Hi all,
still I got the fix about sorting, I use only sorting with two columns, but every time I run BT "Tasks" and "Transfers" tab are sorted by means of three columns. ??
I've noticed this problem today on BT 0.41 and as there is no related change mentioned in the change list, I'm reporting it here although 0.42 (currently) does not seem to manifest the problem.
---
The Lattice Project - Garli 5.13, task 284462300.6323525377828851.2_0
Since 2-3 minutes from its beginning, I've noticed that in Boinc Tasks this task's progress seems to be stuck %-wise - now after elapsed = 22:49:08 (20:52:42), CPU% = 84.56, Progress = 0.992%, Remaining time = 94d,23:37:51, deadline in a week.
OTOH, Boinc Manager showed the task as progressing since its beginning, now (a minute or two later) Elapsed = 22:51:56, Progress = 25.924%, To completion = 2279:39:23 (similar 94.5 days).
The difference is the Progress %, it was nearly always that low in BoincTasks.
Details from the client_state.xml:
<active_task>
<project_master_url>http://boinc.umiacs.umd.edu/</project_master_url>
<result_name>284462300.6323525377828851.2_0</result_name>
<active_task_state>1</active_task_state>
<app_version_num>513</app_version_num>
<slot>15</slot>
<pid>488</pid>
<checkpoint_cpu_time>75285.190000</checkpoint_cpu_time>
<checkpoint_elapsed_time>82302.245195</checkpoint_elapsed_time>
<fraction_done>0.259241</fraction_done>
<current_cpu_time>75433.770000</current_cpu_time>
<swap_size>1019482112.000000</swap_size>
<working_set_size>616804352.000000</working_set_size>
<working_set_size_smoothed>616804352.000000</working_set_size_smoothed>
<page_fault_rate>0.000000</page_fault_rate>
</active_task>
Info obtained using the boinccmd.exe:
31) -----------
name: 284462300.6323525377828851.2_0
WU name: 284462300.6323525377828851.2
project URL: http://boinc.umiacs.umd.edu/
report deadline: Wed Mar 03 21:37:16 2010
ready to report: no
got server ack: no
final CPU time: 0.000000
state: 2
scheduler state: 2
exit_status: 0
signal: 0
suspended via GUI: no
active_task_state: 1
stderr_out:
app version num: 513
checkpoint CPU time: 75888.410000
current CPU time: 76335.410000
fraction done: 0.259241
swap size: 1019482112.000000
working set size: 616808448.000000
estimated CPU time remaining: 8207566.865642
supports graphics: no
...
Later at 29 hours, the progress is finally "non-zero" at 32.379%. I can't tell whether it has always been below 1% or has ben jumping high and low.
...
Yes, definitely jumpin' around. At 30:17:13 and 33.372%, Boinc Tasks is again showing 0.769% ... 0.770% ... 0.805% ... 0.822 slowly rising, but got suspended for now.
Interestingly, at the same time the mini preview gadget window is displaying the progress% correctly.
Few hours later, few seconds prior to upgrade to BT 0.42, the Progress rising towards 1.0%, just after the upgrade the 0.42 is showing 38.901%. But it also could just have been one of these jumps...
For Pirates@Home, Boinc Manager shows in the Project tab: "Scheduler request pending, Requested by project, Scheduler request in progress", whereas Boinc Tasks (sorry, it was the 0.41) was displaying a non-explanatory "sRPCp 7".
Quote from: Corsair on February 22, 2010, 01:01:09 PM
Hi all,
still I got the fix about sorting, I use only sorting with two columns, but every time I run BT "Tasks" and "Transfers" tab are sorted by means of three columns. ??
Which of the 3 columns are sorted?
What column is sorted incorrectly?
Give me the following info.
HKEY_CURRENT_USER software efmer Boinctasks sorting and give me the numbers for:
tasks_p tasks_s tasks_t (the 3 sorting columns)
Quote from: Pepo on February 22, 2010, 01:39:04 PM
For Pirates@Home, Boinc Manager shows in the Project tab: "Scheduler request pending, Requested by project, Scheduler request in progress", whereas Boinc Tasks (sorry, it was the 0.41) was displaying a non-explanatory "sRPCp 7".
Is in the bug list.
Quote from: Pepo on February 22, 2010, 01:37:06 PM
...
Later at 29 hours, the progress is finally "non-zero" at 32.379%. I can't tell whether it has always been below 1% or has ben jumping high and low.
