
148 nextSince = await WaitForScheduledOrManualPoll(nextSince);
\r
152 //Ensure polling is scheduled even in case of error
\r
153 TaskEx.Run(() => PollRemoteFiles(nextSince));
\r
159 /// Wait for the polling period to expire or a manual sync request
\r
161 /// <param name="since"></param>
\r
162 /// <returns></returns>
\r
163 private async Task<DateTime?> WaitForScheduledOrManualPoll(DateTime? since)
\r
165 var sync = _syncEvent.WaitAsync();
\r
166 var wait = TaskEx.Delay(TimeSpan.FromSeconds(Settings.PollingInterval), NetworkAgent.CancellationToken);
\r
167 var signaledTask = await TaskEx.WhenAny(sync, wait);
\r
169 //Wait for network processing to finish before polling
\r
170 var pauseTask=NetworkAgent.ProceedEvent.WaitAsync();
\r
171 await TaskEx.WhenAll(signaledTask, pauseTask);
\r
173 //If polling is signalled by SynchNow, ignore the since tag
\r
174 if (sync.IsCompleted)
\r
176 //TODO: Must convert to AutoReset
\r
177 _syncEvent.Reset();
\r
183 public async Task<DateTime?> ProcessAccountFiles(AccountInfo accountInfo, DateTime? since = null)
\r
185 if (accountInfo == null)
\r
186 throw new ArgumentNullException("accountInfo");
\r
187 if (String.IsNullOrWhiteSpace(accountInfo.AccountPath))
\r
188 throw new ArgumentException("The AccountInfo.AccountPath is empty", "accountInfo");
\r
189 Contract.EndContractBlock();
\r
192 using (ThreadContext.Stacks["Retrieve Remote"].Push(accountInfo.UserName))
\r
195 await NetworkAgent.GetDeleteAwaiter();
\r
197 Log.Info("Scheduled");
\r
198 var client = new CloudFilesClient(accountInfo);
\r
200 //We don't need to check the trash container
\r
201 var containers = client.ListContainers(accountInfo.UserName)
\r
202 .Where(c=>c.Name!="trash")
\r
206 CreateContainerFolders(accountInfo, containers);
\r
208 //The nextSince time fallback time is the same as the current.
\r
209 //If polling succeeds, the next Since time will be the smallest of the maximum modification times
\r
210 //of the shared and account objects
\r
211 var nextSince = since;
\r
215 //Wait for any deletions to finish
\r
216 await NetworkAgent.GetDeleteAwaiter();
\r
217 //Get the poll time now. We may miss some deletions but it's better to keep a file that was deleted
\r
218 //than delete a file that was created while we were executing the poll
\r
220 //Get the list of server objects changed since the last check
\r
221 //The name of the container is passed as state in order to create a dictionary of tasks in a subsequent step
\r
222 var listObjects = (from container in containers
\r
223 select Task<IList<ObjectInfo>>.Factory.StartNew(_ =>
\r
224 client.ListObjects(accountInfo.UserName, container.Name, since), container.Name)).ToList();
\r
226 var listShared = Task<IList<ObjectInfo>>.Factory.StartNew(_ =>
\r
227 client.ListSharedObjects(since), "shared");
\r
228 listObjects.Add(listShared);
\r
229 var listTasks = await Task.Factory.WhenAll(listObjects.ToArray());
\r
231 using (ThreadContext.Stacks["SCHEDULE"].Push("Process Results"))
\r
233 var dict = listTasks.ToDictionary(t => t.AsyncState);
\r
235 //Get all non-trash objects. Remember, the container name is stored in AsyncState
\r
236 var remoteObjects = (from objectList in listTasks
\r
237 where (string)objectList.AsyncState != "trash"
\r
238 from obj in objectList.Result
\r
239 select obj).ToList();
\r
241 //Get the latest remote object modification date, only if it is after
\r
242 //the original since date
\r
243 nextSince = GetLatestDateAfter(nextSince, remoteObjects);
\r
245 var sharedObjects = dict["shared"].Result;
\r
246 nextSince = GetLatestDateBefore(nextSince, sharedObjects);
\r
248 //DON'T process trashed files
\r
249 //If some files are deleted and added again to a folder, they will be deleted
\r
250 //even though they are new.
\r
251 //We would have to check file dates and hashes to ensure that a trashed file
\r
252 //can be deleted safely from the local hard drive.
\r
254 //Items with the same name, hash may be both in the container and the trash
\r
255 //Don't delete items that exist in the container
\r
256 var realTrash = from trash in trashObjects
\r
258 !remoteObjects.Any(
\r
259 info => info.Name == trash.Name && info.Hash == trash.Hash)
\r
261 ProcessTrashedFiles(accountInfo, realTrash);
\r
264 var cleanRemotes = (from info in remoteObjects.Union(sharedObjects)
\r
265 let name = info.Name??""
