Pirates@Home logo

Pirates@Home

Berkeley Open Infrastructure
BOINC!
for Network Computing
Home Help Status Forums Glossary Account

Treachery amongst us?

log in

Advanced search

Message boards : Number Crunching : Treachery amongst us?

Author Message
STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 1953 - Posted: 16 Jun 2005 | 10:28:27 UTC
Last modified: 16 Jun 2005 | 10:45:00 UTC

Sorry to say I'm going to drop this Project, there is some highly suspicious activity going on with the Credits being given out to certain people IMO ... I don't mind a little competition but when I see over 100 Credits being given out for a WU that shouldn't take more than a few minutes to complete & all I'm getting is .15 Credits for the same type WU's then something is rotten in Denmark I think ...

Profile Rusty
Chief Petty Officer
Boatswain
Avatar
Send message
Joined: 29 Jun 04
United States
Away
Credit: 1,911.4
RAC: 0.00
Joined: Jun 29, 2004
Verified: NEVER
Dubloons: 1
Pieces of Eight: 3
Punishment: Mess Duty
Message 1955 - Posted: 16 Jun 2005 | 11:38:55 UTC - in response to Message 1953.

> Sorry to say I'm going to drop this Project, there is some highly suspicious
> activity going on with the Credits being given out to certain people IMO ... I
> don't mind a little competition but when I see over 100 Credits being given
> out for a WU that shouldn't take more than a few minutes to complete & all
> I'm getting is .15 Credits for the same type WU's then something is rotten in
> Denmark I think ...
>

You are right, that should not happen. Especially with the WU's we are giving out. So there must be a problem somewhere, and this project is all about finding and fixing problems in BOINC and the screensavers. Can you point out the WU's that are getting so much credit, so we can track down this problem?


____________
-- Rusty McGee Johnson

STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 1956 - Posted: 16 Jun 2005 | 12:02:45 UTC
Last modified: 16 Jun 2005 | 12:14:02 UTC

This host is littered with granted credit over 100 ... Some have almost 170 credits granted ...

This is just one of them ... [url=http://pirates.vassar.edu/workunit.php?wuid=190995] WU

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1958 - Posted: 16 Jun 2005 | 13:23:12 UTC
Last modified: 16 Jun 2005 | 13:36:37 UTC

Well it's one thing to be hacked from the outside. We can repel all borders, and have done so recently.

It's quite another thing to have a member of the crew gaming the system.

It seems one of us may have found a way to do so.

Now it might be a bug, in which case we track it down and squash it. That's part of our job.

Or it might be he's figured out a way around the credit system. I have my suspicions on what's up, but I want to hear his explaination first.

If he comes clean and helps us fix the problem, then there is no problem.

And if he doesn't come clean...?

...the expanded choice of project preferences is back on-line, and he can choose his prefered punishment, subject to the suggestions of the crew and the will of the captain.

[Edit] I've moved the posts on this topic to a new thread.

____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

kolch
Avatar
Send message
Joined: 22 Jul 04
Australia
BOINC@AUSTRALIA
Credit: 644.4
RAC: 0.00
Joined: Jul 22, 2004
Verified: NEVER
Message 1959 - Posted: 16 Jun 2005 | 13:58:10 UTC

A couple of points...

Either it is too hard, not always possible to, or as a result of a bug that some WU's from this host are claiming high credit. The majority of the results claimed reasonable credit amounts.

The large claims are not always granted.

I think it is most likely a bug, or maybe a cheat made possible by the more relaxed rules of granting credit for this project. Just my thoughts.
____________

STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 1961 - Posted: 16 Jun 2005 | 15:01:10 UTC
Last modified: 16 Jun 2005 | 15:34:55 UTC

The problem I had with the whole thing was that I know this person keeps on top of his computers enough to know he was getting this type of Credit and if it was a bug then he should have reported himself ...

Being that I know this person I really hated to report it myself & it probably isn't going to sit very well with him, but what is right is right I feel ...

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1962 - Posted: 16 Jun 2005 | 16:09:05 UTC - in response to Message 1961.

> The problem I had with the whole thing was that I know this person keeps on
> top of his computers enough to know he was getting this type of Credit and if
> it was a bug then he should have reported himself ...
>
> Being that I know this person I really hated to report it myself & it
> probably isn't going to sit very well with him, but what is right is right I
> feel ...

