Messages

Thanks Fred, but I hope this change does not cause problems with users who like the current way you are sorting and rely on it.

Clicking on the sort direction for status when projects is selected can put all of the active projects at the top.  ie: all the status that is BLANK are at the top and others under them.



However, other status are not in alphabetical order and I cannot seem to make any sense of their position in the column (plotted horizontal to save space above).  Note that "projects" are selected, not tasks or computers, etc.

Is there any reason why "status" events are not in alphabetical order when projects are selected?  I tried holding down CTRL then SHIFT when clicking the sort direction but there was no difference in sorting.

Why do I ask this question?   I wanted to stop all new work so that all tasks complete before I got out of town for a week or two.  Overnight or so all the tasks would have completed and I could power off all my systems before leaving.  When I come back I wanted to power them all up and then remove the "no new tasks".  However, I got a bunch of suspended and no new tasks projects already scattered about my computers and I cant tell the new ones from the old ones.

   AFAICT there is no way to tag the status or use a checkbox ie: "select" as in a database table.  So what I tried was I sorted the status with all the blank and deferred at the top.    I then suspended all the line item projects underneath the last "Deferred" project and marked all the ones above using "no new tasks".

  I thought that when I got back I would power all them up, click the status sort (projects selected of course) and then all the ones that had only "no new tasks" I could then allow new tasking.

Unfortunately, the sort mixes up the ones suspended with the no new tasks as shown below



If the sort was alphabetic then all the "no new tasks" would be grouped together and I could easily allow new tasks.

Maybe there is a different way to do what I want.

Thanks!

I was not aware of that view.

In debugging a problem with the GPUGRID project, I rebooted a boinc 24/7 system and the elapsed time decreased by 1/2 as shown.

before rebooting as suggested from this thread.



after the reboot ---



It seems I had a stuck task and attempted to reboot to clear it.  Why did the elapsed time go from 2d6h to only 12h.  I am using boinc 7.03, maybe that is the problem?

It would be nice if boinctasks could show if the gpu or the cpu is being throttled or had been throttled recently.  Boinctasks could show the temp in red if it was actively being throttled and yellow if it had been throttled at one time or another since the last reboot.

Thanks!

I found what causes this problem to occur and thought I would document it here.  I sometimes turn the monitors off at night but always leave the system running.  Last night a windows auto update caused a reboot and the left monitor became the primary one and the right monitor the secondary.  I noticed this when I turned the monitors back on and I assume this is because the system could not interrogate the pair as they were off and selected the default of left to right after the auto reboot.  I brought up your gadget and dismissed it and then restarted the system.  The reboot put the monitors back with right to left which is what I use and the gadget worked correctly.  I suspect that if I had not brought up the gadget when the monitors were reversed the gadget woudl have been stuck in an off-screen position after the second reboot.

It appears that the gadget popup occures in the window where it last popped up.  If I switch monitors one and two (eg: nvidia control panel) and then hold the cursor over the icon, the gadget pops up but is not visible.  I have to switch 2 and 1 back, wait for the gadget to popup and then move it back to the #2 monitor.  I can then switch 1 and 2 and the gadget pops up correctly.  This can be solved (at a minimum) by always popping up the gadget in the same window (monitor) as the main boinctask program.

BT is not the only program that has this problem, a dialog box in Nero also has this behavior as well as a Casio disk title printer popup.  If the popups were visible there would be no problem.  Unaccountably, it is NOT visible and <ALT><space><M>{cursor} cannot be used to move the hidden gadget, as the BT gadget does not have a (M)ove menu unlike nero or casio.

Best Regards!

Occassionally one of my systems gets the xml account file corrupted.  This is very rare.  When it happens there is no indication of any error from boinctasks.  Normally the message log (called event log in BM) shows the connection to the account manager (BAM! for me) was made.  When making a synchronization using Boinc Manager, a failure to connect pops up the username / password dialog box and a message in the event log shows "wrong password or username" or something to that effect.

When doing this synchronization from BoincTasks, no message shows up when the connection fails.  I only see a message if the connection succeeds and I have to view messages on each system to confirm the synchronization succeeded.

I can always go to boinc tasks to check the date and time of the last connection.  Obviously, if the last connect has been a long time, there is probably something wrong.  AFAICT there is no mechanism at BoincStats (BAM! account manager) to notify me if a system fails to synchronize after x amount of time.

I had been using V2.2.  When I upgraded my core 2 duo (E6400) to core 2 quad (Q6700) I did not realize that chip ran significently hotter and all cpu cores were way above the temp threshold.  I ran at least two weeks with the desktop systems CPUs being heavily throttled before I realized the problem as discussed here .