...
Yes, definitely jumpin' around. At 30:17:13 and 33.372%, Boinc Tasks is again showing 0.769% ... 0.770% ... 0.805% ... 0.822 slowly rising, but got suspended for now.
Interestingly, at the same time the mini preview gadget window is displaying the progress% correctly.
Few hours later, few seconds prior to upgrade to BT 0.42, the Progress rising towards 1.0%, just after the upgrade the 0.42 is showing 38.901%. But it also could just have been one of these jumps...
Added to the bug list: Don't use alternative progress % when the default % is above 0%.
Quote from: fred on February 23, 2010, 01:48:28 AM
Quote from: Pepo on February 22, 2010, 01:37:06 PM
...
Later at 29 hours, the progress is finally "non-zero" at 32.379%. I can't tell whether it has always been below 1% or has ben jumping high and low.
...
Yes, definitely jumpin' around. At 30:17:13 and 33.372%, Boinc Tasks is again showing 0.769% ... 0.770% ... 0.805% ... 0.822 slowly rising, but got suspended for now.
Interestingly, at the same time the mini preview gadget window is displaying the progress% correctly.
Few hours later, few seconds prior to upgrade to BT 0.42, the Progress rising towards 1.0%, just after the upgrade the 0.42 is showing 38.901%. But it also could just have been one of these jumps...
Added to the bug list: Don't use alternative progress % when the default % is above 0%.
Unfortunately I do not understand the issue :) never mind. (Or, is it for cases when/while the application does not report any % progress at all?)
Since the upgrade to 0.42, I've never noticed a progress under 1% - it was continuously moving through 40-50-60-70%.
Occasionally it happens in the Transfers, that more data is actually reported as transferred than the net size (be it the overhead or any error), Boinc Manager displays it like (now in one of my Rosetta uploads) "100% 80.39/79.79KB". For the same upload, BT is displaying "100.755% 81K", the blue background seems to match the progress rectangle. (I've a snapshot picture around.) (* Could we get the "so far transferred data" with resolution to at least 3 decimal places, like 80.4 K or 365 K or 6.72 M?) (* Could we get displayed also the amount "to be transferred"?)
But I've also noticed one transfer to Spinhenge, not noticed the bytes, but the final progress was 111.11% and the blue background was way over the progress' rectangle - touching the number of transferred data in the adjacent column. I guess that the %-tage could freely go over 100% (if much more was transferred than planned), but the colored background should not leave the rectangle...
Quote from: Pepo on February 23, 2010, 12:09:51 PM
Occasionally it happens in the Transfers, that more data is actually reported as transferred than the net size (be it the overhead or any error), Boinc Manager displays it like (now in one of my Rosetta uploads) "100% 80.39/79.79KB". For the same upload, BT is displaying "100.755% 81K", the blue background seems to match the progress rectangle. (I've a snapshot picture around.) (* Could we get the "so far transferred data" with resolution to at least 3 decimal places, like 80.4 K or 365 K or 6.72 M?) (* Could we get displayed also the amount "to be transferred"?)
But I've also noticed one transfer to Spinhenge, not noticed the bytes, but the final progress was 111.11% and the blue background was way over the progress' rectangle - touching the number of transferred data in the adjacent column. I guess that the %-tage could freely go over 100% (if much more was transferred than planned), but the colored background should not leave the rectangle...
I will limit the % to 100% as the rest is probably overhead or retries. It happens more frequently when the connection is busy.
Quote from: Pepo on February 23, 2010, 11:57:52 AM
Quote from: fred on February 23, 2010, 01:48:28 AM
Quote from: Pepo on February 22, 2010, 01:37:06 PM
...
Later at 29 hours, the progress is finally "non-zero" at 32.379%. I can't tell whether it has always been below 1% or has ben jumping high and low.
...
Yes, definitely jumpin' around. At 30:17:13 and 33.372%, Boinc Tasks is again showing 0.769% ... 0.770% ... 0.805% ... 0.822 slowly rising, but got suspended for now.
Interestingly, at the same time the mini preview gadget window is displaying the progress% correctly.
Few hours later, few seconds prior to upgrade to BT 0.42, the Progress rising towards 1.0%, just after the upgrade the 0.42 is showing 38.901%. But it also could just have been one of these jumps...
Added to the bug list: Don't use alternative progress % when the default % is above 0%.
Unfortunately I do not understand the issue :) never mind. (Or, is it for cases when/while the application does not report any % progress at all?)
