.... Also I noticed E@H WUs in "work" tab say they are between an hour and two hours longer than they were before. Are they really longer or is that something to do with benchmark reset or something else.
R.L.,
If your benchmark total has dropped significantly below the previous marks, that could account for longer estimated time, but I see that the last 2 Einstein results you've turned in have been about 1.4 hours longer crunchtime than the balance of your work, so the Duration Correction Factor has bumped the estimate of To Completion time accordingly. This scales upward very quickly, the purpose of which is to prevent Boinc from taking on more work than it can handle. If your rig returns to crunching WUs in the 29K range, then the DCF will adjust downward, albeit incrementally. It may take 12-15 consecutive WUs crunched at the 29K range for the To Completion estimate to return to it's previous level.
I'll leave Bill to take care of the rest of your post, as he or one of his many clones :-) (he seems to be everywhere, all the time) has been doing a terrific job with you, as they do with everyone else.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Now that SETI is back up and (mostly) sending work, the best bet is to unsuspend it, allow new work, and just let the scheduler work it's way out of any debt situation. You may find that Einstein work doesn't get done for a few days; that's okay, it will then do a chunk of Einstein, before deadlines become an issue. It'll take several days, maybe a week or two, to get back to "normal".
I wouldn't worry about longer _estimated_ Einstein times until things _are_ back to normal. Longer _actual_ times, that go on for more than a day or so, I'd start checking to see if your PC is overheating or something.
I'm finding that whenever I reboot my computer and restart BOINC the progress tab for Einstein resets to 0% and the time to completion reverts to its "start" value. And the LTD for Einstein climbs each time BOINC is restarted. Is there a connection here? It seems that Einstein isn't writing its current progress at each checkpoint, then when it is time for it to download a new WU, the LTD has climbed too high.
I am running the other projects shown in my sig, and the the LTD for SETI has decreased, although I don't think this has to do with the recent troubles at Seti. It seems more like it's balancing the over-inflated LTD of Einstein.
[edit]Not running CPDN at the moment[/edit]
Just a shot in the dark, but what is the setting for "write to disk..." in general preferences? I think we'd like to be able to eliminate this as a possible factor. Also, are you getting any error messages, possibly involving write permissions? It sounds as though it is not writing checkpoints, or is corrupting them, or is ignoring them on restart. This is not an area in which I have much knowledge, so I am acting as "early responder" here, to get information that may be used by someone more expert in this area.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Hi Michael,
I've got "write to disk" set at 540 seconds.
HTH
Alan
Alan,
That's rather a long interval, but it shouldn't be a factor in losing all progress on a WU, unless it's shutting down more often than every 9 minutes. You might want to keep in mind, though, that what it is writing to disk in this case is not cumulative, i.e., setting a shorter interval does not use/waste more drive space. It is setting a "fallback" position, to which it is supposed to pick up where it left off, in case of shutdown or computation error, and is overwritten time and again, at the interval you specify in preferences. Recommend setting it at somewhere from 10 to 60 seconds, to minimize work loss upon error. Statistically, you will lose an average of 50% of that setting in progress each time you shut down. Also, same applies when Boinc switches projects, if you do not have "leave in memory..." set to "yes", though I presume you do have that set to "yes" if you run CPDN.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Just a few more thoughts on this.
I'm not losing any progress on the WU's - it's just the "progress" shows as zero and time to completion seems to be reset. It was just that the LTD seems to go more negative than it should, and I was wondering if these were connected.
It must be writing checkpoints, as it picks up at the correct point in the WU when it is Einstein's turn to run. But it's not picking up this info when Boinc starts and Einstein starts as Preempted.
[edit]
Einstein has just started up again and the "Progress" and "Time to completion" are back to what they should be. Also the LTD has gone from ca -3700 to -2396, so this time it looks as though it will download new work when the current WU is nearly finished. My "connect to" is set at 0.25 days.
Alan
Just a few more thoughts on this.
I'm not losing any progress on the WU's - it's just the "progress" shows as zero and time to completion seems to be reset. It was just that the LTD seems to go more negative than it should, and I was wondering if these were connected.
It must be writing checkpoints, as it picks up at the correct point in the WU when it is Einstein's turn to run. But it's not picking up this info when Boinc starts and Einstein starts as Preempted.
[edit]
Einstein has just started up again and the "Progress" and "Time to completion" are back to what they should be. Also the LTD has gone from ca -3700 to -2396, so this time it looks as though it will download new work when the current WU is nearly finished. My "connect to" is set at 0.25 days.
Alan
Alan,
OK, it appears that your LTD situation is improving. Boinc does a fine job of balancing worktime between projects, especially when we don't "muck about" with it in large doses. I understand what you were saying is that Einstein only shows 0% progress while it pre-empted. Is that the way the other project acts while it is pre-empted also? I am one of the minority here who run only Einstein, so I have no personal knowledge of what is a proper display, and for the same reason, I haven't become that familiar with LTD, EDF, etc. Maybe someone who can better address those topics will step in.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
I have never monitored LTD very closely (always just trusted BOINC to do the right thing) but I have noticed the "Progress" and "Time to completion" issues after restarting. But, the bogus numbers always return to what they should be as soon as BOINC gets around to processing the WU again. (I have always just dismissed it as a minor initialization bug and never really "looked" at what was happening.) Furthermore, since I only run Einstein and Seti, when I restart, I have at most one WU with the bogus numbers anyway. Are you saying the bogus numbers only occur with Einstein units or with all projects? (Because, I think I have seen it with Seti units, too - just not sure.) If it's an "all projects" issue, then maybe it should be reported on the BOINC boards.