This same system also had a gtx280 that was being throttled unncessarily.  I returned an ATI Cypress 5850 for warranty and temporarily put a gtx280 in its place.  That card runs hotter then the Cypress and it was also throttled, but not anywhere as much as the CPUs were.

I only use throttle to report temperatures on all my desktop systems (BOINC farm) which are adaquately cooled except when they get dustbugs, and there is no need to actually throttle anything.

Suggestion:  Pop up a caution box when throttle detects a CPU or Graphics Accelerator change to alert the user after a hardware change.

best regards!

I just had to stop all work on the project primaboinca due to they way they are handleing their checkpointing or lack thereof.  I did not see any way to do this all at once for all of my systems.  Is there any way I could have highlighted all of computers running the project 'primaboinca" and do a "no more tasks" with one clck?

I use boincstats as a project manager.  They have a similar, but worse problem, when doing the same thing.

Team names having an "&"  that were created some time ago have improperly escaped the ampersand.  yoyo and cosmology are two that I know of.  When the team names appear in your display they are improperly displayed because the escape code was not properly constructed and only the team founder can fix it (?this is a guess?)  Anyway, this is what your display shows.  What is supposed to show is A&M University.



This is pretty low in priority, and, at least for me, I have deselected the team name column altogether. 

BT not always allowing BOINC to start.  Boinc gives error message "another instance of BOINC is running" and quits.  However, there is no other instance of BOINC running according to Windows 7 task manager.  If I stop BT using the "Exit" then BOINC can be started.  I am guessing that BOINC tries to use port 31416 and when it cant, it then assumes a copy of itself is running

To duplicate this problem BT must have " [X] Start BOINC at login" checked and be iconized.  It can sometimes take 3 tries before the failure shows up.  Once the failure show up, BOINC will never start unless BT exits.

This is what it looks like when the failure occures



The following image was taken immediately after entering the RETURN key on the command line.  Note that BOINC is running



This image was taken a few seconds later and shows the error message.  Note that BOINC is no longer in the task manager



This images was taken after terminating BT (note it is no longer in the task manager, and BOINC is running fine




The purpose of my stopping boinc was to allow CD and DVD burning to proceed at full speed on the few occassions when there is a heavy load on my system.  Maybe there is another way to pause all tasks so multiple DVD burns can run at full speed.

What is the purpose of the Task "Count" in the Projects Tab?

For example, Milkyway is listed in three places under the "Tasks Tab" as ..
Ready to report: 0.05c + 1ATI ..and..
Running: 0.05c + 1ATI ..and..
20[Tasks]: Ready to start

however, under the "Projects Tab" I see only
Tasks:  0/21

It should be 0/22 or better yet:  1/21
or even better to show all 4 possible states
(done/waiting/running/ready) : 1/0/1/20

The tasks counts are either  0/n or they are n/0 (or 0/0)  ie: the zero is either before or after the non-zero number and a tool tip for the Tasks column header to explain the difference would be nice.

Thanks for looking!

<SNIP>
The location:C:\Users\username\AppData\Roaming\eFMer\BoincTasks\crash
you can mail me at fred at efmer dot com
If possible one of a recent BT copy 0.78 or 0.79.
What BOINC version are you using?

OMG!!  There are 95 dump files there!!  from 85mb thru 142mb in size.

I will zip up one each from 79 and 78 (Both are from today).   Lemme know if the zip files arrives ok.  My ISP may not let it thru.
I am running boinc 6.10.65 and 6.10.66 on 3 windows and 5 ubuntu systems.

I am getting crashes sometimes minutes apart.  ie:  BT crashes and asks if I want to send the dump file to you.  I close it, start BT back up and about 5 - 50 minutes later it crashes again.  This also happened in .78 or maybe .76, as I may have skipped an update or two.  About a week ago I had maybe 50 BT icons in the task bar, all but 1 were dead.  That has not happened recently but I did turn off the silent restart.

Do you want the .dmp files e-mailed?  I can also ftp them to a web site I got and just email you the link.  Possibly you already have more .dmp files then you need.

The last crash occured just a few minutes ago and before I would write down the location of the dmp file the message disappeared off the screen.  There was time to make a note of the location but I didnt do it right away. 

I am currently debugging my own app under VS2008 and turned off BT temporarily to make my own debugging easier.   About a month ago, I also stopped  reporting temperature info back to BT from my Linux systems so that cannot be causing the problem.