Cool. This might be a case where the allocated disk quota is eaten up by the fact that some data files (those starting with "l1_") never seem to get deleted, something Gary reported here lately. Bernd is informed and the scheduler people are working on fixing it, I guess, igf not already done.
Yeah, I saw that post. I had recently done a Reset Project, so I only had 12MB of clutter. That doubled when I got a new datapack, but it still is minimal for me, although I guess it could grow to be a problem on smaller drives or if your settings are tighter...
As for Bruce's machine, it appears he fixed it as it turned in full results after that event. It was just real ironic...
As for Bruce's machine, it appears he fixed it as it turned in full results after that event. It was just real ironic...
I suspect that the problem with Bruce's machine was not really to do with BOINC or EAH. Bruce's "cattle station" (as someone once dubbed them :) ) was set up to do other post-processing tasks on the data returned by the EAH volunteers, IIRC. I believe they run EAH as a background task just to make sure that no CPU cycles are wasted. Some time ago, Bruce had a much higher RAC than he currently has so I assume that his involvement in basic EAH crunching has wound down considerably these days. The machines are busy on other things :).
I think that it is likely that the real work on that machine probably filled up the partition so that the background EAH task suddenly didn't have enough available free space any more. That machine actually takes about 7 days to return a result and has a RAC of little more than 100 which shows how little basic EAH work it is doing.
Also, I've checked one of my more modern (and hence faster) machines to see how much clutter (l1_* files) has accumulated since the start of S5R2. There are about 30 or so of these dating back to April 23 which is about when this run commenced. So about 100MB of l1_* clutter plus the clutter of all the old .exes and .pdbs. The entire BOINC directory is less than 200MB so unless you have a super fast machine with a very small partition for BOINC, it shouldn't yet be too much of a problem for participants.
Thanks for Bikeman for drawing Bernd's attention to this to get it looked at before it really does become a problem.
Whilst the initial replication is still 2, I have a WU that has now been issued 6 times and about to go to 7 as of the 4/7.
I have been waiting since the 9/5/07 for it to validate, I just hope I don't get the validation error which will mean another WU to be sent out and flip a coin to see who gets the money.
RE: Cool. This might be a
)
Yeah, I saw that post. I had recently done a Reset Project, so I only had 12MB of clutter. That doubled when I got a new datapack, but it still is minimal for me, although I guess it could grow to be a problem on smaller drives or if your settings are tighter...
As for Bruce's machine, it appears he fixed it as it turned in full results after that event. It was just real ironic...
Brian
RE: As for Bruce's
)
I suspect that the problem with Bruce's machine was not really to do with BOINC or EAH. Bruce's "cattle station" (as someone once dubbed them :) ) was set up to do other post-processing tasks on the data returned by the EAH volunteers, IIRC. I believe they run EAH as a background task just to make sure that no CPU cycles are wasted. Some time ago, Bruce had a much higher RAC than he currently has so I assume that his involvement in basic EAH crunching has wound down considerably these days. The machines are busy on other things :).
I think that it is likely that the real work on that machine probably filled up the partition so that the background EAH task suddenly didn't have enough available free space any more. That machine actually takes about 7 days to return a result and has a RAC of little more than 100 which shows how little basic EAH work it is doing.
Also, I've checked one of my more modern (and hence faster) machines to see how much clutter (l1_* files) has accumulated since the start of S5R2. There are about 30 or so of these dating back to April 23 which is about when this run commenced. So about 100MB of l1_* clutter plus the clutter of all the old .exes and .pdbs. The entire BOINC directory is less than 200MB so unless you have a super fast machine with a very small partition for BOINC, it shouldn't yet be too much of a problem for participants.
Thanks for Bikeman for drawing Bernd's attention to this to get it looked at before it really does become a problem.
Cheers,
Gary.
RE: Whilst the initial
)
Happy day, it has finally been validated after the seventh replication, the wait since the 9/5/07 was worth it.