WU will miss deadline
log in

Advanced search

Message boards : Number crunching : WU will miss deadline

Author Message
Stanley A Bourdon
Send message
Joined: 3 Dec 10
Posts: 4
Credit: 25,253
RAC: 0
Message 779 - Posted: 16 Nov 2012, 16:53:57 UTC

Hi

my current WU appears to be taking about 1.3 hours to do 1% if that holds true it will not meat the deadline that is only 48 hours away. Also the estimated time remaining is about 1220 hours which will put it way beyond the deadline.

Sb

Name batch_761_11_1
Workunit 9283422
Created 16 Nov 2012 | 12:20:27 UTC
Sent 16 Nov 2012 | 12:24:01 UTC
Received ---
Server state In progress
Outcome ---
Client state New
Exit status 0 (0x0)
Computer ID 5723
Report deadline 18 Nov 2012 | 16:10:41 UTC
Run time 0.00
CPU time 0.00
Validate state Initial
Credit 0.00
Application version Tsinghua Nano Tech Research v1.18
____________
Stanley


Boinc Wikipedia - the FAQ in active change

archeye*
Send message
Joined: 3 Oct 12
Posts: 5
Credit: 965
RAC: 0
Message 780 - Posted: 16 Nov 2012, 17:12:49 UTC

Hi

I have about the same,



The last batch of nano WUs all failed with error.

I have kicked this one out, not waiting to see if it gets done in a sensible time or not.

Protein research ones are still being processed though :)

regards,

archeye*
Send message
Joined: 3 Oct 12
Posts: 5
Credit: 965
RAC: 0
Message 781 - Posted: 16 Nov 2012, 20:05:47 UTC

Hi,

update on the nano WUs.

They are running on another of my computers and on return the server is registering an error condition just like the last time a nano batch was sent out.

So "Won't get new tasks" will for me be alongside CAS project until the nano batch is finished with.

Regards,

NATE1
Send message
Joined: 17 Dec 11
Posts: 31
Credit: 66,877
RAC: 0
Message 782 - Posted: 17 Nov 2012, 0:40:48 UTC - in response to Message 779.
Last modified: 17 Nov 2012, 1:29:03 UTC

Hi

my current WU appears to be taking about 1.3 hours to do 1% if that holds true it will not meat the deadline that is only 48 hours away. Also the estimated time remaining is about 1220 hours which will put it way beyond the deadline.

Sb
Tsinghua Nano Tech Research




about 1% every hour. so 110 hours to do a task, running 24/7.
5 day deadline, unless it jumps a large % at some point.
(I seen a number of other projects wu do that)


edit:look like it has to do 511 somethings :) every 7 minutes it does 2. so via my strange math 511 / 2 = 255.5, 255.5 * 7 minutes = 1788.5 minutes, 1788.5 minutes / 60 minutes per hour = 29.8083333.....333 hours, therefor based on that logic 48 hours or 2 day deadline is ok.

Stanley A Bourdon
Send message
Joined: 3 Dec 10
Posts: 4
Credit: 25,253
RAC: 0
Message 784 - Posted: 17 Nov 2012, 5:23:38 UTC - in response to Message 782.


edit:look like it has to do 511 somethings :) every 7 minutes it does 2. so via my strange math 511 / 2 = 255.5, 255.5 * 7 minutes = 1788.5 minutes, 1788.5 minutes / 60 minutes per hour = 29.8083333.....333 hours, therefor based on that logic 48 hours or 2 day deadline is ok.


your computer is hidden, how fast is it?

i am down to 58.42 min per % so am down to 97.35 hours total still well beyond the 2 day deadline

Sb
____________
Stanley


Boinc Wikipedia - the FAQ in active change

Profile skgiven
Avatar
Send message
Joined: 11 Sep 10
Posts: 11
Credit: 123,324
RAC: 0
Message 786 - Posted: 17 Nov 2012, 9:14:48 UTC - in response to Message 784.