Keep in mind that the point of this project is to test BOINC (and screensaver graphics, of course). If a problem has been found it should be reported, investigated, and fixed. Credit isn't the most important thing on this particular project.

I also believe this person is on top of his machines enough to know what is going on. If he's done this intentionally, then he may have even jacked up the credit enough that it would be noticed, just to get our attention. Or not. I'm hoping he'll let us know what is up.

Otherwise we might have to convene a Court Martial! ;-)


____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 1963 - Posted: 16 Jun 2005 | 16:35:23 UTC
Last modified: 16 Jun 2005 | 16:42:32 UTC

Treachery amongst us?

No. But curiosity if it works. Yes it did:(

If it helps, here are my sincere apology.

Hope certain person can refind now the sleep..and calm down

Nobody got a brocken leg, taking a more realistic view, those are just numbers.

That what happen on Sunday morning, that's a real problem with a much higher impact.

Indeed boinc is working great as designed;) In every point of view.

Before this project will close, I just had to get seccured Information.

Experiments are not limited to the people of boinc. Sometime the rat is leaving the kittchen.

What a bug is, depends only of the point of view, talk to "some" developers stuff, they never believe the can be bug(s), even if "everybody" can feel and see it..

finaly some words, it social, as far as I could see, several other people got also more (indirectly) as they estimated to get... therefore this is social.

Also the very unique Setup of this project did it's part, so the question could be (construtive modus) why, dear boinc do you allow this technically??


so relax, it was a limited trial.

alls the "points" so far, have been done on regular way.

Started selfpunishement, using alpha releases of cc:(

I will stop posting here and not giving reply,

if somebody realy wants/needes to know more, feel free to send a email

No experiments, no progress:)


no sig

vahallah!

added:
(how some can be limited to credits! What can you buy for? Nothing.

Much more concerned by the security lacks of the boinc architecture

Reported them last year, no answer got so far..)



Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1964 - Posted: 16 Jun 2005 | 17:09:22 UTC - in response to Message 1963.


> Also the very unique Setup of this project did it's part, so the question
> could be (construtive modus) why, dear boinc do you allow this technically??


I am curious on this point. The app you used for this had a quorum of 1, since they were only minor tests and I wanted to get them back quickly. I've since set the quorum to 2, which should prevent such mischief. [Flash to scene in Hunt for Red October where Tupelov is trying desparately to remove the safety settings from all his torpedoes :-) ]

Do you agree that this is enough to prevent this little trick?

What else can we learn from it?


____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Ageless
Chief Petty Officer
Volunteer tester
Avatar
Send message
Joined: 20 Jul 04
Netherlands
Machinae Supremacy
Credit: 1,295.9
RAC: 0.00
Joined: Jul 20, 2004
Verified: Jul 9, 2011
Dubloons: 3
Pieces of Eight: 7
Punishment: Cat o' Nine Tails
Message 1966 - Posted: 16 Jun 2005 | 18:26:39 UTC
Last modified: 16 Jun 2005 | 19:00:25 UTC

I wonder how he managed to run the unit for over 53,000 seconds. If you check his other WUs, there's one he did in 109,000+ seconds.

Then there's the question of quota. What's the quota per CPU here? 100 per day?
That person has had 223 so far for today.

I think I got 2 last night. Might be wise to set stricter rules. ;)
____________
Jord.

The BOINC FAQ Service.

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 1967 - Posted: 16 Jun 2005 | 18:46:24 UTC - in response to Message 1964.

>
With the apps based on 1 quorum, the result could be estimated.

Some other applications came with the quorum of 2, and there I assumed, 2
possible estimates a+b/2 or only taking a, where a is the lower returned result. I let manage some results, based on random, to "get" a final time of 14 - 40 hours per hello/mello or how all they are called.

But no, the system finaly took and granted the upper result. Why, I cant write.


Right now having problems to show an example, I'm going/went trough the work all the clients returned yesterday,( many comutation error :-( ) but could not find anymore.

This could have the origine, that I merged all what I could, to blur the traps..or a server based script was executed to reinitialize the result -grating based on the lower ones or to build a new "canonical" result.

It helpes much to get more work, while the last rpc times are set in a (user friendly) deep value. Would this number increased, the individual will "rant" but the work could offered to more people.

-->GT the min rpc times, between 2 requests.