Since the upgrade to 0.42, I've never noticed a progress under 1% - it was continuously moving through 40-50-60-70%.
I removed the bug report, when you notice it again let me know.
Quote from: fred on February 23, 2010, 12:55:17 AM
Quote from: Corsair on February 22, 2010, 01:01:09 PM
Hi all,
still I got the fix about sorting, I use only sorting with two columns, but every time I run BT "Tasks" and "Transfers" tab are sorted by means of three columns. ??
Which of the 3 columns are sorted?
What column is sorted incorrectly?
Give me the following info.
HKEY_CURRENT_USER software efmer Boinctasks sorting and give me the numbers for:
tasks_p tasks_s tasks_t (the 3 sorting columns)
I want sorting only by two columns, not by three.
tasks_p 517
tasks_s 267
tasks_t 521
Quote from: Corsair on February 26, 2010, 08:33:34 PM
Quote from: fred on February 23, 2010, 12:55:17 AM
Quote from: Corsair on February 22, 2010, 01:01:09 PM
Hi all,
still I got the fix about sorting, I use only sorting with two columns, but every time I run BT "Tasks" and "Transfers" tab are sorted by means of three columns. ??
Which of the 3 columns are sorted?
What column is sorted incorrectly?
Give me the following info.
HKEY_CURRENT_USER software efmer Boinctasks sorting and give me the numbers for:
tasks_p tasks_s tasks_t (the 3 sorting columns)
I want sorting only by two columns, not by three.
tasks_p 517
tasks_s 267
tasks_t 521
I should be able to solve this in the next version.
I've done some extensive hiking over huge boulders, so not today. ;D
Progressbar is out of its box and way over 100%.
(http://img.techpowerup.org/100228/ProgressBar.png)
Quote from: fred on February 23, 2010, 10:19:22 PM
Quote from: Pepo on February 23, 2010, 11:57:52 AM
Quote from: fred on February 23, 2010, 01:48:28 AM
Quote from: Pepo on February 22, 2010, 01:37:06 PM
...
Later at 29 hours, the progress is finally "non-zero" at 32.379%. I can't tell whether it has always been below 1% or has ben jumping high and low.
...
Yes, definitely jumpin' around. At 30:17:13 and 33.372%, Boinc Tasks is again showing 0.769% ... 0.770% ... 0.805% ... 0.822 slowly rising, but got suspended for now.
Interestingly, at the same time the mini preview gadget window is displaying the progress% correctly.
Few hours later, few seconds prior to upgrade to BT 0.42, the Progress rising towards 1.0%, just after the upgrade the 0.42 is showing 38.901%. But it also could just have been one of these jumps...
Added to the bug list: Don't use alternative progress % when the default % is above 0%.
Unfortunately I do not understand the issue :) never mind. (Or, is it for cases when/while the application does not report any % progress at all?)
Since the upgrade to 0.42, I've never noticed a progress under 1% - it was continuously moving through 40-50-60-70%.
I removed the bug report, when you notice it again let me know.
Just to add to the confusion on this issue:
I'm seeing behavior similar to this
only on one of three hosts (quad core/GPU, Vista SP2) with 3 PrimeGrid tasks running in Panic Mode with deadline warning. BT .42 shows progress as a fraction of a percent, climbing slowly, while BM shows the same tasks at a more believable progress (two at 6~7%, the other at 12.x%. (This agrees with the boinc_task_state.xml.) The fourth CPU core has a PrimeGrid task also in Panic Mode (but
no deadline warning) and BT shows progress in agreement with BM.
My other 2 hosts (both dual core/no GPU, Vista SP2), also have PrimeGrid tasks in Panic Mode (but
no deadline warning) and BT shows progress in agreement with BM.
I don't envy you the witch hunt for this one.
--NEW ISSUE--
Actually, I thought I saw "something, somewhere" in the forum on this but can't find it. Tasks running in Panic Mode (see above) continue to show status "Running High P" when suspended (in addition to suspended, user request, whatever). Tasks running in normal scheduler mode do not continue to show as Running. This, at least, is consistent across all tasks/hosts.
Best wishes.
To answer all questions:
I limited the % in transfers to 100%.
Changed the progress % so it should now progress normally. That is; in the first 1% the calculation was done differently.
Now this is only done in the first 0.01%, why because otherwise you can't see GPU tasks progress in the first couple of seconds.
