6.6.2: ignoreing "Wont't get new tasks"

Message boards : Questions and problems : 6.6.2: ignoreing "Wont't get new tasks"
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Joseph Stateson
Volunteer tester
Avatar

Send message
Joined: 27 Jun 08
Posts: 641
United States
Message 22524 - Posted: 20 Jan 2009, 4:27:56 UTC
Last modified: 20 Jan 2009, 4:29:59 UTC

Although I had specified "no new tasks", an update command downloaded 4 new seti tasks after submitting 6 completed ones. Vista 64, 181.22, 6.6.2

ID: 22524 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 22529 - Posted: 20 Jan 2009, 8:57:18 UTC - in response to Message 22524.  

Please check if those are really 4 tasks and not that it downloaded the simple GUI image files (*.png). I cannot reproduce this problem, not with any of the projects I have set to No new tasks.
ID: 22529 · Report as offensive
Profile Joseph Stateson
Volunteer tester
Avatar

Send message
Joined: 27 Jun 08
Posts: 641
United States
Message 22531 - Posted: 20 Jan 2009, 19:01:48 UTC - in response to Message 22529.  

Please check if those are really 4 tasks and not that it downloaded the simple GUI image files (*.png). I cannot reproduce this problem, not with any of the projects I have set to No new tasks.


seti@home was consistently giveing me WU's from last last night 1/19 (date shown below) up thru 1/20 12:41:59 when I got the message the project is down

Here is the last WU that came in

1/20/2009 10:22:46 AM|SETI@home|Finished download of 09dc08ad.493.9070.12.8.119


All this time (since late last night) I had "No new work" set on the seti@home project.

Right now I cannot verify it because there is no work from the project. Since I read where 6.07 is available, maybe I need to finish the rest of my 6.06 before I can get new work???

If you want to mess with it and have VNC I can email you the ip and my ULTRA VNC password.
ID: 22531 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 22533 - Posted: 20 Jan 2009, 19:11:21 UTC - in response to Message 22531.  

If you want to mess with it and have VNC I can email you the ip and my ULTRA VNC password.

Brrr, no better not. I have DNS problems enough already as it is. ;-)

Can you post the content part for Seti from your client_state.xml file please?
I am only interested in the part from <master_url> to </scheduler_url>

With thanks.
ID: 22533 · Report as offensive
Profile Joseph Stateson
Volunteer tester
Avatar

Send message
Joined: 27 Jun 08
Posts: 641
United States
Message 22535 - Posted: 20 Jan 2009, 19:21:10 UTC - in response to Message 22533.  
Last modified: 20 Jan 2009, 19:23:49 UTC

If you want to mess with it and have VNC I can email you the ip and my ULTRA VNC password.

Brrr, no better not. I have DNS problems enough already as it is. ;-)

Can you post the content part for Seti from your client_state.xml file please?
I am only interested in the part from <master_url> to </scheduler_url>

With thanks.


    <master_url>http://setiathome.berkeley.edu/</master_url>
    <project_name>SETI@home</project_name>
    <symstore></symstore>
    <user_name>BeemerBiker</user_name>
    <team_name>Texas A&amp;M University</team_name>
    <host_venue>home</host_venue>
    <email_hash>3cbc73b6dd56a3e08b5ab71ebf2e3c14</email_hash>
    <cross_project_id>576de0cab03611e6e5e196bf15c1c367</cross_project_id>
    <cpid_time>927830575.000000</cpid_time>
    <user_total_credit>894307.680434</user_total_credit>
    <user_expavg_credit>1240.327092</user_expavg_credit>
    <user_create_time>927830575.000000</user_create_time>
    <rpc_seqno>13</rpc_seqno>
    <hostid>4711247</hostid>
    <host_total_credit>56124.717928</host_total_credit>
    <host_expavg_credit>176.230087</host_expavg_credit>
    <host_create_time>1229474478.000000</host_create_time>
    <nrpc_failures>1</nrpc_failures>
    <master_fetch_failures>0</master_fetch_failures>
    <min_rpc_time>1232480519.142178</min_rpc_time>
    <next_rpc_time>0.000000</next_rpc_time>
    <short_term_debt>44866.835404</short_term_debt>
    <long_term_debt>8040.647550</long_term_debt>
    <cpu_backoff_interval>0.000000</cpu_backoff_interval>
    <cpu_backoff_time>0.000000</cpu_backoff_time>
    <cuda_debt>-9441.462621</cuda_debt>
    <cuda_backoff_interval>0.000000</cuda_backoff_interval>
    <cuda_backoff_time>0.000000</cuda_backoff_time>
    <resource_share>100.000000</resource_share>
    <duration_correction_factor>37.041857</duration_correction_factor>
    <sched_rpc_pending>1</sched_rpc_pending>
    <send_time_stats_log>0</send_time_stats_log>
    <send_job_log>0</send_job_log>
    <dont_request_more_work/>
    <attached_via_acct_mgr/>
    <ams_resource_share>100.000000</ams_resource_share>
    <scheduler_url>http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi</scheduler_url>


OK i did see the DCF so how can boincview get the DCF from this XML file?