Others are outside of boinc, more relating to network technologie

More work=more "points"..

Forbit the framework, but it's part of the design of boinc..

If I'm not confusing the movie, in red hunt october they had to destroy the
"remote console" by hand to regain the controll over the fireing system

>Do you agree that this is enough to prevent this little trick?
I dont know it for sure. It looks like only above 2 quorums, the "security" will prevent the game of numbers.

What are you doing when a client, let's say running with 2000 MHz is getting work (from any kind of project) and the cpu speed is reduced to let's say 500 MHz? even quorum based, the result could be "different".



Very legal way to optimize the returning "points" is simple to open the show grafic..

This 4.45 Host has some work, over rpc I suspended all work on WU level and an other rpc was enabling the grafic

(working on a "own" scheduling the scheduler where it's possible to run the 4.4x in time frames, as long as there isn't a "problem with deadlines)

A far bid more Credits" can be collected.




>What else can we learn from it? II

This project is much more important to keep it alife to test, evolute and optimize the the client-server nad the server part.

with or without "public work"

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 1968 - Posted: 16 Jun 2005 | 18:54:04 UTC - in response to Message 1966.
Last modified: 16 Jun 2005 | 18:57:41 UTC

> I wonder how he managed to run the unit for over 53,000 seconds. If you check
> his other WUs, there's <a>

The work was done on regular way somehow in 120-150 sec. others took 15 minutes

The assistance came later, but before the work was uploaded.

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1972 - Posted: 16 Jun 2005 | 20:17:46 UTC - in response to Message 1968.

> > I wonder how he managed to run the unit for over 53,000 seconds. If you
> > check his other WUs, there's ...
>
> The work was done on regular way somehow in 120-150 sec. others took 15
> minutes
>
> The assistance came later, but before the work was uploaded.

Basically, he hacked the reply to change the CPU time after the WU finished but before it was uploaded.

For a quorum of 1, you get all you request, if the WU is valid. We've been using the 'trivial' validator, so any WU which completes succesfuly and does a little time is valid.

For a quorum of 2 you get some combination of the two requests. It should be the lower of the two, as described here, but it's possible to set up a project to use a different policy. It may be that we are configured to take an average rather than the lower of the two. I'll have to check, as I've not focused on this end of the project.

The redundancy of setting the quorum > 1 is supposed to protect against this, but it won't protect against a conspiracy. If the quorum was 2 but Mr. peg_leg was conspiring with Mr. one_eye they both could have still received higher credit because they both shared so many WU's. But that is in part due to the small nature of our project.

A required quorum of 3 gives further protection against conspiracy, but would be harder on a test project like this. I trust we will not have to institute such a policy.

It is also possible to put more careful checks into the validator to scrutinize WU's more closely after they are uploaded. So there are ways to defend against this for a real 'production' project.

Mr. peg_leg should be commended for his keen eye in all this.

The question is, should Mr. one_eye receive praise or punishment for his actions? What do the crew think? ;-)
____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Ageless
Chief Petty Officer
Volunteer tester
Avatar
Send message
Joined: 20 Jul 04
Netherlands
Machinae Supremacy
Credit: 1,295.9
RAC: 0.00
Joined: Jul 20, 2004
Verified: Jul 9, 2011
Dubloons: 3
Pieces of Eight: 7
Punishment: Cat o' Nine Tails
Message 1973 - Posted: 16 Jun 2005 | 20:21:48 UTC - in response to Message 1972.

> > The question is, should Mr. one_eye receive praise or punishment for his
> actions? What do the crew think? ;-)

I think he should get a T-shirt, then be slashed with the cat o' nine tails. Since he didn't say anything about it himself.

Peg_leg should get a T-shirt and some gold bullion. If you know where you hid some of yours. ;)
____________
Jord.

The BOINC FAQ Service.

STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 1975 - Posted: 16 Jun 2005 | 21:07:48 UTC

Peg_leg should get a T-shirt and some gold bullion
=========

Send the T-Shirt & the hell with the gold bullion, T-Shirts are much more valuable ... hehe

I never meant no harm to one eye and I still consider him a friend even though he quit our team. As a Team Founder I just felt it was just something I had to report, if the Team Founders don't keep a check on their own Team Members then there's not much hope for the Projects when it comes to cheating IMO ...

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1980 - Posted: 17 Jun 2005 | 1:16:45 UTC - in response to Message 1967.

