A major computer re-configuration prompted start on a new model two weeks ago. My stats show the credits I received for only half of the time-steps (43208, 54010, 75614, 86416). Going through boinc.log I see that computation was normal and that boinc talked to base every time ("Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded"). Still my account only shows half of the trickle results got credit. Why?

---

If you check the result ID 786718, you'll see that your computer has uploaded 8 trickles for that result, and that you have got the correct amount of credit for 8 trickles. Why they don't show on your main stats page, I don't know, although there seem to have problems lately with trickles not being generated on all the stats pages.
HTH
Alan

---

When first reading your answer and checking I though this was OK. I just looked in detail and found following:

04 May 2005 10:11:42 102841 786718 2sl0_300151672_1 1 118822 389005 3.2738
03 May 2005 20:15:10 133215 786718 2sl0_300151672_1 1 108020 339397 3.1420
01 May 2005 19:30:22 133215 786718 2sl0_300151672_1 1 97218 305394 3.1413
30 Apr 2005 23:26:42 133215 786718 2sl0_300151672_1 1 86416 271469 3.1414
29 Apr 2005 18:55:05 133215 786718 2sl0_300151672_1 1 75614 237305 3.1384
28 Apr 2005 15:31:34 102841 786718 2sl0_300151672_1 1 64812 216381 3.3386
27 Apr 2005 21:47:35 133215 786718 2sl0_300151672_1 1 54010 168939 3.1279
26 Apr 2005 14:22:45 133215 786718 2sl0_300151672_1 1 43208 134590 3.1149
25 Apr 2005 01:40:07 102841 786718 2sl0_300151672_1 1 32406 112968 3.4860
23 Apr 2005 17:29:14 102841 786718 2sl0_300151672_1 1 21604 78569 3.6368
23 Apr 2005 02:48:53 102841 786718 2sl0_300151672_1 1 10802 44326 4.1035

In other words, half of the trickles come from another Host. Inspection shows that this indeed another PC (diff operating system, diff CPU type). Looking at results sent in for that host I noticed that by coincidence (?) this one is working on a WU with very similar name: 2sa0_300151272_1 instead of 2sl0_300151672_1 (3rd position a instead of 'l'). Am I seeing ghosts or am I missing something

---

> In other words, half of the trickles come from another Host. Inspection shows
> that this indeed another PC (diff operating system, diff CPU type). Looking at
> results sent in for that host I noticed that by coincidence (?) this one is
> working on a WU with very similar name: 2sa0_300151272_1 instead of
> 2sl0_300151672_1 (3rd position a instead of 'l'). Am I seeing ghosts or am I
> missing something

There's definitely something wrong, but the other WU is a red herring. That host is a hyper-threaded P4 running 2 simultaneous workunits.

Result id 786718 is registered as being run on the other host (102841) rather than yours (133215). Looking at the trickle history for both hosts I can see that 6 trickles have been registered to your host and 5 to the other one, just as is shown in the trickles for the result.

I've no idea how the result could have ended up cross-linked to 2 hosts like this (it was downloaded to the other host at 22 Apr 2005 9:11:23 UTC), but I've drawn it to Tolu's attention.

I'd advise you to force a discard of that workunit as you're duplicating the work being done by the host that the server expects the result to be returned by, and you might only be receiving temporary credits for the work you're doing (you may well be losing them in the periodic recalculation).

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

I see something strange om my computer, suddenly I dont get an trickles registered for 710083, http://climateapps2.oucs.ox.ac.uk/cpdnboinc/trickle.php?resultid=710083 but still is calculating on my machine:
1whv_300109677 - PH 3 TS 084535 - 22/10/2055 03:30 - H:M:S=0485:43:36 AVG= 2.90 DLT= 1.64

For my other model the trickles is updated as can be seen here:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/trickle.php?resultid=713515

---

My own stats are all over the place for this week, probably related to the server problems last weekend.

Just about every one of my hosts has returned trickles that aren't showing up and 4 of the 6 completed results I've uploaded this week don't appear to have been registered as such on the server. They have a shortfall of 26 uncredited trickles between them.

I expect it'll all get sorted out eventually, but we might have to wait a while.

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

This trickle business is a little like the problems reported when transferring a model to a different computer to complete it.

I wonder if, during one of the recent server down times, a server was replaced, and some software is different.

We really need a full explanation of what has been happening during these times to try and work it out.
Hannah appears to be back. Perhaps she could post something, either here or on the community forum. As well as getting it fixed.

Les

One thing now comes to my mind that would fit perfectly into the problem category of lost trickles :

I tried to modify a CPDN preference setting and it said it had updated the profile but the old value was still there. I tried it two more times without success and then decided to modify the XML file instead and update the profile from there.

As updating the database through the XML file worked, I didn't post about it. Maybe I should have done :-/


On Informix and DB2, the first I would do now is a drop distributions and refresh the distributions or maybe even rebuild the indexes, on MySQL such a thing can be done only while the database engine is not serving requests I think.

---