Also, since I got that 6.6.2 by guesswork, it might be that my version is not the same as yours.
ID: 22535 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 22540 - Posted: 20 Jan 2009, 19:30:55 UTC - in response to Message 22535.  
Last modified: 20 Jan 2009, 19:47:53 UTC

<dont_request_more_work/>

Yup, it's set to no new tasks. Perhaps that Seti is using a different scheduler that is doing this, I'll have to wait until their weekly outage is over to test that. It's also possible that you are getting CUDA work, that the NNT setting is ignoring that option.

OK i did see the DCF so how can boincview get the DCF from this XML file?

It reads the correct line. Otherwise by magic. ;-)

Also, since I got that 6.6.2 by guesswork, it might be that my version is not the same as yours.

How did you come to that conclusion? I probably got mine from the same place that you did. No, no need to advertise it.

I'll forward this to the developers to double check.
ID: 22540 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 22545 - Posted: 20 Jan 2009, 21:26:44 UTC - in response to Message 22540.  
Last modified: 20 Jan 2009, 21:28:15 UTC

It seems to point in the direction of CUDA work requests ignoring the NNT option.
BOINC 6.6.2 has a separate work-fetch module for CPU and GPU.
ID: 22545 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5077
United Kingdom
Message 22548 - Posted: 20 Jan 2009, 22:31:44 UTC

This issue has been independently reported by Claggy at SETI Main: message 855551.

Note that Claggy got Astropulse (i.e. non-CUDA) tasks, so it couldn't have been the CUDA request which failed to respect NNT.
ID: 22548 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 22549 - Posted: 20 Jan 2009, 22:40:23 UTC - in response to Message 22548.  

You'll find Claggy's message here on the forums as well and I answered him separately. :-)
ID: 22549 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5077
United Kingdom
Message 22557 - Posted: 21 Jan 2009, 9:53:00 UTC - in response to Message 22549.  

You'll find Claggy's message here on the forums as well and I answered him separately. :-)

Well, if the forum had been working the last few days, I might have seen the posts as they came in ;-)

But I still think Claggy's Astropulse allocation (in the other thread) rather scuppers your 'CUDA disobeys NNT' theorem (in this thread). No AP for CUDA yet. If David is looking for a needle in a haystack again, best to have all the evidence in one place.
ID: 22557 · Report as offensive
Profile Joseph Stateson
Volunteer tester
Avatar

Send message
Joined: 27 Jun 08
Posts: 641
United States
Message 22566 - Posted: 21 Jan 2009, 13:35:37 UTC - in response to Message 22557.  
Last modified: 21 Jan 2009, 13:48:32 UTC

You'll find Claggy's message here on the forums as well and I answered him separately. :-)

Well, if the forum had been working the last few days, I might have seen the posts as they came in ;-)

But I still think Claggy's Astropulse allocation (in the other thread) rather scuppers your 'CUDA disobeys NNT' theorem (in this thread). No AP for CUDA yet. If David is looking for a needle in a haystack again, best to have all the evidence in one place.


Hmm - not sure what you are saying. However, overnight I got the biggest ever job allocation I have ever seen.

Snapshot here (used snagit's scrolling capture to get the whole frigging list) Windows photo does not display the .png with a scroll bar but IE7 handled it correctly.

Observations:
(1) I did have "no new tasks" enabled all last night for seti.
(2) After clicking "update" on all projects, I got two more seti, both were 6.08 but the download failed as shown here
(3) There were no more seti 6.06 downloaded thank god
(4) Estimated times for AQUA is over 10 hours, but they finish in less than 2. if the estimate is 10 hours, then why are so many AQUAs requested? I have 56 pending AQUAs with a deadline of feb 10th. That adds up to 24 days and one would think the scheduler would know there are only 19 days before that event occurs.
(5) Cosmology, eijnstein and milkway do not have this problem because they refused to give me any more work.
(6) Poem has a more resonable 2 hour completion (matches the actual) as does rosetta, but rosetta got bumped by the higher priority Poem and there are 3 waiting to run.
(7) Note the date format changes unexpectedly from 24 hour to 12/am/pm and back

I put a work hold on all projects until this clears up. Once all the jobs complete I will put in the 64 bit version of 6.6.2 like I meant to originally. I assume that seti will still manage to get some jobs to me since it is definately ignoring the no new work.

Vista 64, 6.6.2 (32 bit boinc), Q6700 cpu, gtx280, 181.22

HEY - DID NOT GET A SINGLE NVIDIA KERNEL RESET!!!
ID: 22566 · Report as offensive
Profile Joseph Stateson
Volunteer tester
Avatar

Send message
Joined: 27 Jun 08
Posts: 641
United States
Message 22571 - Posted: 21 Jan 2009, 14:44:28 UTC - in response to Message 22566.  

(4) Estimated times for AQUA is over 10 hours, but they finish in less than 2. if the estimate is 10 hours, then why are so many AQUAs requested? I have 56 pending AQUAs with a deadline of feb 10th. That adds up to 24 days and one would think the scheduler would know there are only 19 days before that event occurs.


OK, Just realized this cpu has 4 cores so it possible to do all those. I just have never had more than 1 - 2 day backlog on any system and this list of tasks was by far the largest I have ever seen.
ID: 22571 · Report as offensive

Message boards : Questions and problems : 6.6.2: ignoreing "Wont't get new tasks"

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.