\r
266 where !name.EndsWith(".ignore", StringComparison.InvariantCultureIgnoreCase) &&
\r
267 !name.StartsWith(FolderConstants.CacheFolder + "/",
\r
268 StringComparison.InvariantCultureIgnoreCase)
\r
269 select info).ToList();
\r
271 var differencer = _differencer.PostSnapshot(accountInfo, cleanRemotes);
\r
273 ProcessDeletedFiles(accountInfo, differencer.Deleted.FilterDirectlyBelow(SelectiveUris));
\r
275 // @@@ NEED To add previous state here as well, To compare with previous hash
\r
279 //Create a list of actions from the remote files
\r
280 var allActions = MovesToActions(accountInfo,differencer.Moved.FilterDirectlyBelow(SelectiveUris))
\r
282 ChangesToActions(accountInfo, differencer.Changed.FilterDirectlyBelow(SelectiveUris)))
\r
284 CreatesToActions(accountInfo, differencer.Created.FilterDirectlyBelow(SelectiveUris)));
\r
286 //And remove those that are already being processed by the agent
\r
287 var distinctActions = allActions
\r
288 .Except(NetworkAgent.GetEnumerable(), new LocalFileComparer())
\r
291 //Queue all the actions
\r
292 foreach (var message in distinctActions)
\r
294 NetworkAgent.Post(message);
\r
297 Log.Info("[LISTENER] End Processing");
\r
300 catch (Exception ex)
\r
302 Log.ErrorFormat("[FAIL] ListObjects for{0} in ProcessRemoteFiles with {1}", accountInfo.UserName, ex);
\r
306 Log.Info("[LISTENER] Finished");
\r
312 /// Returns the latest LastModified date from the list of objects, but only if it is before
\r
313 /// than the threshold value
\r
315 /// <param name="threshold"></param>
\r
316 /// <param name="cloudObjects"></param>
\r
317 /// <returns></returns>
\r
318 private static DateTime? GetLatestDateBefore(DateTime? threshold, IList<ObjectInfo> cloudObjects)
\r
320 DateTime? maxDate = null;
\r
321 if (cloudObjects!=null && cloudObjects.Count > 0)
\r
322 maxDate = cloudObjects.Max(obj => obj.Last_Modified);
\r
323 if (maxDate == null || maxDate == DateTime.MinValue)
\r
325 if (threshold == null || threshold == DateTime.MinValue || threshold > maxDate)
\r
331 /// Returns the latest LastModified date from the list of objects, but only if it is after
\r
332 /// the threshold value
\r
334 /// <param name="threshold"></param>
\r
335 /// <param name="cloudObjects"></param>
\r
336 /// <returns></returns>
\r
337 private static DateTime? GetLatestDateAfter(DateTime? threshold, IList<ObjectInfo> cloudObjects)
\r
339 DateTime? maxDate = null;
\r
340 if (cloudObjects!=null && cloudObjects.Count > 0)
\r
341 maxDate = cloudObjects.Max(obj => obj.Last_Modified);
\r
342 if (maxDate == null || maxDate == DateTime.MinValue)
\r
344 if (threshold == null || threshold == DateTime.MinValue || threshold < maxDate)
\r
349 readonly AccountsDifferencer _differencer = new AccountsDifferencer();
\r
350 private List<Uri> _selectiveUris=new List<Uri>();
\r
353 /// Deletes local files that are not found in the list of cloud files
\r
355 /// <param name="accountInfo"></param>
\r
356 /// <param name="cloudFiles"></param>
\r
357 private void ProcessDeletedFiles(AccountInfo accountInfo, IEnumerable<ObjectInfo> cloudFiles)
\r
359 if (accountInfo == null)
\r
360 throw new ArgumentNullException("accountInfo");
\r
361 if (String.IsNullOrWhiteSpace(accountInfo.AccountPath))
\r
362 throw new ArgumentException("The AccountInfo.AccountPath is empty", "accountInfo");
\r
363 if (cloudFiles == null)
\r
364 throw new ArgumentNullException("cloudFiles");
\r
365 Contract.EndContractBlock();
\r
370 //Only consider files that are not being modified, ie they are in the Unchanged state
\r
371 var deleteCandidates = FileState.Queryable.Where(state =>
\r
372 state.FilePath.StartsWith(accountInfo.AccountPath)
\r
373 && state.FileStatus == FileStatus.Unchanged).ToList();
\r
376 //TODO: filesToDelete must take into account the Others container
\r
377 var filesToDelete = (from deleteCandidate in deleteCandidates
\r
378 let localFile = FileInfoExtensions.FromPath(deleteCandidate.FilePath)
\r
379 let relativeFilePath = localFile.AsRelativeTo(accountInfo.AccountPath)
\r
381 !cloudFiles.Any(r => r.RelativeUrlToFilePath(accountInfo.UserName) == relativeFilePath)