> Some other applications came with the quorum of 2, and there I assumed, 2
> possible estimates a+b/2 or only taking a, where a is the lower returned
> result. I let manage some results, based on random, to "get" a final time of
> 14 - 40 hours per hello/mello or how all they are called.

It seems the quorum on hello, yello, and sextant were all set at 1, and that explains how this was possible. Nothing more is needed.

>
> But no, the system finaly took and granted the upper result. Why, I cant
> write.

I can offer a theory here. Your result and another were both returned, but not yet processed. Yours was validated first, and with a quorum of 1 it became the "canonical" result. The other then should have received the same credit.

Or are there counterexamples?

____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 1983 - Posted: 17 Jun 2005 | 1:46:08 UTC - in response to Message 1980.
Last modified: 17 Jun 2005 | 1:47:21 UTC

captn,

finaly found an wonderful example, please look here


2 results and a quorum of 1?

might your are right, the result was returned quicker as the regular one

now going to bed looking to the reply tomorrow.

Stupid I am, why didn't I asked for a full week/month/year cputime ;)

The "boinc time" calculation started somewhere at 1.1.1970.
Wondering what would happen, when a time is reported so high, that it would fall below 1970...

more mysteries?? take a short look into the q+a of LHC, where a brave mate got a time warp. This host has due bios reset a brocken time epoche

enjoy!

[url=http://lhcathome.cern.ch/forum_thread.php?id=1310
] http://lhcathome.cern.ch/forum_thread.php?id=1310[/url]

Profile Contact
Volunteer tester
Avatar
Send message
Joined: 29 Aug 04
Canada
BOINC Synergy
Credit: 26,644.6
RAC: 0.00
Joined: Aug 29, 2004
Verified: Nov 10, 2011
Dubloons: 3
Pieces of Eight: 3
Punishment: Misfit
Message 1984 - Posted: 17 Jun 2005 | 2:29:47 UTC

Most exciting pirate thread to date!
Big respect to all here.
To peg leg for his keen eye in all this and courage to report knowing full well of possible damage to personal relationship even beyond this one project. And for declining the gold bullion you show you truly are not a ‘rich man’

Respect for the Capt. for making all the right moves.

Praise for one eye, not punishment.
If he comes clean and helps us fix the problem, then there is no problem.

Real question remains. why, dear boinc do you allow this technically
I have had fleeting thoughts of doing this myself, so i’m sure have many others.
This should be front page news. Brought to you by the team of peg leg and one eye. Sponsored by Pirates!

____________

Click and enter your name for your BOINC Statistics

STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 1986 - Posted: 17 Jun 2005 | 9:36:41 UTC
Last modified: 17 Jun 2005 | 9:38:22 UTC

To peg leg for his keen eye in all this and courage to report knowing full well of possible damage to personal relationship even beyond this one project.
==========

Yes, it was a very difficult decision for me to make as I do or did have personal contact with one eye. I actually waited about 24 hr's to see what or where this was leading to before finally reporting the WU's. In the end though I feel I had to do what was right no matter what the cost was to me ... :/

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1989 - Posted: 17 Jun 2005 | 11:27:40 UTC - in response to Message 1983.
Last modified: 17 Jun 2005 | 11:37:01 UTC

> captn,
>
> finaly found an wonderful example, please look here
>
> 2 results and a quorum of 1?

Yes, this shows that BOINC is acting as expected. The first result (yours, with the "modified" claimed CPU) is taken as the canonical result, but the second result is already "out there", so when it returns it is also granted credit.

After a little reflection, I think the proper paraphrase from Red October is (please imagine this being said by Sean Connery):

"The Capt'n is now resetting all his workunits to a quorum of two. He will not make the same mistake twice."

I believe that is all that is required to prevent this little mischief, but let's see, If you like, Mr. one_eye, please try your trick on the latest starboard workunits to see what happens. I expect we will then be able to link to a WU which shows the attempt and how it fails. If Mr. peg_leg ends up running the other Result from the same WU then he will keep you honest again. ;-)


____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 1990 - Posted: 17 Jun 2005 | 11:28:42 UTC - in response to Message 1984.


> This should be front page news. Brought to you by the team of peg leg and one
> eye. Sponsored by Pirates!

Good idea. Consider it done.


