[Eulero]

Good evening,

i'm downloading a file (eD2K link with AICH) with ZZUL-TRA (but i think the problem is the same with official), that has 1 complete source, and about 600 incomplete sources.

In Verbose i read dozens of req for AICH packet, all failed!

24/04/2012 17.29.28: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 79.18.XXX.XXX 'http://emule-project.net «MorphXT 12.6»' (eMule v0.50a [MorphXT 12.6],OnQueue/OnUploadQueue/None)
24/04/2012 17.30.08: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 87.9.XXX.XXX 'http://emule-project.net' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.30.10: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 93.40.XXX.XXX 'http://www.adunanza.net' (eMule v0.48a [AdunanzA 3.16],OnQueue/OnUploadQueue/None)
24/04/2012 17.31.08: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 94.37.XXX.XXX 'zoronocha «MorphXT 12.6»' (eMule v0.50a [MorphXT 12.6],OnQueue/OnUploadQueue/None)
24/04/2012 17.31.39: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 87.11.XXX.XXX 'http://emule-project.net' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.32.21: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 95.246.XXX.XXX 'http://emule-project.net' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.32.54: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 7411480@88.191.XXX.XXX (93.56.XXX.XXX) 'Alessandra' (aMule v2.2.2 [AdunanzA 3.14b3],None/OnUploadQueue/None)
24/04/2012 17.33.24: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 9107326@88.191.XXX.XXX (93.36.224.19) 'http://www.adunanza.net' (eMule v0.48a [AdunanzA 3.16],OnQueue/OnUploadQueue/None)
24/04/2012 17.33.29: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 2.234.XXX.XXX 'http://www.adunanza.net' (eMule v0.48a [AdunanzA 3.16],OnQueue/OnUploadQueue/None)
24/04/2012 17.34.33: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 95.253.XXX.XXX 'http://emule-Luke' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.35.14: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 87.11.XXX.XXX 'paperino' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.35.24: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 80.116.XXX.XXX 'http://emule-project.net' (aMule v2.2.6,OnQueue/OnUploadQueue/None)
24/04/2012 17.35.33: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 79.41.XXX.XXX 'http://emule-project.net' (eMule v0.50a,OnQueue/OnUploadQueue/None)
24/04/2012 17.36.57: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 80.116.XXX.XXX 'MARCHINO' (eMule v0.49c,None/OnUploadQueue/None)
24/04/2012 17.37.13: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 2.226.XXX.XXX 'Tremendone' (eMule v0.48a [AdunanzA 3.16],OnQueue/OnUploadQueue/None)
24/04/2012 17.37.18: AICH Packet Request: Failed to create recoverydata - Hashset not ready or requested Hash differs from Masterhash for SomeFile.mkv to 217.133.XXX.XXX 'anonymous «MorphXT 12.6»' (eMule v0.50a [MorphXT 12.6],OnQueue/OnUploadQueue/None)
....
....

It's a normal behavior?

If no one sends the recovery packet (excluding releaser), eMule is able to recover the corrupted parts?


thanks in advance.

[Stulle]

Nope, not normal. Try using the official client, just to be sure. Seems like something went wrong with the AICH Hashset your client retrieved... or failed to retrieve. Could also be that your initial link or the file in itself is bogus.

This post has been edited by Stulle: 24 April 2012 - 04:52 PM

[Eulero]

i don't know if it's useful, but i did this test:

From the details i copied the eD2K link with the option "add full hashset" enabled.

Then i asked at releaser to send me the complete link with "add full hashset" enabled.

The two links match.


now i just opened the officiale 0.50a , but i still have the same failures.


This is the Details window:

[xilolee]

That file seems actually incomplete.
Anyway when a part is corrupted, emule starts asking for ed2k hash... The request fails and it appears that there isn't a masterhash:

Requesting AICH Hash (highid or lowid) from client xxx
AICH Request failed, Trying to ask another client (file yyy, client xxx)
Unable to request AICH Recoverydata because we have no trusted Masterhash

Given that the file seems to have got (actually) 1039 sources, we'll see hundreds of those messages.

[Riso64Bit]

i can confirm this behavior.

the first & only one completed source is an MorphXT whit HideOverShare option enabled. if i try to connet to releaser i can watch the n° of parts shared (for example 30%, The client then displays the percentage of remaining to share (70%)). some Suspect Client (for me are the cause of problem)are in possession of adjacent parts not yet shared (every at the "bottom" of the file but whit different %) & if i try to connect these clients my Xtreme receive Every time corrupted parts but try to recover asking all clients for AICH (or MasterHASH) but generate the same message posted from Eulero & xilolee.

I have write a part of DLP to detect these bad clients & i don't receive corrupted part anymore.
From my tests these bad clients are all amule clients.
default name: "http://emule-project.net"
clientversion: "aMule v2.2.6"
ALL clints are iTALiAN (1 is GREEK), ALL dynamic IP)
I have All the HASH (@Stulle i can send you an ED2k link & all the HASH of these bad clients).

