I wonder if tasks that have been committed to a client but have not yet
been delivered to the client are being counted. I.e. those tasks where the
request result has been received by the client, but the file transfers are
not yet complete.
jm7
Kamran Karimi
<kkarimi@dwavesys
.com> To
Sent by: <[email protected]>
<boinc_dev-bounce cc
[email protected]
u> Subject
Re: [boinc_dev] <total_limit> in
config_aux.xml not respected
05/12/2011 05:20
PM
I tried again after setting total_limit to 1: The client downloads 1 job
for each scheduler request, but does it twice (two scheduler requests,
one after the other), for a total of 2 downloaded jobs . After that the
client shows a message from the server indicating that the limit on
tasks in progress has been reached.
-Kamran
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Kamran Karimi
Sent: May-12-11 10:40 AM
To: [email protected]
Subject: Re: [boinc_dev] <total_limit> in config_aux.xml not respected
Just a note that this is on a quad-core machine (Xeon 5507, 4-cores, no
hyper-threading). It may or may not have an effect on the reported
problem.
-Kamran
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Kamran Karimi
Sent: May-12-11 10:33 AM
To: [email protected]
Subject: [boinc_dev] <total_limit> in config_aux.xml not respected
Hi all,
After a server upgrade yesterday, it seems that the total_limit tag in
config_aux.xml is not respected. The relevant entry is:
<app>
<app_name>Fokker_Planck</app_name>
<total_limit>
<jobs>2</jobs>
</total_limit>
</app>
Which is meant to make sure only 2 jobs are downloaded to the volunteer
computer. What actually happens is that the client downloads 4 jobs
(first 2, and then another scheduler request is send and another 2 are
downloaded). I tried <cpu_limit> too, but it didn't change anything. BTW
there is no <project> tag in this file.
The relevant entries in config.xml (which I assume are being ignored
when config_aux.xml is present) are:
<one_result_per_user_per_wu> 1 </one_result_per_user_per_wu>
<max_wus_in_progress> 1 </max_wus_in_progress>
<max_wus_in_progress_gpu> 1 </max_wus_in_progress_gpu>
<gpu_multiplier>1</gpu_multiplier>
<cuda_multiplier>1</cuda_multiplier>
So nothing there could cause 4 workunits to be downloaded.
Any idea how I can enforce the desired behaviour?
-Kamran
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