____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Mchl
Volunteer tester
Avatar
Send message
Joined: 23 Sep 04
Poland
BOINC@Poland
Credit: 1,549.7
RAC: 0.00
Joined: Sep 23, 2004
Verified: May 1, 2009
Punishment: Mess Duty
Message 1995 - Posted: 17 Jun 2005 | 13:34:12 UTC - in response to Message 1967.

> >What else can we learn from it? II
>
> This project is much more important to keep it alife to test, evolute and
> optimize the the client-server nad the server part.
>
> with or without "public work"
>
I've been always repeating that!
Aye!



____________
Me Pirrrate name be Mad William Flint an' don't ye dare t' call me with any other name me lad!
Arrr!

Profile Femue
Send message
Joined: 22 Mar 05
Germany
Ruhr-SETIs
Credit: 330.5
RAC: 0.00
Joined: Mar 22, 2005
Verified: NEVER
Dubloons: 3
Message 2035 - Posted: 19 Jun 2005 | 8:09:38 UTC - in response to Message 1989.
Last modified: 19 Jun 2005 | 8:15:42 UTC

> "The Capt'n is now resetting all his workunits to a quorum of two. He will
> not make the same mistake twice."

Captn,

i have a few results pending since last night, and they seem that they won't get credited: http://pirates.vassar.edu/workunit.php?wuid=196891 (see also wuid 196943, 196985, 196986, 196987, 197107).

Shouldn't there be a quorum of two? Will they ever leave the pending-status?

Aye
Femue

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 2036 - Posted: 19 Jun 2005 | 13:12:55 UTC - in response to Message 2035.

> > "The Capt'n is now resetting all his workunits to a quorum of two. He
> will not make the same mistake twice."
>
> Captn,
>
> i have a few results pending since last night, and they seem that they won't
> get credited....

The Capt'n won't make the same mistake twice, he'll make a different mistake. :-)

I set the quorum to 2, but the number of target results to 1. So one WU was sent out, and now it's waiting for 2 to come back to fill the quorum. Interesting situation.

I belive I can repair this in the database and get companions for all those lonely WU's. Standby...


____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

STE\/E
Send message
Joined: 13 Aug 04
United States
SETI.USA
Credit: 37,071.9
RAC: 0.00
Joined: Aug 13, 2004
Verified: Oct 4, 2011
Punishment: Mess Duty
Message 2037 - Posted: 19 Jun 2005 | 14:23:41 UTC

I have a couple too Capt'n ... :/

Result ID Claimed credit
448086 0.23
448625 0.18
449394 0.23
449544 0.19
449973 0.24
450136 0.18
450140 0.18
450487 0.37
450612 0.36
450619 0.36
450694 0.19
453355 2.84
453363 2.87
453371 2.88
453377 2.90
453379 2.89
453383 2.88
453406 2.89
453410 2.87
453416 2.89
453417 2.88
453418 2.90
453426 2.88
453441 2.89
453445 2.88
453446 2.84
453450 2.85
453453 2.82
453455 2.85
453456 2.88
453463 2.84
453473 2.84
453492 2.78
453498 2.63
453499 3.00
453503 3.00
453507 2.98
453508 2.84
453509 2.98
453511 2.87
453522 2.46
453523 2.94
453524 2.56
453526 2.55
453528 3.00
453529 2.83
453530 2.94
453531 2.56
453534 3.00
453536 2.94
453537 2.56
453538 2.67
453542 2.56
453543 2.84
453545 2.94
453546 2.55
453547 2.85
453549 2.84
453550 2.65
453551 2.93
453553 2.84
453555 2.56
453558 2.84
453559 2.55
453560 2.84
453563 2.56
453566 2.85
453567 2.55
453570 2.85
453571 2.54
453572 2.99
453573 2.55
453574 2.85
453575 2.55
453576 2.93
453577 2.55
453579 2.55
453580 2.85
453584 2.85
453586 2.83
453587 2.54
453588 2.82
453589 2.53
453590 2.80
453591 2.52
453592 2.80
453593 2.52
453594 2.80
453595 2.53
453596 2.98
453599 2.52
453600 2.98
453601 2.53
453602 2.80
453605 2.53
453606 2.98
453607 2.53
453608 2.80
453609 2.54
453610 3.00
453612 2.79
453613 2.55
453614 2.80
453616 2.80
453617 2.82
453618 2.55
453620 2.56
453622 2.55
453623 2.81
453624 2.84
453625 3.00
453626 2.84
453628 2.82
453629 2.56
453631 2.56
453633 2.55
453636 3.00
453637 2.55
453641 2.86
453642 2.99
453643 2.55
453644 2.97
453645 2.82
453646 2.82
453650 2.55
453658 2.56
453659 2.95
453660 2.56
453661 2.98
453662 2.81
453664 2.86
453667 2.93
453668 2.86
453672 2.56
453673 2.55
453675 2.82
453676 2.87
453677 2.85
453681 2.84
453682 2.95
453683 2.84
453685 2.56
453688 2.92
453691 2.55
453692 2.94
453694 2.98
453695 2.55
453696 2.70
453697 2.79
453698 2.95
453699 2.88
453700 2.74
453702 2.86
453704 2.86
453705 2.83
453706 2.84
453707 2.95
453708 2.56
453709 2.99
453710 2.93
453712 2.90
453713 2.86
453715 2.50
453722 2.92
453723 2.83
453724 2.98
453725 2.55
453726 2.97
453727 2.86
453728 2.76
453729 2.85
453738 2.60
Pending credit: 450.04