\r
382 select localFile).ToList();
\r
386 //Set the status of missing files to Conflict
\r
387 foreach (var item in filesToDelete)
\r
389 //Try to acquire a gate on the file, to take into account files that have been dequeued
\r
390 //and are being processed
\r
391 using (var gate = NetworkGate.Acquire(item.FullName, NetworkOperation.Deleting))
\r
395 StatusKeeper.SetFileState(item.FullName, FileStatus.Conflict, FileOverlayStatus.Deleted);
\r
398 UpdateStatus(PithosStatus.HasConflicts);
\r
399 StatusNotification.NotifyConflicts(filesToDelete, String.Format("{0} local files are missing from Pithos, possibly because they were deleted", filesToDelete.Count));
\r
400 StatusNotification.NotifyForFiles(filesToDelete, String.Format("{0} files were deleted", filesToDelete.Count), TraceLevel.Info);
\r
404 var deletedFiles = new List<FileSystemInfo>();
\r
405 foreach (var objectInfo in cloudFiles)
\r
407 if (Log.IsDebugEnabled)
\r
408 Log.DebugFormat("Handle deleted [{0}]",objectInfo.Uri);
\r
409 var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);
\r
410 var item = FileAgent.GetFileAgent(accountInfo).GetFileSystemInfo(relativePath);
\r
411 if (Log.IsDebugEnabled)
\r
412 Log.DebugFormat("Will delete [{0}] for [{1}]", item.FullName,objectInfo.Uri);
\r
415 if ((item.Attributes & FileAttributes.ReadOnly) == FileAttributes.ReadOnly)
\r
417 item.Attributes = item.Attributes & ~FileAttributes.ReadOnly;
\r
422 Log.DebugFormat("Deleting {0}", item.FullName);
\r
424 var directory = item as DirectoryInfo;
\r
425 if (directory!=null)
\r
426 directory.Delete(true);
\r
429 Log.DebugFormat("Deleted [{0}] for [{1}]", item.FullName, objectInfo.Uri);
\r
431 _lastSeen.TryRemove(item.FullName, out lastDate);
\r
432 deletedFiles.Add(item);
\r
434 StatusKeeper.SetFileState(item.FullName, FileStatus.Deleted, FileOverlayStatus.Deleted);
\r
436 Log.InfoFormat("[{0}] files were deleted",deletedFiles.Count);
\r
437 StatusNotification.NotifyForFiles(deletedFiles, String.Format("{0} files were deleted", deletedFiles.Count), TraceLevel.Info);
\r
443 /// Creates a Sync action for each changed server file
\r
445 /// <param name="accountInfo"></param>
\r
446 /// <param name="changes"></param>
\r
447 /// <returns></returns>
\r
448 private IEnumerable<CloudAction> ChangesToActions(AccountInfo accountInfo, IEnumerable<ObjectInfo> changes)
\r
450 if (changes == null)
\r
451 throw new ArgumentNullException();
\r
452 Contract.EndContractBlock();
\r
453 var fileAgent = FileAgent.GetFileAgent(accountInfo);
\r
455 //In order to avoid multiple iterations over the files, we iterate only once
\r
456 //over the remote files
\r
457 foreach (var objectInfo in changes)
\r
459 var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);
\r
460 //If a directory object already exists, we may need to sync it
\r
461 if (fileAgent.Exists(relativePath))
\r
463 var localFile = fileAgent.GetFileSystemInfo(relativePath);
\r
464 //We don't need to sync directories
\r
465 if (objectInfo.Content_Type == @"application/directory" && localFile is DirectoryInfo)
\r
467 using (new SessionScope(FlushAction.Never))
\r
469 var state = StatusKeeper.GetStateByFilePath(localFile.FullName);
\r
470 _lastSeen[localFile.FullName] = DateTime.Now;
\r
471 //Common files should be checked on a per-case basis to detect differences, which is newer
\r
473 yield return new CloudAction(accountInfo, CloudActionType.MustSynch,
\r
474 localFile, objectInfo, state, accountInfo.BlockSize,
\r
475 accountInfo.BlockHash);
\r
480 //Remote files should be downloaded
\r
481 yield return new CloudDownloadAction(accountInfo, objectInfo);
\r
487 /// Creates a Local Move action for each moved server file
\r
489 /// <param name="accountInfo"></param>
\r
490 /// <param name="moves"></param>
\r
491 /// <returns></returns>
\r
492 private IEnumerable<CloudAction> MovesToActions(AccountInfo accountInfo, IEnumerable<ObjectInfo> moves)
\r
495 throw new ArgumentNullException();
\r
496 Contract.EndContractBlock();
\r
497 var fileAgent = FileAgent.GetFileAgent(accountInfo);
\r