Quote from: jjwhalen on February 28, 2010, 07:02:45 PM
--NEW ISSUE--
Actually, I thought I saw "something, somewhere" in the forum on this but can't find it. Tasks running in Panic Mode (see above) continue to show status "Running High P" when suspended (in addition to suspended, user request, whatever). Tasks running in normal scheduler mode do not continue to show as Running. This, at least, is consistent across all tasks/hosts.
Best wishes.
This is something I can't place.
Did you see this yourself? Where did you see it. The problem may already been solved.
Quote from: efmer on February 28, 2010, 10:21:40 PM
Quote from: jjwhalen on February 28, 2010, 07:02:45 PM
--NEW ISSUE--
Actually, I thought I saw "something, somewhere" in the forum on this but can't find it. Tasks running in Panic Mode (see above) continue to show status "Running High P" when suspended (in addition to suspended, user request, whatever). Tasks running in normal scheduler mode do not continue to show as Running. This, at least, is consistent across all tasks/hosts.
Best wishes.
This is something I can't place.
Did you see this yourself? Where did you see it. The problem may already been solved.
Sorry but yes, I'm reproducing this in v0.42 on demand, including as I type this. With 3 cores running in Panic Mode (with deadline warning), and the client either Snoozed or Suspended, I'm getting a Status column thus: "Running High P., deadline warning,suspended: user request". The 4th core task, in Panic Mode but without deadline warning, reads: "Running High P., suspended: user request".
With tasks running in normal scheduler mode and the client either Snoozed or Suspended, the Status column 'correctly' reads "suspended: user request". This is consistent on 3 hosts. Host environments are as listed earlier.
B/W
Quote from: jjwhalen on March 01, 2010, 12:24:40 AM
Quote from: efmer on February 28, 2010, 10:21:40 PM
Quote from: jjwhalen on February 28, 2010, 07:02:45 PM
--NEW ISSUE--
Actually, I thought I saw "something, somewhere" in the forum on this but can't find it. Tasks running in Panic Mode (see above) continue to show status "Running High P" when suspended (in addition to suspended, user request, whatever). Tasks running in normal scheduler mode do not continue to show as Running. This, at least, is consistent across all tasks/hosts.
Best wishes.
This is something I can't place.
Did you see this yourself? Where did you see it. The problem may already been solved.
Sorry but yes, I'm reproducing this in v0.42 on demand, including as I type this. With 3 cores running in Panic Mode (with deadline warning), and the client either Snoozed or Suspended, I'm getting a Status column thus: "Running High P., deadline warning,suspended: user request". The 4th core task, in Panic Mode but without deadline warning, reads: "Running High P., suspended: user request".
With tasks running in normal scheduler mode and the client either Snoozed or Suspended, the Status column 'correctly' reads "suspended: user request". This is consistent on 3 hosts. Host environments are as listed earlier.
B/W
Thanks, I added it to the Bug list
Quote from: jjwhalen on February 28, 2010, 07:02:45 PM
My other 2 hosts (both dual core/no GPU, Vista SP2), also have PrimeGrid tasks in Panic Mode (but no deadline warning) and BT shows progress in agreement with BM.
The warning is only activated when a task is within the set time period. Tasks may go into Panic Mode long before that. And Panic Mode has it's own color, so it's a warning as well.
Quote from: efmer on March 01, 2010, 12:56:03 AM
Quote from: jjwhalen on February 28, 2010, 07:02:45 PM
My other 2 hosts (both dual core/no GPU, Vista SP2), also have PrimeGrid tasks in Panic Mode (but no deadline warning) and BT shows progress in agreement with BM.
The warning is only activated when a task is within the set time period. Tasks may go into Panic Mode long before that. And Panic Mode has it's own color, so it's a warning as well.
I do know about the 'Deadline warning' option pullldowns in Settings, and have them set appropriately. These hosts had not processed the particular (PrimeGrid) applications before and the 'default' Time Left is ridiculously long (1500 hours +), which is
months beyond the Report Deadline at a 15% Resource share--hence the Panic Mode. This is always a problem, especially with Wrapper applications like PrimeGrid. I know to expect it & it's not an issue. I already know based on Elapsed Time that the actual time to completion will be ~32 hours, and the client will "learn" that.
The only issue I'm requesting you look at here is that tasks in Panic Mode are indicating Status of both Running [High P.] & Suspended at the same time--which is an oxymoron ::) EDIT--(As per the TO DO list :))
B/W