Quote

26/04/2012 16.48.56: [iTALiANaMuleCorruptDataSender](reduce score)- Client 176.92.XX.XXX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)
26/04/2012 16.48.58: [iTALiANaMuleCorruptDataSender](reduce score)- Client 84.222.XX.XXX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)
26/04/2012 16.54.37: [iTALiANaMuleCorruptDataSender](reduce score)- Client 82.55.XXX.XX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)

This is not the first time we have this problem.

My question is: Why if I hav a link whit MASTERHASH full (+AICH+ED2K link) & if i receive corrupted parts emule try to ask other client for verify the corrupted parts?

This post has been edited by Riso64Bit: 26 April 2012 - 03:32 PM

[SS1900]

http://forum.emulefu...5797/#post59522

[Stulle]

Sounds interesting. Riso, if those clients just a couple of hashes it might be worth forwarding them to me.

[xilolee]

Quote

26/04/2012 19.20.34: Download session started. User: 190.71.224.192 'http://emule-project.net' (aMule v2.2.6,OnQueue/OnUploadQueue/None) in SetDownloadState(). New State: 0

26/04/2012 19.21.15: Requesting AICH Hash (HighId) from client xxx.xxx.xxx.xxx 'http://emule-project.net «MorphXT 12.6»' (eMule v0.50a [MorphXT 12.6],NoNeededParts/OnUploadQueue/None)

26/04/2012 19.21.15: AICH Request failed, Trying to ask another client (file SomeFile, Part: 830, Clientxxx.xxx.xxx.xxx 'http://emule-project.net «MorphXT 12.6»' (eMule v0.50a [MorphXT 12.6],NoNeededParts/OnUploadQueue/None))

26/04/2012 19.22.30: Download session ended: Disconnected: CClientReqSocket::Disconnect(): Close User: 190.71.224.192 'http://emule-project.net' (aMule v2.2.6,Downloading/OnUploadQueue/None) in SetDownloadState(). New State: 1, Length: 1:56 min, Payload: 27.30 MB, Transferred: 30.51 KB, Req blocks not yet completed: 3.

http://www.freeimagehosting.net/jqwq2




user hash: DD708919730ED49747E3D1F52AA36F6C



ANOTHER:

Quote

26/04/2012 23.56.08: Download session started. User: 1@0.0.0.0 (93.56.240.8) 'http://emule-project.net' (aMule v2.2.6,OnQueue/OnUploadQueue/None) in SetDownloadState(). New State: 0

27/04/2012 0.00.59: Download session ended: Disconnected: CClientReqSocket::Disconnect(): Close User: 1@0.0.0.0 (93.56.240.8) 'http://emule-project.net' (aMule v2.2.6,Downloading/OnUploadQueue/None) in SetDownloadState(). New State: 1, Length: 4:51 min, Payload: 55.66 MB, Transferred: 62.19 KB, Req blocks not yet completed: 3.

user hash: 630A2A37C70E3E409ED3642EB07A6FCA




ps: they don't download anything from me, although i have 88.5% of the file (actually)

This post has been edited by xilolee: 26 April 2012 - 10:29 PM

[Riso64Bit]

from amule changelog i can't find any fix or not fixed bug relative this problem.


new hash added

Quote

27/04/2012 10.22.20: [iTALiANaMuleCorruptDataSender](reduce score)- Client 84.222.77.XX'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)
27/04/2012 10.23.04: [iTALiANaMuleCorruptDataSender](reduce score)- Client 190.71.224.XXX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)
27/04/2012 10.25.26: [iTALiANaMuleCorruptDataSender](reduce score)- Client 79.40.42.XXX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)
27/04/2012 10.25.51: [iTALiANaMuleCorruptDataSender](reduce score)- Client 176.92.80.XXX 'http://emule-project.net' (aMule v2.2.6,Connecting/None/None)

This post has been edited by Riso64Bit: 27 April 2012 - 08:35 AM

[tHeWiZaRdOfDoS]

Reduce score is nonsense... if you're almost sure that it's sending bad blocks BAN or FILTER it completely.

Riso64Bit, on 26 April 2012 - 03:36 PM, said:

the first & only one completed source is an MorphXT whit HideOverShare option enabled. if i try to connet to releaser i can watch the n° of parts shared (for example 30%, The client then displays the percentage of remaining to share (70%)). some Suspect Client (for me are the cause of problem)are in possession of adjacent parts not yet shared (every at the "bottom" of the file but whit different %) & if i try to connect these clients my Xtreme receive Every time corrupted parts but try to recover asking all clients for AICH (or MasterHASH) but generate the same message posted from Eulero & xilolee.

THIS is the interesting part IMHO. Basically HOS hurts the release-process as eMule won't ask the original source for the hashes for the parts as they appear to be hidden... another case for anti-hideos.