499 //In order to avoid multiple iterations over the files, we iterate only once
\r
500 //over the remote files
\r
501 foreach (var objectInfo in moves)
\r
503 var previousRelativepath = objectInfo.Previous.RelativeUrlToFilePath(accountInfo.UserName);
\r
504 //If the previous file already exists, we can execute a Move operation
\r
505 if (fileAgent.Exists(previousRelativepath))
\r
507 var previousFile = fileAgent.GetFileSystemInfo(previousRelativepath);
\r
508 using (new SessionScope(FlushAction.Never))
\r
510 var state = StatusKeeper.GetStateByFilePath(previousFile.FullName);
\r
511 _lastSeen[previousFile.FullName] = DateTime.Now;
\r
513 //For each moved object we need to move both the local file and update
\r
514 yield return new CloudAction(accountInfo, CloudActionType.RenameLocal,
\r
515 previousFile, objectInfo, state, accountInfo.BlockSize,
\r
516 accountInfo.BlockHash);
\r
517 //For modified files, we need to download the changes as well
\r
518 if (objectInfo.Hash!=objectInfo.PreviousHash)
\r
519 yield return new CloudDownloadAction(accountInfo,objectInfo);
\r
522 //If the previous file does not exist, we need to download it in the new location
\r
525 //Remote files should be downloaded
\r
526 yield return new CloudDownloadAction(accountInfo, objectInfo);
\r
533 /// Creates a download action for each new server file
\r
535 /// <param name="accountInfo"></param>
\r
536 /// <param name="creates"></param>
\r
537 /// <returns></returns>
\r
538 private IEnumerable<CloudAction> CreatesToActions(AccountInfo accountInfo, IEnumerable<ObjectInfo> creates)
\r
540 if (creates == null)
\r
541 throw new ArgumentNullException();
\r
542 Contract.EndContractBlock();
\r
543 var fileAgent = FileAgent.GetFileAgent(accountInfo);
\r
545 //In order to avoid multiple iterations over the files, we iterate only once
\r
546 //over the remote files
\r
547 foreach (var objectInfo in creates)
\r
549 if (Log.IsDebugEnabled)
\r
550 Log.DebugFormat("[NEW INFO] {0}",objectInfo.Uri);
\r
552 var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);
\r
553 //If the object already exists, we probably have a conflict
\r
554 if (fileAgent.Exists(relativePath))
\r
556 Log.DebugFormat("[SKIP EXISTING] {0}", objectInfo.Uri);
\r
557 //If a directory object already exists, we don't need to perform any other action
\r
558 var localFile = fileAgent.GetFileSystemInfo(relativePath);
\r
559 StatusKeeper.SetFileState(localFile.FullName, FileStatus.Conflict, FileOverlayStatus.Conflict);
\r
563 //Remote files should be downloaded
\r
564 yield return new CloudDownloadAction(accountInfo, objectInfo);
\r
570 /// Notify the UI to update the visual status
\r
572 /// <param name="status"></param>
\r
573 private void UpdateStatus(PithosStatus status)
\r
577 StatusNotification.SetPithosStatus(status);
\r
578 //StatusNotification.Notify(new Notification());
\r
580 catch (Exception exc)
\r
582 //Failure is not critical, just log it
\r
583 Log.Warn("Error while updating status", exc);
\r
587 private static void CreateContainerFolders(AccountInfo accountInfo, IEnumerable<ContainerInfo> containers)
\r
589 var containerPaths = from container in containers
\r
590 let containerPath = Path.Combine(accountInfo.AccountPath, container.Name)
\r
591 where container.Name != FolderConstants.TrashContainer && !Directory.Exists(containerPath)
\r
592 select containerPath;
\r
594 foreach (var path in containerPaths)
\r
596 Directory.CreateDirectory(path);
\r
600 public void SetSyncUris(Uri[] uris)
\r
602 SelectiveUris=uris.ToList();
\r
605 protected List<Uri> SelectiveUris
\r
607 get { return _selectiveUris;}
\r
608 set { _selectiveUris = value; }
\r
611 public void AddAccount(AccountInfo accountInfo)
\r
613 //Avoid adding a duplicate accountInfo
\r
614 _accounts.TryAdd(accountInfo.UserName, accountInfo);
\r
617 public void RemoveAccount(AccountInfo accountInfo)
\r
619 AccountInfo account;
\r
620 _accounts.TryRemove(accountInfo.UserName,out account);
\r
621 SnapshotDifferencer differencer;
\r
622 _differencer.Differencers.TryRemove(accountInfo.UserName, out differencer);
\r