I wouldn't worry too much about the 'deadline', just if it's running and progressing or not. You will probably still get credit and more importantly the WU will still be of use even if its a day or so late.
____________

NATE1
Send message
Joined: 17 Dec 11
Posts: 31
Credit: 66,877
RAC: 0
Message 787 - Posted: 17 Nov 2012, 11:06:49 UTC - in response to Message 784.



your computer is hidden, how fast is it?

i am down to 58.42 min per % so am down to 97.35 hours total still well beyond the 2 day deadline

Sb


stock 2.2 ghz amd laptop

NATE1
Send message
Joined: 17 Dec 11
Posts: 31
Credit: 66,877
RAC: 0
Message 788 - Posted: 17 Nov 2012, 13:42:54 UTC - in response to Message 787.



your computer is hidden, how fast is it?

i am down to 58.42 min per % so am down to 97.35 hours total still well beyond the 2 day deadline

Sb


stock 2.2 ghz amd laptop


stats on one wu:
% 21.124
elapsed 16 hours
remaining 535 hours

not bad when you consider it started with a remaining time of 1044 hours.
therefor 16 hours of elapsed time = almost 50% drop in remaining time.:)

NATE1
Send message
Joined: 17 Dec 11
Posts: 31
Credit: 66,877
RAC: 0
Message 791 - Posted: 18 Nov 2012, 15:00:33 UTC - in response to Message 788.
Last modified: 18 Nov 2012, 15:09:06 UTC



your computer is hidden, how fast is it?

i am down to 58.42 min per % so am down to 97.35 hours total still well beyond the 2 day deadline

Sb


stock 2.2 ghz amd laptop


stats on one wu:
% 21.124
elapsed 16 hours
remaining 535 hours

not bad when you consider it started with a remaining time of 1044 hours.
therefor 16 hours of elapsed time = almost 50% drop in remaining time.:)


it' done 38.12236667 hours, but this: [Errno 9] Bad file descriptor

16 Nov 2012 | 21:16:36 UTC 18 Nov 2012 | 12:52:09 UTC 137,240.52 81,259.22 pending


Stderr output

<core_client_version>7.0.28</core_client_version>
<![CDATA[
<stderr_txt>
wrapper: starting
16:16:14 (1500): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
wrapper: starting
16:34:33 (4364): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
wrapper: starting
07:35:44 (5076): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
close failed: [Errno 9] Bad file descriptor
wrapper: starting
12:08:34 (4948): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
close failed: [Errno 9] Bad file descriptor
wrapper: starting
17:07:26 (4776): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
wrapper: starting
17:56:55 (3436): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
wrapper: starting
19:11:38 (4420): wrapper: running start_lammps.exe ( -var restart 0 -in lammps_script -var looprun 3000 -var loopnumber 258 -var thermon 129 -var dumpn 1290 -var e_steps 0 -var l_script lammps_script -var vx 0.00026 -var vy 0.00127 -var vz 0.00046)
close failed: [Errno 9] Bad file descriptor
07:50:02 (4420): wrapper: running parse_result.exe ()
close failed: [Errno 9] Bad file descriptor
07:50:34 (4420): called boinc_finish

</stderr_txt>
]]>

adrianxw
Send message
Joined: 21 Oct 11
Posts: 6
Credit: 13,247
RAC: 0
Message 792 - Posted: 18 Nov 2012, 16:26:59 UTC

Mine are also going to go past the 2 day deadline, but probably not by much. What concerns me is that when they do, do they get sent to another cruncher thus wasting resources?

They need to look at the allowed run time or people are going to dump them.

NATE1
Send message
Joined: 17 Dec 11
Posts: 31
Credit: 66,877
RAC: 0
Message 794 - Posted: 18 Nov 2012, 16:56:58 UTC - in response to Message 792.

Mine are also going to go past the 2 day deadline, but probably not by much. What concerns me is that when they do, do they get sent to another cruncher thus wasting resources?

They need to look at the allowed run time or people are going to dump them.


most projects work this way:
if the deadline is missed the workunit will be sent to another host, but, if you can get the wu back before the other host it was sent to, you get credit.
problem is: "most" projects. some will not let you miss the deadline by even 1 second. so just let it run and see how this project is set up, as far as giving credits to late workunits.

Stanley A Bourdon
Send message
Joined: 3 Dec 10
Posts: 4
Credit: 25,253
RAC: 0
Message 800 - Posted: 19 Nov 2012, 14:54:15 UTC

Hi

it was late but got validated

It was at about 53 hours run time and 48% done the last time i looked at it than a couple of hours later it uploaded

strange behavior

still these need a longer deadline

Sb
____________
Stanley


Boinc Wikipedia - the FAQ in active change

Profile skgiven
Avatar
Send message
Joined: 11 Sep 10
Posts: 11
Credit: 123,324
RAC: 0
Message 801 - Posted: 19 Nov 2012, 19:50:50 UTC - in response to Message 800.

