User w/ Nearly 4,000 Abandoned Tasks on 1 Computer


log in

Advanced search

Message boards : Number crunching : User w/ Nearly 4,000 Abandoned Tasks on 1 Computer

Author Message
Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1394950 - Posted: 25 Jul 2013, 23:31:34 UTC

Was just looking through my own WU database and ran across a wingman that has abandoned every WU we've ever shared (30 of them since early May). In looking at his current computer:

http://setiathome.berkeley.edu/results.php?hostid=7017265

it appears that he simply abandons every WU he receives (and it looks like he's received over 600 today), with a current total of nearly 4,000 in the "Error" bucket! Since his total credit is still at 0, I assume that this has been happening since this computer was added on June 10, 2013. (And his previous computer, which appears to be the same as the current one but with a different ID, has a total credit of 0, too!)

How does this happen, how does he keep on getting so many tasks when he just keeps trashing them, and isn't there some way this be stopped???? I know that the tasks just get resent, but this seems like such a waste of resources for the project to keep sending out all these tasks every day for absolutely no purpose!

Profile ignorance is no excuse
Avatar
Send message
Joined: 4 Oct 00
Posts: 9529
Credit: 44,433,274
RAC: 0
Korea, North
Message 1395192 - Posted: 26 Jul 2013, 16:15:49 UTC - in response to Message 1394950.

It's unlikely that someone is actively abandoning WU's it's more likely his PC/BOINC is not able to run them or something has gone horribly wrong on that PC to basically reject work or fail to start the work.
____________
In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

End terrorism by building a school

rob smithProject donor
Volunteer tester
Send message
Joined: 7 Mar 03
Posts: 8425
Credit: 57,489,740
RAC: 74,760
United Kingdom
Message 1395197 - Posted: 26 Jul 2013, 16:22:07 UTC

I've PM's him to see if that helps...
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Profile Gary CharpentierProject donor
Volunteer tester
Avatar
Send message
Joined: 25 Dec 00
Posts: 12577
Credit: 6,882,843
RAC: 6,618
United States
Message 1395477 - Posted: 27 Jul 2013, 3:43:45 UTC - in response to Message 1395192.

It's unlikely that someone is actively abandoning WU's it's more likely his PC/BOINC is not able to run them or something has gone horribly wrong on that PC to basically reject work or fail to start the work.

With the 100% abandoned, I suspect that they aren't actually downloading to his machine. Or once they get there they aren't being saved to his hard drive. Might be a permissions issue with the BOINC install.

____________

Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1395599 - Posted: 27 Jul 2013, 15:58:27 UTC - in response to Message 1395197.

I've PM's him to see if that helps...

The PM doesn't seem to have had any impact. His machine is still blowing them away at a prodigious clip. In the most recent 24 hour period, it looks like he's received almost 800 tasks. If the machine's been going at that clip since its inception on June 10, that would be over 37,000 abandoned tasks! (And that's not counting the results of the machine's earlier incarnation.)

Isn't there some way for the project to put that machine on a "blocked" list so the scheduler won't keep wasting time and bandwidth?

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 7
United States
Message 1395613 - Posted: 27 Jul 2013, 16:32:01 UTC
Last modified: 27 Jul 2013, 16:38:08 UTC

I've PM's him to see if that helps...

He has zero RAC on all 4 BOINC projects - not likely to visit a website.

The PM doesn't seem to have had any impact. His machine is still blowing them away at a prodigious clip. In the most recent 24 hour period, it looks like he's received almost 800 tasks. If the machine's been going at that clip since its inception on June 10, that would be over 37,000 abandoned tasks! (And that's not counting the results of the machine's earlier incarnation.)

Isn't there some way for the project to put that machine on a "blocked" list so the scheduler won't keep wasting time and bandwidth?


I thought the daily limits would cut off a runaway host - that's supposed to be their purpose. But he's showing for each app:

Number of tasks completed 0
Max tasks per day 33

and for the individual apps with a max daily of 33:
Number of tasks today 61 (v7cpu), 52 cuda32, 97 cuda42, and 43 cuda50.

Why hasn't the max daily gone to zero or one? And why allow more than the max?
The staff needs to look into why the controls are not cutting him off.
The app details:
http://setiathome.berkeley.edu/host_app_versions.php?hostid=7017265
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1397442 - Posted: 1 Aug 2013, 4:47:52 UTC - in response to Message 1395613.

