BTW, the 38.8 hour task was at about 60 percent and already overdue. It was held up because the tasks before were also inordinate long and as it is, repairs only get 3.5 days from point of receipt. Now my tab is mobile again, just as it was designed to be, mobile.

Mentioned before here jeanguy. each rig seems to have a different threshold but there is a time based line which, if crossed, triggers an anti cheat mechanism. This then give a flat rate figure (amount of points) regardless of how much longer the wu takes. The catch for us honest workers though is that particularly long wu's, such as recent FAHV offerings, get caught as well

I just had my first problems with the larger work units. Three have errors over file size being too big
here is an excerpt from the log
#[05:44:08] Vina exited normal 0.
#[05:44:08] Finished task #138 cpu time used 773.468558
#[05:44:08] Starting task 139,CPU time is 116243.708347.
#[05:44:08] ./ZINC12971592.pdbqt size = 25 6 #../../projects/www.worldcommunitygrid.org/fahv.x3ZCM_A_IN_Y3b_rig.pdbqt #size = 2792 0
#[05:57:14] Vina exited normal 0.
#[05:57:14] Finished task #139 cpu time used 761.706083
#05:57:15 (6028): called boinc_finish
#</stderr_txt>
#<message>
#upload failure: <file_xfer_error>
#<file_name>FAHV_x3ZCM_A_IN_Y3b_rig_0226067_0036_1_0</file_name>
#<error_code>-131 (file size too big)</error_code>
#</file_xfer_error>
this is for workunits
FAHV_x3ZCM_A_IN_Y3b_rig_0226067_0036
FAHV_x3ZCM_A_IN_Y3b_rig_0226194_0017
FAHV_x3ZCM_A_IN_Y3b_rig_0226063_0010

this is on an 17 4771 with 8 gb ram and an ssd for boinc. each one had over 30 hrs of runtime.

I find the posts in this thread rather alarming. I am certainly glad I am not running on an android device.

I have my cache set very low. Usually about a day but right now set at .75. This means I have no problem with getting these things cleared on time.

I have had none run over 25 hours that I know of and think they run less than 23 usually.

Problem is that I just had another end with an error. That makes 3 of them.

While I am getting credited with points on them, over 450 each, I seriously question the value of running them. I currently have 2 running, one just started adn the other at 80% after 16:20 run time and just recieved 3 more.

I could be running WUs that actually run and finish without error.

Is there really a point to keep sending these dogs out?

What is the point of running unreliable WUs?

If there is some real value to someone of us running these things that is fine. I will make sure to check back in here tomorrow and see what folks have to say.

If there is no indication of a good reason to run them I will do 1 of 2 things. Quit running faat WUs or simply abort these aparently stupid ones.

I really don't like either option as this is just foisting the problem off on other crunchers.

These errors are all the 131 errors and I fail to see how running them on another box, as a normal WU or a repair unit is going to change that.

While I know little or nothing of how these units are built it seems that they generate way too much output. This has been going on for some time and not improving. The early ones would finish without this error.

I would, I am sure, be a pain to simply pull all of them and cause time loss in research. Would this time loss be more than the time lost on simply wasted time running these 131 WUs?

As I understand it they do not even get uploaded and therefore any possibility of them having any useful output is moot. They just go to /dev/nul.

Sorry, I may be very thick but I just don't see the point.

Please note, I have just done a quick check and I am seeing the updated workunits going out. The new estimated work units start with 880194.

Thanks,
-Uplinger