Quote

My question is: Why if I hav a link whit MASTERHASH full (+AICH+ED2K link) & if i receive corrupted parts emule try to ask other client for verify the corrupted parts?

eMule can't calculate the part and block hashes using the masterhash but calculates the masterhash using the block hashes... you can't revert that process. Once a part has been detected as being corrupt eMule sends that packet to verify which blocks are corrupt. Usually that lowers the amount of "lost" data due to corruption from 9.28MB to 180kB - quite some bargain, isn't it?

[fox88]

tHeWiZaRdOfDoS, on 28 April 2012 - 09:14 AM, said:

Usually that lowers the amount of "lost" data due to corruption from 9.28MB to 180kB - quite some bargain, isn't it?

Somehow I thought that eMule could recalculate part hash after receiving every 180kB. That would almost half expectation of the amount to redownload when/if AICH is not available.

This post has been edited by fox88: 28 April 2012 - 06:33 AM

[Eulero]

tHeWiZaRdOfDoS, on 28 April 2012 - 08:14 AM, said:

THIS is the interesting part IMHO. Basically HOS hurts the release-process as eMule won't ask the original source for the hashes for the parts as they appear to be hidden... another case for anti-hideos.

As you can see from the first post, i used ZZUL-TRA with AntiHideOS [netfinity /WiZaRd].
So, this means that it does not "detect" correctly the HOS ?


Anyway, my question is why all other sources "Failed to create recoverydata"?

This post has been edited by Eulero: 28 April 2012 - 07:45 AM

[tHeWiZaRdOfDoS]

Anti-HideOS is only displaying the hidden parts and maybe requesting parts that are hidden but only if no others can be asked for - Morph on the other hand features anti-anti-HideOS which blocks such requests.
I'm not sure but I think the AICH recovery process isn't altered - yet - it should be considering the scenario you encountered.


EDIT: I just checked the code and it seems that eMule isn't validating the available parts before sending an AICH recovery request. This *seems* a bit strange given that a client may propagate having the AICH hash even without having a single byte of the file. It may also be the reason for having to send hundreds of requests in some situations without getting a proper response.

[xilolee]

Or the releaser didn't send the masterhash to anyone?

[Riso64Bit]

@xilolee:

xilolee, on 28 April 2012 - 01:26 PM, said:

Or the releaser didn't send the masterhash to anyone?

I have received the full link whit masterhash form the releaser.this isn't the problrem (for me)

Xtreme mod have this bug (at today NOT FIXED):

Quote

Known Bugs:
A bug in Hashset request processing. This bug may cause some trouble while releasing new files.

if the releaser have MorphXT (MorphXT is'n based on Xtreme.it's True?),why have the same problem?

@tHeWiZaRdOfDoS:
1)thanks for the info about MasterHash
2)reduce score is only for debug, for check new bad hash & detect false positive.

This post has been edited by Riso64Bit: 28 April 2012 - 12:48 PM

[xilolee]

No one had got the masterhash, so the releaser didn't send it.
Maybe the bug is in the HideOs?

[Riso64Bit]

@xilolee: ok we made a test? read the PM.

[xilolee]

Where is it??

[Riso64Bit]

from OUR test the problem isn't on the releaser (test whit 0.50a to Xtreme & Xtreme to 0.50a)

Quote

[16:22] **Riso64Bit**: 28/04/2012 16.21.46: SXRecv: Client source request; XXX.XXX.XXX.XX 'XILOLEE' (eMule v0.50a,None/None/None), File="Nuovo Archivio WinRAR.rar"
[16:22] XILOLEE: 28/04/2012 16.22.10: Received valid MD4 Hashset (FileIdentifiers) form XXX.XXX.XXX.XX '**Riso64Bit** ['f@v] «Xtreme 8.1»' (eMule v0.50a [Xtreme 8.1],ReqHashSet/OnUploadQueue/None), file: Nuovo Archivio WinRAR.rar
[16:22] **Riso64Bit**: 28/04/2012 16.22.12: Adding client to upload list: Accept new client, because other client Upload end soon Client: XXX.XXX.XXX.XX 'XILOLEE' (eMule v0.50a,None/None/None)
[16:23] XILOLEE: 28/04/2012 16.22.10: Received valid AICH Part Hashset form XXX.XXX.XXX.XX '**Riso64Bit** ['f@v] «Xtreme 8.1»' (eMule v0.50a [Xtreme 8.1],ReqHashSet/OnUploadQueue/None), file: Nuovo Archivio WinRAR.rar
[16:24] XILOLEE: 28/04/2012 16.22.36: Download session started. User: XXX.XXX.XXX.XX '**Riso64Bit** ['f@v] «Xtreme 8.1»' (eMule v0.50a [Xtreme 8.1],OnQueue/OnUploadQueue/None) in SetDownloadState(). New State: 0

EDIT: MASTERHASH sended & received


test whit MorphXT:

@Stulle: is normal?
"Hashing :nameoffile" (should be "Hashing: nameoffile"?)

This post has been edited by Riso64Bit: 28 April 2012 - 03:07 PM

[Eulero]

i think that this test is incomplete.

if there isn't corruption , you can't verify the request of recoverydata.