Project: 3051 (Run 9, Clone 32, Gen 24) long 1-4 interaction

[GTron]

One of my folders was issued Project: 3051 (Run 9, Clone 32, Gen 24) and would consistently get "Warning: long 1-4 interactions" about 2 min 20 sec after completing 93%, followed 4 sec later by "CoreStatus = 0 (0)" and "Client-core communications error: ERROR 0x0".

This dedicated folder is a q6600/G0 @2.4 running Ubuntu 7.04 and has been very stable.

It was initially issued this WU Saturday 2/23 (MDT) while still running the beta 1 client. Late Tuesday 2/26, after it had failed 4 times, I stopped it, upgraded to beta 2, and restarted it at 0%. That restart and the 2 subsequent runs also failed. On the 8th attempt, I stopped and restarted the WU at 65% and it then ran to completion early this morning, 2/29. (Re-downloading of the core did occur as expected under both clients.)

@DanEnsign, aside from being notified of a WU getting "long 1-4 interactions" (since they don't otherwise get reported back) what information would help you?

[bruce]

Interesting. . . you have received full credit for that WU. I wonder if it was processed on the same machine or a different one.
Hi GTron (team 0),
Your WU (P3051 R9 C32 G24) was added to the stats database on 2008-02-29 06:19:36 for 1440 points of credit.

This was from a Linux machine with UserID 14F1XXXXXXXXXXXXXX

[DanEnsign]

@DanEnsign, aside from being notified of a WU getting "long 1-4 interactions" (since they don't otherwise get reported back) what information would help you?

Your story about stopping then restarting the clients, which resulted in the WU getting completed, may itself be a useful clue to what the 'Client-core communications error: ERROR 0x0' headaches are.

But for now I'm happy you could finish the WU.

Dan

ADDITIONAL : I should add, this WU looks perfect from my end -- no indication of constraints* issues that might lead to "long 1-4 interactions" messages.

* constraints == little doohickeys that keep (some) bond lengths constant, because their vibrations aren't well described by harmonic/Hookean potentials (like a spring).

[GTron]

@Bruce -- Wonder no more, it was all processed on a single machine. Thanks, nice to know the info on the points, matches up with when the WU finished and was returned though I thought deadlines might have intervened. I've never really followed or worried about the points in that detail, they're primarily an indication that contributions to the cause are being made. I fold in "memory" of my late parents.

@Dan -- It had been several months since the last previous "long 1-4 interactions". I had several last fall and summer, iirc, and posted a few times on the old forum. I usually, but not always, was able to complete those WUs by stopping and restarting them shortly (<5%) before the failure point. The 65-93% spread for this WU is larger than any of the previous times. Back then, I was worried about things like buffer overflows and expressed concern about a stop/restart apparently yielding different results than a straight run. In any case, I am VERY glad you are looking into it. THANKS