Hi Everyone,

It looks like I fixed the bug with the validator. I'll be keeping an eye on it this weekend to make sure everything is running okay.

---

Great news, thanks for all the efforts.

---

The Server Status still shows it as not running.

The Server Status still shows it as not running.

That problem should be fixed now as well.

---

All my 57 WU's which were waiting for validation now say
"Completed, validation inconclusive", I presume this will sort itself out as the backlog is processed ?

It should! Let me know if it doesn't (but it looks like most of them have already gotten validated).

---

Yes, most have, thanks.
The one's that still say "Completed, validation inconclusive", do not appear to be a problem with the validator per se, more like a problem with the initial replication.
This workunit for instance is a typical example http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=277437505 indicates an initial replication and required quorum of 2, but only one WU was sent out.

Yes, most have, thanks.
The one's that still say "Completed, validation inconclusive", do not appear to be a problem with the validator per se, more like a problem with the initial replication.
This workunit for instance is a typical example http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=277437505 indicates an initial replication and required quorum of 2, but only one WU was sent out.

This is actually kind of a unique thing milkyway@home does. We send out a single workunit, and then if we need to use it for our optimization techniques we'll up the quorum and send another one or two out to validate it before we start using that result.

We also use adaptive replication for results we don't use with a min of 10% (for users who rarely/never have errors validating) and a max of 100% (for users who send junk results all the time) we'll send out extra workunits for validation as well.

---

for users who send junk results all the time

Wouldn't it make sense to block such users/computers? There are for example still lots of those using the old Gipsel apps, so they never return anything useful, but since they get just validation errors instead of computation errors their quota is not decreased.

---

BTW, I don't think everything is working as it should, I'm getting quite a lot of "no work available" answers from the server (much more than usual), OTOH this task for example has been created 8:54:16 UTC and not send out yet at the time of this posting. Usually here at Milkyway a task is send out just few minutes after it has been created.

---

Hm i have nearly 300 wus without validating including a wingman finished too but not resend to a third one :( and it raising hard :/

---

DSKAG Austria Research Team: http://www.research.dskag.at

Another thing: I don't know how important it is for you, that every WU is crunched, but since we have 2 as max # of error tasks I see every now and than a WU like this: WU 272446371.

349310657 	315228 	26 Nov 2012 | 15:20:50 UTC 	10 Dec 2012 | 1:34:20 UTC 	Aborted by user 	0.00 	0.00 	--- 	MilkyWay@Home v1.02 (opencl_amd_ati)
357292680 	398141 	8 Dec 2012 | 15:21:49 UTC 	8 Dec 2012 | 15:24:42 UTC 	Error while computing 	0.00 	0.00 	--- 	MilkyWay@Home v1.02 (opencl_amd_ati)
357294294 	293662 	8 Dec 2012 | 15:25:34 UTC 	8 Dec 2012 | 22:29:36 UTC 	Completed, can't validate 	1,086.41 	3.90 	0.00 	MilkyWay@Home Anonymous platform (ATI GPU)
357550873 	482581 	9 Dec 2012 | 2:31:24 UTC 	9 Dec 2012 | 10:27:58 UTC 	Completed, can't validate 	432.11 	82.03 	0.00 	MilkyWay@Home v1.02 (opencl_nvidia)
358008631 	144013 	9 Dec 2012 | 22:09:07 UTC 	9 Dec 2012 | 22:10:41 UTC 	Error while computing 	1.13 	0.02 	--- 	MilkyWay@Home v1.02 (opencl_nvidia)
358156544 	--- 	--- 	--- 	Didn't need 	0.00 	0.00 	--- 	

Eventually it would be good not to count "aborted by user" as a work unit error.

---