1) Message boards : Number crunching : WUs starting from beginning at each start of BOINC (Message 6357)
Posted 4 Mar 2005 by Vax
Post:
Well, the problem still exists. The same workunit (on the machine at my office) has now restarted back at 0% completion 4 more times (since I last wrote) each time LHC starts back up on BOINC.

I confirmed last night that the same problem is happening on LHC workunits on my home computer.

If it was just happening on my work machine, I'd think it's a workunit from hell, but since it's happening on both machines, it must be something wrong with the sixtrack software, or how it works with BOINC.

Real drag though, in an hour, the workunit gets to about 90-94% completion when LHC swaps out, if the computer (or sixtrack) was just a snick faster, I'd manage to get that sucker done. Instead I find out 4 hours later that it restarted back at 0 again.

Unless someone has a different solution, I'm going to have to detach LHC from BOINC. I hate to spend the effort on something that is just wasting resources that could be used to analyse the other 3 programs I'm running on BOINC.

Tom

> Having the same problem here.
>
> I have 4 projects running in BOINC. LHC was idle while something else was
> running. I noted LHC was sitting at somewhere above 96% completion when it
> got pulled from memory to swap to another project. When it kicked back in, it
> restarted back at 0% completion and is chugging away (finally back to 31%
> completion).
>
> This is the workunit particulars:
> LHC@home - 2005-03-03 11:22:45 - Restarting result
> v64lhc87-24s12_14575_1_sixvf_3481_2 using sixtrack version 4.64
> Tom
>
2) Message boards : Number crunching : WUs starting from beginning at each start of BOINC (Message 6316)
Posted 3 Mar 2005 by Vax
Post:
Having the same problem here.

I have 4 projects running in BOINC. LHC was idle while something else was running. I noted LHC was sitting at somewhere above 96% completion when it got pulled from memory to swap to another project. When it kicked back in, it restarted back at 0% completion and is chugging away (finally back to 31% completion).

BOINC is running almost 24/7, so it's not an issue with turning it off or on.

I do not know if this particular workunit has been doing this continually, I only noticed it today. I do suspect that other LHC workunits (on my work and home machine) have done the same thing in the past.

This is the workunit particulars:
LHC@home - 2005-03-03 11:22:45 - Restarting result v64lhc87-24s12_14575_1_sixvf_3481_2 using sixtrack version 4.64

On a side note, I do not know how often I have had workunits swap out at 96-99% completion. Maybe something for BOINC to consider is to look at the estimated remaining time for a workunit and, if it's less than a certain amount of time, finish the darn thing before swapping out to do another project. It seems silly for a workunit to sit for hours, waiting to start again, while other projects swap in and out and it only had 10-15 minutes left to finish. Just a thought (pet peeve hat off).

Tom



©2024 CERN