I've started getting the quota error, but i am still able to download WU's every so often, picked up 3 AP in the last hour or so.

Don't forget: the quota is (or should be) application dependent.

Gruß,
Gundolf

It isn't, and to do that there needs to be a DCF/app which we also don't have.

What does DCF have to do with the quota, I don't understand

It wasn't, in traditional BOINC, and that led to absurdities like the Christmas where the download pipe was maxxed out for 10 days or more by a comparatively small number of hosts trashing Astropulse tasks. That should have been nipped in the bud by quota, but it wasn't - because quota was shared, and reset, by successful MB tasks.

'Quota by application' would be a "Good Thing", in the opinion of most people who've studied the problem: and for similar reasons 'DCF by application' would also be a "Good Thing" (save me writing all these posts about flop counts, for starters!).

Both of these "Good Things" were being trialled at Beta, and it seems as if a cut-down version of them (or possibly just the quota part) have been brought across to the Main project. Probably it's the cutting down which has gone wrong.

Seems it should have STAYED in Beta until they knew it would work. Seems the PROJECT has become Beta. Lots of work but....this is absurd.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

It wasn't, in traditional BOINC, and that led to absurdities like the Christmas where the download pipe was maxxed out for 10 days or more by a comparatively small number of hosts trashing Astropulse tasks. That should have been nipped in the bud by quota, but it wasn't - because quota was shared, and reset, by successful MB tasks.

'Quota by application' would be a "Good Thing", in the opinion of most people who've studied the problem: and for similar reasons 'DCF by application' would also be a "Good Thing" (save me writing all these posts about flop counts, for starters!).

Both of these "Good Things" were being trialled at Beta, and it seems as if a cut-down version of them (or possibly just the quota part) have been brought across to the Main project. Probably it's the cutting down which has gone wrong.

What I don't understand, Richard, is why the $#^@&%#^@ Boinc code was not withdrawn from Main as soon as these problems surfaced until it was fixed.

That, to me, is inexcusable.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

I guess what amazes me the most is that they know what code they put in that caused this and instead of taking it back out until they fix it, they leave it in and cause all this rabble....Maybe they enjoy rocking the boat from time to time?

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I've started getting the quota error, but i am still able to download WU's every so often, picked up 3 AP in the last hour or so.

Don't forget: the quota is (or should be) application dependent.

Gruß,
Gundolf

It isn't, and to do that there needs to be a DCF/app which we also don't have.

What does DCF have to do with the quota, I don't understand

It wasn't, in traditional BOINC, and that led to absurdities like the Christmas where the download pipe was maxxed out for 10 days or more by a comparatively small number of hosts trashing Astropulse tasks. That should have been nipped in the bud by quota, but it wasn't - because quota was shared, and reset, by successful MB tasks.

'Quota by application' would be a "Good Thing", in the opinion of most people who've studied the problem: and for similar reasons 'DCF by application' would also be a "Good Thing" (save me writing all these posts about flop counts, for starters!).

Both of these "Good Things" were being trialled at Beta, and it seems as if a cut-down version of them (or possibly just the quota part) have been brought across to the Main project. Probably it's the cutting down which has gone wrong.

Seems it should have STAYED in Beta until they knew it would work. Seems the PROJECT has become Beta. Lots of work but....this is absurd.

To do that, they would have had to ask NVidia not to sell Fermi cards into the open market until the Beta testing was finished. And I've written earlier today about the lack of beta testers skilled in observing and writing accurate reports. And of course, with no cards on the open market, who would do the beta testing?

Sorry, capitalism doesn't work that way. With centralised state control of production and supply, just maybe.

Well once I run out of work I will try an idea I have and if it works I will let others know....few more hours.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

To do that, they would have had to ask NVidia not to sell Fermi cards into the open market until the Beta testing was finished. And I've written earlier today about the lack of beta testers skilled in observing and writing accurate reports. And of course, with no cards on the open market, who would do the beta testing?

Sorry, capitalism doesn't work that way. With centralised state control of production and supply, just maybe.

So, it's OK for them to continue whacking thousands of GPU users so they can test for the tiny fraction of a percent that have the new Fermi cards?

My position stands.....
I can understand the need for testing things. And if that absolutely cannot be accomplished in Beta, I can accept the need to try it on Main. But when you test something and find out it is not working, you shut it down and repair it....you don't just let it continue to bang away broken, eh?

The code should have been yanked for review the minute it was known that it was adversely affecting thousands of users.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

I've started getting the quota error, but i am still able to download WU's every so often, picked up 3 AP in the last hour or so.

Don't forget: the quota is (or should be) application dependent.

Gruß,
Gundolf

It isn't, and to do that there needs to be a DCF/app which we also don't have.

What does DCF have to do with the quota, I don't understand

It wasn't, in traditional BOINC, and that led to absurdities like the Christmas where the download pipe was maxxed out for 10 days or more by a comparatively small number of hosts trashing Astropulse tasks. That should have been nipped in the bud by quota, but it wasn't - because quota was shared, and reset, by successful MB tasks.