Profile Neil Woodvine
Avatar
Send message
Joined: 14 Mar 05
United Kingdom
BOINC Synergy
Credit: 2,185.6
RAC: 0.00
Joined: Mar 14, 2005
Verified: NEVER
Dubloons: 1
Punishment: Keel Haul
Message 2043 - Posted: 19 Jun 2005 | 21:22:26 UTC

Aye one or two here aswell

452501 0.40
452504 0.40
452828 1.79
452829 1.75
452830 1.75
452831 1.75
452832 1.77
452838 1.77
452839 1.77
452841 1.76
452845 1.75
452852 1.76
452854 1.73
452856 1.75
452857 1.81
452867 1.77
452870 1.78
452871 1.79
452873 1.74
452875 1.79
452877 1.73
452881 1.71
452883 1.72
452884 1.71
452889 1.70
452890 1.69
452893 1.69
452895 1.69
452897 1.69
452900 1.69
452902 1.69
452903 1.69
452905 1.69
452906 1.69
452907 1.70
452913 1.70
452914 1.69
452915 1.69
452920 1.70
452921 1.69
452922 1.68
452924 1.68
452925 1.68
452928 1.68
452931 1.68
452934 1.67
452943 1.67
452954 1.67
452958 1.68
452959 1.68
452961 1.68
452963 1.68
452964 1.68
452965 1.68
452967 1.69
452969 1.68
452970 1.69
452972 1.68
452974 1.67
452986 1.67
452987 1.67
452989 1.67
452991 1.67
452999 1.68
453000 1.67
453002 1.67
453003 1.67
453004 1.68
453005 1.67
453006 1.68
453011 1.68
453013 1.68
453014 1.68
453016 1.68
453018 1.68
453019 1.69
453020 1.68
453021 1.68
453022 1.68
453026 1.69
453030 1.68
453031 1.69
453032 1.68
453035 1.68
453036 1.68
453037 1.68
453039 1.68
453041 1.68
453042 1.69
453097 1.68
453099 1.69
453100 1.69
453102 1.69
453103 1.69
453104 1.69
453105 1.69
453138 1.75

____________

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 2044 - Posted: 20 Jun 2005 | 14:03:09 UTC - in response to Message 2036.

> I belive I can repair this in the database and get companions for all those
> lonely WU's. Standby...

The first thing I tried didn't seem to work, and I don't think it is related to the DB corruption we just went through. Will try some other things in a bit...

____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Ananas
Send message
Joined: 23 Mar 05
Germany
Nordlichter
Credit: 378.4
RAC: 0.00
Joined: Mar 23, 2005
Verified: Mar 28, 2009
Pieces of Eight: 5
Punishment: Mess Duty
Message 2048 - Posted: 20 Jun 2005 | 16:30:08 UTC



Not hard to make a result claim a lot more than it should have.