RE: .... Also I noticed E@H
)
R.L.,
If your benchmark total has dropped significantly below the previous marks, that could account for longer estimated time, but I see that the last 2 Einstein results you've turned in have been about 1.4 hours longer crunchtime than the balance of your work, so the Duration Correction Factor has bumped the estimate of To Completion time accordingly. This scales upward very quickly, the purpose of which is to prevent Boinc from taking on more work than it can handle. If your rig returns to crunching WUs in the 29K range, then the DCF will adjust downward, albeit incrementally. It may take 12-15 consecutive WUs crunched at the 29K range for the To Completion estimate to return to it's previous level.
I'll leave Bill to take care of the rest of your post, as he or one of his many clones :-) (he seems to be everywhere, all the time) has been doing a terrific job with you, as they do with everyone else.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
That's what I thought! Now I
)
That's what I thought! Now I know. I may be slowly learning something
Thanks
Now that SETI is back up and
)
Now that SETI is back up and (mostly) sending work, the best bet is to unsuspend it, allow new work, and just let the scheduler work it's way out of any debt situation. You may find that Einstein work doesn't get done for a few days; that's okay, it will then do a chunk of Einstein, before deadlines become an issue. It'll take several days, maybe a week or two, to get back to "normal".
I wouldn't worry about longer _estimated_ Einstein times until things _are_ back to normal. Longer _actual_ times, that go on for more than a day or so, I'd start checking to see if your PC is overheating or something.
I'm finding that whenever I
)
I'm finding that whenever I reboot my computer and restart BOINC the progress tab for Einstein resets to 0% and the time to completion reverts to its "start" value. And the LTD for Einstein climbs each time BOINC is restarted. Is there a connection here? It seems that Einstein isn't writing its current progress at each checkpoint, then when it is time for it to download a new WU, the LTD has climbed too high.
I am running the other projects shown in my sig, and the the LTD for SETI has decreased, although I don't think this has to do with the recent troubles at Seti. It seems more like it's balancing the over-inflated LTD of Einstein.
[edit]Not running CPDN at the moment[/edit]
Red Squirrel, Just a shot
)
Red Squirrel,
Just a shot in the dark, but what is the setting for "write to disk..." in general preferences? I think we'd like to be able to eliminate this as a possible factor. Also, are you getting any error messages, possibly involving write permissions? It sounds as though it is not writing checkpoints, or is corrupting them, or is ignoring them on restart. This is not an area in which I have much knowledge, so I am acting as "early responder" here, to get information that may be used by someone more expert in this area.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Hi Michael, I've got "write
)
Hi Michael,
I've got "write to disk" set at 540 seconds.
[edit] Nearly forgot - no error messages.
HTH
Alan
RE: Hi Michael, I've got
)
Alan,
That's rather a long interval, but it shouldn't be a factor in losing all progress on a WU, unless it's shutting down more often than every 9 minutes. You might want to keep in mind, though, that what it is writing to disk in this case is not cumulative, i.e., setting a shorter interval does not use/waste more drive space. It is setting a "fallback" position, to which it is supposed to pick up where it left off, in case of shutdown or computation error, and is overwritten time and again, at the interval you specify in preferences. Recommend setting it at somewhere from 10 to 60 seconds, to minimize work loss upon error. Statistically, you will lose an average of 50% of that setting in progress each time you shut down. Also, same applies when Boinc switches projects, if you do not have "leave in memory..." set to "yes", though I presume you do have that set to "yes" if you run CPDN.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Just a few more thoughts on
)
Just a few more thoughts on this.
I'm not losing any progress on the WU's - it's just the "progress" shows as zero and time to completion seems to be reset. It was just that the LTD seems to go more negative than it should, and I was wondering if these were connected.
It must be writing checkpoints, as it picks up at the correct point in the WU when it is Einstein's turn to run. But it's not picking up this info when Boinc starts and Einstein starts as Preempted.
[edit]
Einstein has just started up again and the "Progress" and "Time to completion" are back to what they should be. Also the LTD has gone from ca -3700 to -2396, so this time it looks as though it will download new work when the current WU is nearly finished. My "connect to" is set at 0.25 days.
Alan
RE: Just a few more
)
Alan,
OK, it appears that your LTD situation is improving. Boinc does a fine job of balancing worktime between projects, especially when we don't "muck about" with it in large doses. I understand what you were saying is that Einstein only shows 0% progress while it pre-empted. Is that the way the other project acts while it is pre-empted also? I am one of the minority here who run only Einstein, so I have no personal knowledge of what is a proper display, and for the same reason, I haven't become that familiar with LTD, EDF, etc. Maybe someone who can better address those topics will step in.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Alan, I have never
)
Alan,
I have never monitored LTD very closely (always just trusted BOINC to do the right thing) but I have noticed the "Progress" and "Time to completion" issues after restarting. But, the bogus numbers always return to what they should be as soon as BOINC gets around to processing the WU again. (I have always just dismissed it as a minor initialization bug and never really "looked" at what was happening.) Furthermore, since I only run Einstein and Seti, when I restart, I have at most one WU with the bogus numbers anyway. Are you saying the bogus numbers only occur with Einstein units or with all projects? (Because, I think I have seen it with Seti units, too - just not sure.) If it's an "all projects" issue, then maybe it should be reported on the BOINC boards.
Stick