As far as I can tell, that "Max tasks per day" number doesn't mean diddly-squat (or perhaps that's exactly what it means).

I just started another machine running S@H a few hours ago. Naturally, it doesn't have much of a track record so far, only 2 tasks completed and validated, and another 18 with validation pending, all cuda50. I would expect the initial limits to be enforced until the new machine's reliability is proven. That's hardly the case, as my application details show:

SETI@home v7 7.00 windows_intelx86
Number of tasks completed 0
Max tasks per day 33
Number of tasks today 44
Consecutive valid tasks 0
Average turnaround time 0.00 days

SETI@home v7 7.00 windows_intelx86 (cuda32)
Number of tasks completed 0
Max tasks per day 33
Number of tasks today 63
Consecutive valid tasks 0
Average turnaround time 0.00 days

SETI@home v7 7.00 windows_intelx86 (cuda42)
Number of tasks completed 0
Max tasks per day 33
Number of tasks today 5
Consecutive valid tasks 0
Average turnaround time 0.00 days

SETI@home v7 7.00 windows_intelx86 (cuda50)
Number of tasks completed 2
Max tasks per day 35
Number of tasks today 52
Consecutive valid tasks 2
Average processing rate 153.18175576871
Average turnaround time 0.02 days

For every application type, except for cuda42, the "Number of tasks today" already exceeds the "Max tasks per day", in just the first 2 hours that this machine has been active. So, unless the scheduler is taking into account the Application Details of my other hosts (which are doing pretty well), the Max Tasks number appears to be meaningless! It's no wonder there are runaway hosts like this guy's (who appears to have been sent another 900+ tasks for auto-trashing today), probably quite a few of them. Does the S@H staff actually care??

Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1402861 - Posted: 14 Aug 2013, 1:03:51 UTC - in response to Message 1394950.

http://setiathome.berkeley.edu/results.php?hostid=7017265

For a while it looked like there might be good news on this machine. It seemed to have been shut down about 10 days ago and, even though the last 69 tasks it downloaded will likely be in limbo until they eventually time out, at least it wasn't abandoning hundreds of tasks every day.

Unfortunately, the same machine appears to have been reincarnated yesterday under a new ID:

http://setiathome.berkeley.edu/results.php?hostid=7065266

And in the first 24 hours or so, he's already abandoned at least 331 tasks. (I actually got the resend for one of his Astropulse tasks early this morning.) Wonder how long it'll take to get back up to the nearly 4,000 abandoned task level that it seemed to maintain under the old ID?

j tramer
Send message
Joined: 6 Oct 03
Posts: 242
Credit: 5,386,384
RAC: 16
Canada
Message 1403109 - Posted: 14 Aug 2013, 16:09:47 UTC

i cant get work at all....I think I give up on this shit

Juha
Volunteer tester
Send message
Joined: 7 Mar 04
Posts: 180
Credit: 142,443
RAC: 124
Finland
Message 1403111 - Posted: 14 Aug 2013, 16:38:44 UTC - in response to Message 1403109.

i cant get work at all....I think I give up on this shit

Or you could start paying some attention to what the project does.

The project released a new application, Setiathome v7, two and a half months ago and Astropulse v6 over a year ago.

You need to upgrade your applications. Either reset the project to allow the server to send current applications to your host or get the newest Lunatics installer and run it.

rob smithProject donor
Volunteer tester
Send message
Joined: 7 Mar 03
Posts: 8425
Credit: 57,489,740
RAC: 74,760
United Kingdom
Message 1403139 - Posted: 14 Aug 2013, 18:09:09 UTC

Since j tramer is running optimised apps (s)he will have to run the Lunatics installer as resetting will not help.
There is another way - remove the app_info.xml file from the SETI@Home project directory, and that will, at the next contact force the latest stock apps to be download automatically.
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Juha
Volunteer tester
Send message
Joined: 7 Mar 04
Posts: 180
Credit: 142,443
RAC: 124
Finland
Message 1403157 - Posted: 14 Aug 2013, 19:23:52 UTC - in response to Message 1403139.

Since j tramer is running optimised apps (s)he will have to run the Lunatics installer as resetting will not help.

I thought resetting removed app_info.xml but looks like it must be removed manually or removed by detach+attach. Oops.

Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1403305 - Posted: 15 Aug 2013, 3:44:29 UTC

Hijacked! Oh, well....guess this thread wasn't going anywhere, anyway.