Of course it did not take my PC that long :


CPU time 40802.6875

worker: CPU time: 40802.140625 seconds, Wall clock time: 40867.546875 seconds


but there is no checksum that makes the server reject edited client_state.xml entries.
____________

Ben Christy
Send message
Joined: 13 Apr 05
United States
Team ToonTown
Credit: 13.3
RAC: 0.00
Joined: Apr 13, 2005
Verified: NEVER
Punishment: Canon Fodder
Message 2050 - Posted: 20 Jun 2005 | 17:08:48 UTC
Last modified: 20 Jun 2005 | 17:10:38 UTC

Wormholio: shouldn't the program that BOINC sites use be changed too. a simple check to prevent target less than quorum?
____________

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 2051 - Posted: 20 Jun 2005 | 17:25:11 UTC - in response to Message 2050.

> Wormholio: shouldn't the program that BOINC sites use be changed too. a simple
> check to prevent target less than quorum?

Ideally, yes.

____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Ben Christy
Send message
Joined: 13 Apr 05
United States
Team ToonTown
Credit: 13.3
RAC: 0.00
Joined: Apr 13, 2005
Verified: NEVER
Punishment: Canon Fodder
Message 2052 - Posted: 20 Jun 2005 | 17:34:07 UTC - in response to Message 1980.
Last modified: 20 Jun 2005 | 17:58:26 UTC

......
> >
> > But no, the system finaly took and granted the upper result. Why, I cant
> > write.
>
> I can offer a theory here. Your result and another were both returned, but
> not yet processed. Yours was validated first, and with a quorum of 1 it
> became the "canonical" result. The other then should have received the same
> credit.
>
> Or are there counterexamples?
>
-------------

this is a good reason to always report unexpected results... especially unusual credit recieved even if it appears to be a gift. you would complain if you requested 2 credits but only recieved .02 so you should also complain when your 2 credit request results in 200. in this case it is an indication that a cheater has shanghied you for the ride. next time someone else will get the top deck if you don't make him walk the plank now.
____________

Ben Christy
Send message
Joined: 13 Apr 05
United States
Team ToonTown
Credit: 13.3
RAC: 0.00
Joined: Apr 13, 2005
Verified: NEVER
Punishment: Canon Fodder
Message 2054 - Posted: 20 Jun 2005 | 17:50:10 UTC - in response to Message 2048.

>
>
> Not hard to make a <a> href="http://pirates.vassar.edu/result.php?resultid=455495">result[/url] claim a
> lot more than it should have.
>
> Of course it did not take my PC that long :
>
>
> CPU time 40802.6875
>
> worker: CPU time: 40802.140625 seconds, Wall clock time: 40867.546875 seconds
>
>
> but there is no checksum that makes the server reject edited client_state.xml
> entries.
>

How hard would it be to impliment and how secure would it be if public keys were used to encode results before they were put in the queue for transmition back to host site?(which would decode with its private key)

this would put another layer of protection between programs and users since they would need to hack the program instead of just the result file.

other than time to program the implimentation I can see no down side
____________

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 2057 - Posted: 20 Jun 2005 | 18:01:57 UTC - in response to Message 2051.

...especially unusual credit recieved even if it appears ..

there are many reasons to have "unespected" values.

For example when a project offers work like at LHC, based on simulation.

Even with the same expected cpu time, the finaly used CPU Time
can vary much from several seconds, several minutes of even using the full expected cpu time.

So technically this will be unsuccessful to to. Even a history (sever side)
can't offer the "security" desired.

with a quorum of 2(3) the excess can be put down/narrowed, but not the cheat itself..

(time = basis for "credits")

As mentioned above, lookto that situation:
1 host receives work and then the cpu speed is dropped by half. Whats the sitution now?


> > Wormholio: shouldn't the program that BOINC sites use be changed too. a
> simple
> > check to prevent target less than quorum?

The only way I see to prevent for "sure", it's to put an additional tag into the client_state, with the checksum of the cputime, md5 based of what ever.

At the time when the workunit gets the state of "uploading" or "rRep"

Now it can be talked about which part should do this,
the core client or the application?

Profile Wormholio
Captain
Avatar
Send message
Joined: 6 Jun 04
United States
Away
Credit: 4,009.8
RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008
Dubloons: 3
Pieces of Eight: 10
Punishment: Aztec curse
Message 2058 - Posted: 20 Jun 2005 | 18:06:23 UTC - in response to Message 2057.

> Now it can be talked about which part should do this,
> the core client or the application?

It would be simple to have the scheduler running on the project site check that the number of targets requested is at least the quorum, and decline to put out thework and log an error if it is not.