Your task:
batch_761_11_1 9283422 16 Nov 2012 | 12:24:01 UTC 19 Nov 2012 | 6:57:45 UTC Completed and validated 199,659.40 162,052.70 675.56 Tsinghua Nano Tech Research v1.18

name batch_761_11
application Tsinghua Nano Tech Research
created 16 Nov 2012 | 12:20:23 UTC
canonical result 19951042
granted credit 675.56

Work Unit:
minimum quorum 2
initial replication 2
max # of error/total/success tasks 8, 12, 4
Task - - Computer - - Sent - - Time reported or deadline - - - - - - - - - - Status - - - Run time (s) - CPU time (s) - Credit - Application
19951042 11872 16 Nov 2012 | 12:24:00 UTC 18 Nov 2012 | 9:35:15 UTC Completed and validated 139,607.29 134,716.70 675.56 Tsinghua Nano Tech Research v1.18
19951043 5723 16 Nov 2012 | 12:24:01 UTC 19 Nov 2012 | 6:57:45 UTC Completed and validated 199,659.40 162,052.70 675.56 Tsinghua Nano Tech Research v1.18
19977521 --- --- --- Didn't need 0.00 0.00 ---

As you can see, these tasks are sent out twice to begin with, but another could be sent out if needed...

From what you said, it sounds like they might have a cut off point; perhaps after so many steps.

PS. Still terrible credit vs runtime here!
____________

Tex1954
Send message
Joined: 23 Apr 11
Posts: 38
Credit: 812,304
RAC: 44
Message 805 - Posted: 21 Nov 2012, 14:12:58 UTC
Last modified: 21 Nov 2012, 14:14:22 UTC

I too have a couple long WU's that look like they will miss deadlines. Both have run over 21 hours and one of them is barely over 6% done... The time remaining seems wrong.

I'll let them run anyway...

8-)

Tex1954
Send message
Joined: 23 Apr 11
Posts: 38
Credit: 812,304
RAC: 44
Message 806 - Posted: 21 Nov 2012, 21:30:10 UTC

One of the two long WU's finished after about 28 hours in time I guess since it's in my pending tasks; the other is already late now...

8-)

astroWX
Send message
Joined: 27 Sep 10
Posts: 29
Credit: 147,739
RAC: 0
Message 808 - Posted: 22 Nov 2012, 23:17:42 UTC

I have one in the same condition. (Another one, started at the same time, on the same machine, finished a couple days ago.)

____________
Greetings from the US Pacific Northwest.

Tex1954
Send message
Joined: 23 Apr 11
Posts: 38
Credit: 812,304
RAC: 44
Message 809 - Posted: 23 Nov 2012, 1:30:49 UTC
Last modified: 23 Nov 2012, 1:31:37 UTC

Well, this one LONG WU was timed out by the system...



And at the time, this WU had run a LONG TIME! (like 56 hours and tons more to go)



Soo, I aborted it...

Somebody needs to adjust things for the future I think...

8-)

Profile skgiven
Avatar
Send message
Joined: 11 Sep 10
Posts: 11
Credit: 123,324
RAC: 0
Message 811 - Posted: 24 Nov 2012, 11:28:48 UTC - in response to Message 809.

I thought the estimated run time was based on the average time of previous tasks using the same app, rather than on the project. Either way if you run a few short tasks and then get a long one it skews the times up, and these long tasks are variable. That said the estimated run time and cut off times could be adjusted server side, by the researchers. I don't think progression reports linearly anyway.

Going by the project/server status CAS has two project apps at present, Tsinghua Nano Tech Research, and ICT Protein Structure Prediction(2nd Generation).
I guess you could choose to run one and not the other from your; from
the CAS@home preferences page choose Edit CAS@home preferences and select one app.
Don't know how long you would have two sets of task types however.

____________

Profile [AF>WildWildWest] RLDF
Send message
Joined: 30 Oct 10
Posts: 1
Credit: 3,030,905
RAC: 0
Message 813 - Posted: 24 Nov 2012, 17:17:18 UTC

Why giving too long workunits ?
i got one with a deadline november, 26th, 10pm (50h) but the estimated time is 900h !!!
is this serious ?

we're ok to crunch for science but not to waste time for nothing...
electricity isn't free ...

wenjing wu
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 13 Sep 10
Posts: 161
Credit: 751,216
RAC: 0
Message 817 - Posted: 25 Nov 2012, 13:30:10 UTC - in response to Message 813.

Sorry , Everyone! The Nano tech application(Lammps) has very dynamical computing time, so we have a program testing each job and do an estimation of the computing time of each job. However this estimation is not accurate enough , and the deadline is not set up accordingly.. I will fix this bug ASAP.

Thanks for your feedback. And also my apologies for the late reply due to my traveling schedule!
____________
加油!CAS@home!我们帮助科学家跟时间赛跑!
Go CAS@home! We help scientists to race against time!


Post to thread

Message boards : Number crunching : WU will miss deadline