I did just notice an interesting WU among this guy's latest abandoned tasks:

http://setiathome.berkeley.edu/workunit.php?wuid=1243921183

It's an old v6 WU that appears to have been validated back on May 14th, but with 0.00 credits for each machine. Then a third machine finally reported it today but got a validate error, followed by a resend that got a computation error, another that got this guy's auto-abandonment, and now it's off to "he who shall not be named" for another opinion. Is there any way that this WU can ever get off the merry-go-round?

Profile Donald L. JohnsonProject donor
Avatar
Send message
Joined: 5 Aug 02
Posts: 6212
Credit: 710,890
RAC: 1,183
United States
Message 1403331 - Posted: 15 Aug 2013, 6:21:13 UTC - in response to Message 1403305.

Hijacked! Oh, well....guess this thread wasn't going anywhere, anyway.

I did just notice an interesting WU among this guy's latest abandoned tasks:

http://setiathome.berkeley.edu/workunit.php?wuid=1243921183

It's an old v6 WU that appears to have been validated back on May 14th, but with 0.00 credits for each machine. Then a third machine finally reported it today but got a validate error, followed by a resend that got a computation error, another that got this guy's auto-abandonment, and now it's off to "he who shall not be named" for another opinion. Is there any way that this WU can ever get off the merry-go-round?

If Mark's results match the _2 (Validate error) results, it should validate and be done. Shame the _1 won't get any credit, though...
____________
Donald
Infernal Optimist / Submariner, retired

Profile James Sotherden
Avatar
Send message
Joined: 16 May 99
Posts: 8835
Credit: 34,921,050
RAC: 59,373
United States
Message 1403334 - Posted: 15 Aug 2013, 6:42:27 UTC - in response to Message 1403331.

Hijacked! Oh, well....guess this thread wasn't going anywhere, anyway.

I did just notice an interesting WU among this guy's latest abandoned tasks:

http://setiathome.berkeley.edu/workunit.php?wuid=1243921183

It's an old v6 WU that appears to have been validated back on May 14th, but with 0.00 credits for each machine. Then a third machine finally reported it today but got a validate error, followed by a resend that got a computation error, another that got this guy's auto-abandonment, and now it's off to "he who shall not be named" for another opinion. Is there any way that this WU can ever get off the merry-go-round?

If Mark's results match the _2 (Validate error) results, it should validate and be done. Shame the _1 won't get any credit, though...

Mark makes sure his riggs run true. If its a good work unit it will validate.
We have all had are share of _5 work units that bite the dust.
____________

Old James

N9JFE David SProject donor
Volunteer tester
Avatar
Send message
Joined: 4 Oct 99
Posts: 11632
Credit: 14,358,570
RAC: 13,450
United States
Message 1403433 - Posted: 15 Aug 2013, 13:47:47 UTC - in response to Message 1403334.

Hijacked! Oh, well....guess this thread wasn't going anywhere, anyway.

I did just notice an interesting WU among this guy's latest abandoned tasks:

http://setiathome.berkeley.edu/workunit.php?wuid=1243921183

It's an old v6 WU that appears to have been validated back on May 14th, but with 0.00 credits for each machine. Then a third machine finally reported it today but got a validate error, followed by a resend that got a computation error, another that got this guy's auto-abandonment, and now it's off to "he who shall not be named" for another opinion. Is there any way that this WU can ever get off the merry-go-round?

If Mark's results match the _2 (Validate error) results, it should validate and be done. Shame the _1 won't get any credit, though...

Mark makes sure his riggs run true. If its a good work unit it will validate.
We have all had are share of _5 work units that bite the dust.

Mark did validate it against the 2nd original user, Charlie1. 1 pulse and 1 gaussian. BillyMc had a CUFFT error in line 62 and gets an invalid.

____________
David
Sitting on my butt while others boldly go,
Waiting for a message from a small furry creature from Alpha Centauri.


Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 289
Credit: 33,961,645
RAC: 90,139
United States
Message 1403473 - Posted: 15 Aug 2013, 15:39:39 UTC - in response to Message 1403331.


If Mark's results match the _2 (Validate error) results, it should validate and be done. Shame the _1 won't get any credit, though...

Looks like the _1 got credit after all. His status actually got changed from "Completed, marked as invalid", with 0.00 credit to "Completed and validated" with full credit. But the _2, which also appears to match, still ends up with the "Validate error". Verrrry interesting!

Message boards : Number crunching : User w/ Nearly 4,000 Abandoned Tasks on 1 Computer

Copyright © 2014 University of California