____________
-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Profile Neil Woodvine
Avatar
Send message
Joined: 14 Mar 05
United Kingdom
BOINC Synergy
Credit: 2,185.6
RAC: 0.00
Joined: Mar 14, 2005
Verified: NEVER
Dubloons: 1
Punishment: Keel Haul
Message 2061 - Posted: 20 Jun 2005 | 18:19:43 UTC

I think I may have stumbled across one of the ways one eye had boosted his credit, I saw something similar in his explanations on the board.

The other day I finally figured out how some of the crunchers are able to get so many workunits to crunch ( where I normally only get a few , but this maybe to do with the new scheduler on the client , not really sure ) which is to click update and then rest my stapler on the enter button on the keypad and walkaway and have a cup of tea.

Now in my excitement I got a little carried away downloading the sextants and overloaded my comp with units ( won't do that again, sorry =/ ) when I realised that they were going to come in late I tried to pause the already late units and make the client work on the work units further down the list which had a chance of finishing on time ( coincidentally the same time you took the schedulers offline yesterday ).

After I paused the first ten or so units instead of going on to the next unit in the queue it went back to the first ( which may have something to do with the panic/deadline mode in the new scheduler ) and no matter how much I fiddled with it , it wouldn’t work on the work units that would make it on time.

So I unpaused them all and figured I'd at least crunch them even if I didn't get credit as they would be late. When I came back an hour later I noticed that the work units were taking 12 minutes instead of the unsual 8 and incrementing in the amount slightly as it moved to the next work unit, all of which were claiming higher than usual credit. When I opened task manager it showed multiple instances of the sextant running ( which is what reminded me of the picture in one eye's post ). I believe if I had kept pausing unspausing the units to make more instances I would have been able to boost the requested scores of all the remaining work units.

Well not wanting a public flogging and feeling like I had been caught with my hand in the cookie jar I reset the the project to clear all these work units and condemn them to the depths.

I hope that this information helps and sorry for wasting those work units =/

____________

one eye
Send message
Joined: 19 Sep 04
Switzerland
Credit: 40,405.3
RAC: 0.00
Joined: Sep 19, 2004
Verified: NEVER
Punishment: Walk Plank
Message 2062 - Posted: 20 Jun 2005 | 18:22:14 UTC - in response to Message 2058.

> It would be simple to have the scheduler running on the project site check
> that the number of targets requested is at least the quorum, and decline to
> put out thework and log an error if it is not.
>
That would work also but my I put this?:


That's a nomal view of a regular predictor Work Unit (quorum=3).

The "points" returned are realy close, even the final used cpu time differs
form the minimum/maximum of over 100%

The implemented "rules" for taxing it as an error or outside "normal" would be not far or at least not effective as it should.

I believe, a scheduler based verifying and "filtering" can't be done for
obtaining a high leveled secured validation. But could be wrong.

Profile Neil Woodvine
Avatar
Send message
Joined: 14 Mar 05
United Kingdom
BOINC Synergy
Credit: 2,185.6
RAC: 0.00
Joined: Mar 14, 2005
Verified: NEVER
Dubloons: 1
Punishment: Keel Haul
Message 2073 - Posted: 20 Jun 2005 | 21:45:40 UTC

Well unfortunatley i've been able to duplicate the results from the other night
according to task manager i have currently 34 sextant processes and 4 starboards but i'm not seeing the abnormal credit i had yesterday , the only slight difference i've seen is a 1.9 credit for a sextant as opposed to my normal 1.7 =/

i'll keep fiddling though

@one eye , the sooner these bugs get out in the open ( in the hands of a dev ) the sooner they can get fixed i'd hate to see what would happen if someone who isn't as "honest" as you would do with the knowledge.

____________

Profile Kajunfisher
Avatar
Send message
Joined: 3 Apr 05
United States
The Final Front Ear
Credit: 53.2
RAC: 0.00
Joined: Apr 3, 2005
Verified: NEVER
Dubloons: 1
Punishment: Cat o' Nine Tails
Message 2092 - Posted: 21 Jun 2005 | 20:49:56 UTC

I have 2 wu's where I am the only one in the quorum, credit still pending:

197081

197006

I have 1 wu where there is another host that errored out, credit still pending:

196994
____________

Post to thread

Message boards : Number Crunching : Treachery amongst us?

Home Help Status Forums Glossary Account


Return to Pirates@Home main page


Copyright © 2013 Capt. Jack Sparrow