Rickjb's ErrorSCC_WUs_200504.txt

Dear WCG Techs
When I scanned my crunching "farm" earlier today (6 May 2020 AEST) with BoincTasks, I noticed an anomoly
with an SCC WU that was about to finish.
After it finished I looked up its result on WCG.
It was deemed Valid but had claimed far more CPU time than it had actually run.
The Valid runtimes for SCC under Linux are currently quite variable, ranging from about 30m to more than 2h,
but not 5h or more.

I went though all results for SCC on this device that are currently online and have recorded the names of all
SCC WUs with anomolousclaimed runtimes. The list appears near the bottom of this file.

I also looked through Valid SCC results for all devices, and only the one device has given anomolous results.

Here are details of the WU I noticed while it was running on BoincTasks:
------------------------------------------------------------------------
Device Name: Callisto-VM

Result Name 	OS type 	OS version 	App Version Number 	Status 	Sent Time 	Time Due /
 Return Time 	CPU Time / Elapsed Time (hours) 	Claimed/ Granted BOINC Credit
SCC1_ 0003840_ FoxO1-A_ 89973_ 0-- 	Linux 	3.2.0-4-amd64 	708 	Valid 	5/5/20 21:22:16 	5/6/20 03:31:06

6.15 / 6.15 	168.6 / 168.6
---------
Log file:
<core_client_version>7.0.27</core_client_version>
<![CDATA[
<stderr_txt>
INFO: result number = 0
INFO: No state to restore. Start from the beginning.
[13:10:35] Number of tasks = 1
[13:10:35] Running task 0,CPU time at start of task 0 was 0.000000
[13:10:35] ./cmpd-1189973.pdbqt size = 20 3 ../../projects/www.worldcommunitygrid.org/scc1.FoxO1-A.pdbqt size = 468 0
[13:41:54] Finished task #0 cpu time used 18446745959.383400
13:41:54 (13029): called boinc_finish(0)

</stderr_txt>
]]>
Thus it ran for only 31 min, but claimed 6.15h !!
---------------------------------------------------

More typical is:
Name -- CPU Time / Elapsed Time
SCC1_ 0003855_ FoxO1-A_ 9691_ 0-- 0.76 / 0.76
Log file:
<core_client_version>7.0.27</core_client_version>
<![CDATA[
<stderr_txt>
INFO: result number = 0
INFO: No state to restore. Start from the beginning.
[12:04:38] Number of tasks = 1
[12:04:38] Running task 0,CPU time at start of task 0 was 0.000000
[12:04:38] ./cmpd-2609691.pdbqt size = 26 5 ../../projects/www.worldcommunitygrid.org/scc1.FoxO1-A.pdbqt size = 468 0
[12:50:09] Finished task #0 cpu time used 2753.408077
12:50:09 (12139): called boinc_finish(0)

</stderr_txt>
]]>

--------------------------------------------------
Here are the names of all affected WUs that are still up in my WCG account Results section:
SCC1_ 0003840_ FoxO1-A_ 89973_ 0-- (the one above)
SCC1_ 0003849_ FoxO1-A_ 92170_ 0--
SCC1_ 0003917_ FoxO1-B_ 35883_ 0--
SCC1_ 0003861_ FoxO1-A_ 6450_ 0--
SCC1_ 0003909_ FoxO1-B_ 76386_ 0--
SCC1_ 0003861_ FoxO1-A_ 88142_ 0--
SCC1_ 0003861_ FoxO1-A_ 88142_ 0-- (claimed 16.25h for 38m work)
SCC1_ 0003921_ FoxO1-B_ 93377_ 1-- (still PV, will check wingman's when she's done)
SCC1_ 0003917_ FoxO1-B_ 35883_ 0--
SCC1_ 0003921_ FoxO1-B_ 93377_ 1--
SCC1_ 0003918_ FoxO1-B_ 45681_ 0--
SCC1_ 0003921_ FoxO1-B_ 11396_ 0--
SCC1_ 0003921_ FoxO1-B_ 26482_ 0--
SCC1_ 0003866_ FoxO1-A_ 71351_ 0--
SCC1_ 0003918_ FoxO1-B_ 75346_ 0-- (return time 5/4/20 10:43:17)

--------------------------------------------------
Probable explanation:
One of my devices (17-3770K, Debian Linux 7.x) crashed, dead, about 1 week ago.
1 RAM stick of 2 died.
I've temporarily replaced it with 2 dissimilar sticks, both in Channel B (so single-channel operation instead of dual).
I stress tested the machine before putting it back online, but I seem to have missed its flaw(s).
The machine has not crashed with the mixed-RAM setup.
I will replace both RAMsticks shortly.
--------------------------------------------------

Sincere apologies, but it may have highlighted a weakness in the verification system.

It would be interesting to compare my results to those of the resends, if those are issued.
 - Rickjb

--------------------------------------------------