'Quota by application' would be a "Good Thing", in the opinion of most people who've studied the problem: and for similar reasons 'DCF by application' would also be a "Good Thing" (save me writing all these posts about flop counts, for starters!).

Both of these "Good Things" were being trialled at Beta, and it seems as if a cut-down version of them (or possibly just the quota part) have been brought across to the Main project. Probably it's the cutting down which has gone wrong.

Seems it should have STAYED in Beta until they knew it would work. Seems the PROJECT has become Beta. Lots of work but....this is absurd.

To do that, they would have had to ask NVidia not to sell Fermi cards into the open market until the Beta testing was finished. And I've written earlier today about the lack of beta testers skilled in observing and writing accurate reports. And of course, with no cards on the open market, who would do the beta testing?

Sorry, capitalism doesn't work that way. With centralised state control of production and supply, just maybe.

Berkely pays so crappy that not one of these project Scientists can afford 1 Fermi card? Why don't they pitch in? I am sorry but I don't believe you.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

To do that, they would have had to ask NVidia not to sell Fermi cards into the open market until the Beta testing was finished. And I've written earlier today about the lack of beta testers skilled in observing and writing accurate reports. And of course, with no cards on the open market, who would do the beta testing?

Sorry, capitalism doesn't work that way. With centralised state control of production and supply, just maybe.

So, it's OK for them to continue whacking thousands of GPU users so they can test for the tiny fraction of a percent that have the new Fermi cards?

My position stands.....
I can understand the need for testing things. And if that absolutely cannot be accomplished in Beta, I can accept the need to try it on Main. But when you test something and find out it is not working, you shut it down and repair it....you don't just let it continue to bang away broken, eh?

The code should have been yanked for review the minute it was known that it was adversely affecting thousands of users.

No, it's the other way round. It's the old code that was erroring out when the users exercised freedom of choice to install new Fermi cards.

The only thing that could have been 'yanked' was people's shiny new (expensive) toys.

And you know how much people would have enjoyed that....

A statistic to consider is that there are fewer than 4000 hosts with RAC above 2600. IOW, less than 1.5% of the ~279000 active hosts should be affected by a fixed 100 quota daily even if the supplied work were almost all VHAR shorties.

I've been trying to trace the failure to allow more work on the next day in the BOINC source code, on the assumption that most of what's being used here is either the older stable server code or the CreditNew version from trunk. Very tentatively I think earlier speculation in this thread is probably correct that it only affects those running anonymous platform. Darwin's experiment of switching to stock to get work and then reoptimizing seems to confirm. If anyone has controverting evidence please note it here.

                                                                  Joe

I just got 162 units and only 2 are Vlar...got it figured out.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I made detach/attach, I lost a lot of results, because current the UL server is offline, so maybe I lost Cr. 35,000+ (~ 2 days) (how I can get the Cr.?) and got:

1x CPU MB
1x GPU 6.09_cuda23
130x GPU 6.08_cuda

WUs incl. the apps.

After a shot time BOINC got:

Message from server: No work sent
Message from server: No work is available for Astropulse v5
Message from server: Fermi-class GPU needed
Message from server: (reached daily quota of 1 tasks)

After app_info.xml file BOINC got:

Message from server: No work sent
Message from server: No work is available for Astropulse v5
Message from server: Your app_info.xml file doesn't have a usable version of SETI@home Enhanced.
Message from server: (reached daily quota of 100 tasks)


The server are confused..

I guess it's better to go to backup projects until it's fixed.
So I'll not lose more wattage for nothing.
The best would be to go back to the earlier quota.

---

I made detach/attach, I lost a lot of results, because current the UL server is offline, so maybe I lost Cr. 35,000+ (~ 2 days) (how I can get the Cr.?) and got:

1x CPU MB
1x GPU 6.09_cuda23
130x GPU 6.08_cuda

WUs incl. the apps.

After a shot time BOINC got:

Message from server: No work sent
Message from server: No work is available for Astropulse v5
Message from server: Fermi-class GPU needed
Message from server: (reached daily quota of 1 tasks)

After app_info.xml file BOINC got:

Message from server: No work sent
Message from server: No work is available for Astropulse v5
Message from server: Your app_info.xml file doesn't have a usable version of SETI@home Enhanced.
Message from server: (reached daily quota of 100 tasks)


The server are confused..

I guess it's better to go to backup projects until it's fixed.
So I'll not lose more wattage for nothing.
The best would be to go back to the earlier quota.

I didn't do any detaching or resets wouldn't want to lose any work.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I just got 162 units and only 2 are Vlar...got it figured out.

What did you? Share your experiences! :-)

---

I didn't do any detaching or resets wouldn't want to lose any work.

I do everything for my loved project! :o)

I did it and posted it, maybe it help to solve the problem.

---

I made detach/attach, I lost a lot of results, because current the UL server is offline...

No, you lost the results because you detached - entirely predictable.

Do you remember a few days ago, during the last outage, I talked 'area 51' through the recovery of 2,000 tasks? That was possible because his were merely 'aborted'. You have done the most drastic thing possible: deleting all the data files, the result files, and the application files (including any optimised apps) and any customisations to app_info.xml. I hope you had a backup.

so maybe I lost Cr. 35,000+ (~ 2 days) (how I can get the Cr.?) ...

