no credit granted, though WU ended with "success"

blaubo
blaubo
Joined: 22 Dec 06
Posts: 10
Credit: 287155
RAC: 0
Topic 192777

Hello,
for the first time as long as I am part of E@H there was no credit granted though the result in the WU (33745679) for my box says success. Same box as usual and no new software installed. Could anybody help me by explaining that, so that I can avoid the same outcome in the future?
Thanks

MattDavis
MattDavis
Joined: 18 Jan 05
Posts: 68
Credit: 10565341
RAC: 0

no credit granted, though WU ended with "success"

Someone might be able to explain this better than me. From what I understand, there is currently a bug where sometimes units crunched by different OSes don't validate with each other.

You crunched the unit with Windows and your partner crunched it with Linux. Because of this bug, the unit was sent out again. Since the new partner was Linux, the two Linux units validated with each other, and you got 0.

From what I understand this is rare.

blaubo
blaubo
Joined: 22 Dec 06
Posts: 10
Credit: 287155
RAC: 0

Thank you, I am relieved that

Thank you, I am relieved that nothing in my setup broke.

Though it's too bad about the lost crunching time for the project.

zombie67 [MM]
Joined: 10 Oct 06
Posts: 121
Credit: 503746728
RAC: 1261920

Why not just enable

Why not just enable Homogenous Redundancy on the server?

Reno, NV Team: SETI.USA

Ananas
Ananas
Joined: 22 Jan 05
Posts: 272
Credit: 2500681
RAC: 0

HR seems to have a flaw, from

HR seems to have a flaw, from my observations, it seems
that the scheduler doesn't give out the next WU that would
be suitable for the requested platform, it rather seems
to wait for a specific platform, when an already "tagged"
result is on top of the stack.

I might be wrong though, it's just an idea I got when I
watched the behaviour of some project schedulers, not
based on examining the code.

Without HR, the chance of tracking down not fully portable
code snippets is higher too. The more platforms you have
with compareable results, the better your code quality can
be.

HR is nothing but a dirty workaround for not so good code.

zombie67 [MM]
Joined: 10 Oct 06
Posts: 121
Credit: 503746728
RAC: 1261920

FWIW, Docking@Home uses HR

Message 67080 in response to message 67079

FWIW, Docking@Home uses HR quite successfully.

Reno, NV Team: SETI.USA

KSMarksPsych
KSMarksPsych
Moderator
Joined: 15 Oct 05
Posts: 2702
Credit: 4090227
RAC: 0

RE: FWIW, Docking@Home uses

Message 67081 in response to message 67080

Quote:
FWIW, Docking@Home uses HR quite successfully.

I believe Docking did a lot of modification to the backend to get it to work, and they are still working on fine tuning it. IIRC, they're HR is pretty fine grained. Don't know if they would need the same here.

I'm not sure how much of their HR code made it back into SVN.

Kathryn :o)

Einstein@Home Moderator

Yin Gang
Yin Gang
Joined: 23 Feb 05
Posts: 52
Credit: 120187750
RAC: 0

I just got two results with

Message 67082 in response to message 67076

I just got two results with no credit granted. As MattDavis said, the other two results of the same wu have been crunched by linux while mine by windows.

http://einsteinathome.org/workunit/33772527
http://einsteinathome.org/workunit/33748310

Quote:

Someone might be able to explain this better than me. From what I understand, there is currently a bug where sometimes units crunched by different OSes don't validate with each other.

You crunched the unit with Windows and your partner crunched it with Linux. Because of this bug, the unit was sent out again. Since the new partner was Linux, the two Linux units validated with each other, and you got 0.

From what I understand this is rare.


Welcome To Team China!

RandyC
RandyC
Joined: 18 Jan 05
Posts: 6626
Credit: 111139797
RAC: 0

RE: Someone might be able

Message 67083 in response to message 67076

Quote:

Someone might be able to explain this better than me. From what I understand, there is currently a bug where sometimes units crunched by different OSes don't validate with each other.

You crunched the unit with Windows and your partner crunched it with Linux. Because of this bug, the unit was sent out again. Since the new partner was Linux, the two Linux units validated with each other, and you got 0.

From what I understand this is rare.

Not so rare when your data file is shared with a couple of MACs and a Linux system. I've had 3 out of 5 WUs go belly up because of this bug. I have another pending that will probably also be odd-man out.

When you spend 165 hrs crunching 5 WUs and only get credit for two of them, it SUCKs big time!!!!

[edit typo]

Seti Classic Final Total: 11446 WU.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.