You don't. The credit goes to people who do the job properly - crunching the work and reporting the results.

It's called testing. We make the mistakes, and report the outcome, so that other people can learn and avoid making the same mistake. Sometimes we even learn something ourselves. Welcome to the club.

A statistic to consider is that there are fewer than 4000 hosts with RAC above 2600. IOW, less than 1.5% of the ~279000 active hosts should be affected by a fixed 100 quota daily even if the supplied work were almost all VHAR shorties.

I just want to point out that you are making an assumption about the value of those 4000 hosts. A better stat would be: what the overall "crunch" from those 4000 are in comparison to those below 2600?

I haven't done the math, but it wouldn't surprise me if there is a greater crunch being completed by the top 20% compared to the rest of the assembled 80% below.

In the end it is about the total crunch on the project. That's what produces the science. Every cruncher is valuable (I've had machines that hardly crossed into 4 digits for instance). The math belies the truth of the matter however. Limiting the top machines to limping along seems counter productive (and certainly a game changer in the longer run - whose is going to drop $5K on another machine just to watch it run at minimal utilization?).

Regards,

---

A statistic to consider is that there are fewer than 4000 hosts with RAC above 2600. IOW, less than 1.5% of the ~279000 active hosts should be affected by a fixed 100 quota daily even if the supplied work were almost all VHAR shorties.

I just want to point out that you are making an assumption about the value of those 4000 hosts. A better stat would be: what the overall "crunch" from those 4000 are in comparison to those below 2600?

I haven't done the math, but it wouldn't surprise me if there is a greater crunch being completed by the top 20% compared to the rest of the assembled 80% below.

In the end it is about the total crunch on the project. That's what produces the science. Every cruncher is valuable (I've had machines that hardly crossed into 4 digits for instance). The math belies the truth of the matter however. Limiting the top machines to limping along seems counter productive (and certainly a game changer in the longer run - whose is going to drop $5K on another machine just to watch it run at minimal utilization?).

Regards,

At this point running at minimal utilization would be a whole lot better than sitting dead in the water, like I am currently. I am hopeful the upload server will be turned on, and maybe things will clear on my end. I won't do a thing until my 29 units get uploaded and reported. If that doesn't get things moveing again, then I'll look at more options.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

I'm DOA as well Steve and have been since Saturday afternoon. The other cruncher is offline as the old nVidia card finally won't stay running long enough for the machine to do much more than boot (sorry wingmen but there's a few days cache slowly going stale).

The main cruncher had been working itself up to a +30,000 RAC slowly (the daily stats were encouraging) but as you can see now my RAC has already declined to 20,000 and will continue to slide I am sure into the middle or lower teens before this is cleared up.

C'est la vie (or, c'est la SETI).

Regards,

---

I made detach/attach, I lost a lot of results, because current the UL server is offline...

No, you lost the results because you detached - entirely predictable.

Do you remember a few days ago, during the last outage, I talked 'area 51' through the recovery of 2,000 tasks? That was possible because his were merely 'aborted'. You have done the most drastic thing possible: deleting all the data files, the result files, and the application files (including any optimised apps) and any customisations to app_info.xml. I hope you had a backup.

so maybe I lost Cr. 35,000+ (~ 2 days) (how I can get the Cr.?) ...

You don't. The credit goes to people who do the job properly - crunching the work and reporting the results.

It's called testing. We make the mistakes, and report the outcome, so that other people can learn and avoid making the same mistake. Sometimes we even learn something ourselves. Welcome to the club.

I knew that I'll lose all my WUs & results & Cr. if I detach/attach. I used this upper phrase, because all people know this.
I know that I need to UL and report first, before I do this. But how if the UL server is not available? And no, I have no backup. I wouldn't know how to do it, I made this in past and didn't worked for me.

So I decided to do this, although I knew what I'll lose everything. Maybe if someone will report the scheduler answers to the SETI@home crew the problem is easier to solve.
It was not the first time that I lose results, I lost much more results on my GPU machine because the scheduler didn't accepted the too big report file. Detach/attach was there also the only possibility to have again well contact to the scheduler. (IIRC, you could add 500,000+ Cr. to my account.)

---

I knew that I'll lose all my WUs & results & Cr. if I detach/attach. I used this upper phrase, because all people know this.
I know that I need to UL and report first, before I do this. But how if the UL server is not available? And no, I have no backup. I wouldn't know how to do it, I made this in past and didn't worked for me.

So I decided to do this, although I knew what I'll lose everything

I'm sorry, but this is plain foolish and is a waste of the project's limited resources. To detach, reset or plain mess up your cache through ignorance - (ie, as a pure mistake) is one thing - and we have all probably done something like this during our contribution to this project, but to do it intentionally is vandalism - pure and simple. Your computer and your electric bill are nobody's business but yours, but your action has squandered the bandwidth and server resources necessary to distribute the tasks to you. Your action is totally indefensible. If the upload server is not available, WAIT. If the download server is not available, WAIT. Are these concepts so difficult to grasp (those are rhetorical questions, no answers are required)?

---