As posted by a number of other participants, I have had a number of WUs go to 100%, and new WU download, with no increase to my credits. As also mentioned, the credits in them selves are not the issue, but ... I have been a SETI 'clasic' participant for a number of years with about 14,000 hours of credit - it is a feedback that your are doing something usefull. Since the BOINC version is new for me, I do not know if 1) There is an installation problem, so I have reinstalled 3 times -- with the resultant waste of WU processing; 2) configuration problem, so I have been tinkering with the settings -- with the resultant waste of my time; 3) BOINC, or SETI problem, so I have to spend time on these message boards with their own set of problems ( the Search function does not work on some subset of users - includes me). I am considering going back to the classic version.

I do not mind donating my spare cycles; nor do I mind helping to debug new and complex software. I do mind wasting my cycles and time.

In addition, I find the layout of the BOINC web site very difficult to use.

---

...don

Your WUs are credited when the server has 3 results that match.

---

<img src="http://www.boincstats.com/stats/banner.php?cpid=f098ec1961b5cdf855cbac04a444071f">

Looking at your results pages I see that you have not successfully returned any results. Most of them say "client error". Try completely uninstalling BOINC, make sure all of you current BOINC files are deleted and then reinstall BOINC.

I have also been a SETI Classic user for a number of years but I like BOINC much better now that it is working properly. I like being able to see if the results I return are valid. With classic there was no way to tell.

---

> Looking at your results pages I see that you have not successfully returned
> any results. Most of them say "client error". Try completely uninstalling
> BOINC, make sure all of you current BOINC files are deleted and then reinstall
> BOINC.
>
> I have also been a SETI Classic user for a number of years but I like BOINC
> much better now that it is working properly. I like being able to see if the
> results I return are valid. With classic there was no way to tell.
>

Don, are you running BOINC with the same userid that you installed it with? The errors suggest directory or file permissions problems with BOINC and SETI not able to write files needed to run properly.

> Don, are you running BOINC with the same userid that you installed it with?
> The errors suggest directory or file permissions problems with BOINC and SETI
> not able to write files needed to run properly.
Yes, I am using the same ID.

---

...don

> Looking at your results pages I see that you have not successfully returned
> any results. Most of them say "client error". Try completely uninstalling
> BOINC, make sure all of you current BOINC files are deleted and then reinstall
> BOINC.
>

What 'results' page are you looking at?

I have already uninstalled/reinstalled twice; I would much rather debug the "client error" than shotgun the installation.

---

...don

> Your WUs are credited when the server has 3 results that match.


Is that true? Perhaps on some level that makes sense, but to me, it does not. If the results match or not, I still spent the resources to process them.

---

...don

> Most of them say "client error".

Many of the errors are the failure to create a file C:\Documents and Settings\Matt Lebofsky\...

1) I do not have a C:\Documents and Settings\ ... ANYTHING. My systems is installed on G:

2) I have not clue who Matt Lebosfsky is, and am most certain that he has never been near my computer.


To me this seems to be a BOINC/SETI issue, not my 'client'.

---

...don

> > Most of them say "client error".
>
> Many of the errors are the failure to create a file C:Documents and
> SettingsMatt Lebofsky...
>
> 1) I do not have a C:Documents and Settings ... ANYTHING. My systems is
> installed on G:
>
> 2) I have not clue who Matt Lebosfsky is, and am most certain that he has
> never been near my computer.
>
>
> To me this seems to be a BOINC/SETI issue, not my 'client'.
>

No it doesn't say it was trying to create a file C:Documents and settings(slash)Settings(slash)Matt Lebofsky.....

It says it failed to create a file, and then lists the procedure that got the error, the source file the procedure is located in and the line number of the statement that issued the error message.

Moost of the results for your machine are "Couldn't start the app for this result: error -108". Do you also see a "could not create file" message? If so, post the whole message, including the function, source code file and line number. Might be able to get a little more info as to whats wrong.

Edit: removed suggestion to update from 4.09 to 4.13. Saw a recent result where you did do that.

> ...
> 2) I have not clue who Matt Lebosfsky is, and am most certain that he has
> never been near my computer.
> ...

Stripchart version 2.0
----------------------
Author: Matt Lebofsky
BOINC/SETI@home - University of California, Berkeley

( from BOINC-source-root/stripchart/readme )

>
> > Don, are you running BOINC with the same userid that you installed it
> with?
> > The errors suggest directory or file permissions problems with BOINC and
> SETI
> > not able to write files needed to run properly.
> Yes, I am using the same ID.
>

In a different message you said BOINC is on G:. Is that local or remote? That is, is it a hard drive partition on a drive physically attached to your hard drive controller? Or is it attached some other way, like on another PC and accessed over the network, or an externak USB/Firewire drive plugged into your system?

Perhaps you could run the computer management tool, disk management, and tell us how G: is set up. Properties for the drive (right click the drive icon in the lower pane) will show the device type, what is it? The graphic view of the drive shows the partitions on the drive, what are they and what format are they in?

That way we're more likely to figure out why you're getting those file create errors.

I looked up the source code for the latest error you got, shows it failed to open (for output) "state.sah" for in the slot(slash)0 directory. And it was while trying to take a checkpoint, which gets done every minute.

Anything you can think of that would cause that open to fail?

> I looked up the source code for the latest error you got, shows it failed to
> open (for output) "state.sah" for in the slot(slash)0 directory. And it was
> while trying to take a checkpoint, which gets done every minute.
>
> Anything you can think of that would cause that open to fail?


Maybe some BOINC add-ons open it for their stats.

BOINC opens those files exclusive, it fails if any other program has them opened (even if it's just for reading) just at that moment.

> > I looked up the source code for the latest error you got, shows it failed
> to
> > open (for output) "state.sah" for in the slot(slash)0 directory. And it
> was
> > while trying to take a checkpoint, which gets done every minute.
> >
> > Anything you can think of that would cause that open to fail?
>
>
> Maybe some BOINC add-ons open it for their stats.
>
> BOINC opens those files exclusive, it fails if any other program has them
> opened (even if it's just for reading) just at that moment.
>

_All_ of Don's workunits fail in some fashion related to writing to disk, either they can't get started or they start but can't continue. That suggests something unique to his system. Now its a matter of finding out what.

A few things to check :

- BOINC installed on a network share on a computer that runs S@H or BOINC too

- _very_ long BOINC base pathname or character in pathname that shouldn't be used

- 4.13 not fresh installed but installed over 4.12 (can take over damaged settings then - e.g. venue stuff)

- conflict with virus scanner (exclude BOINC directory for test)

- scandisk needed on FAT file system

- setting for disk_interval not OK ("Write to disk at most every xx")


I wonder if state.sah has already been there when result 13283957 crashed, what it looked like and which file attributes it had - after 3570 seconds it should already have existed.

> I wonder if state.sah has already been there when result 13283957 crashed,
> what it looked like and which file attributes it had - after 3570 seconds it
> should already have existed.
>

State.sah gets created when seti starts, and its rewritten every checkpoint - once a minute by default. The code shows an open for write, which turns into (by Windows) a delete file followed by a create file. So it worked almost 60 times, assuming the "write to disk" setting hasn't been changed.

But you're right, would be nice to get _all_ the output and files from when the program crashed. It would also be nice to get the extended error info from the fopen call that failed, but the program only issues a error message when it fails. Too bad that stuff isn't saved.

Would also be nice to get specifics on his G: drive, and disk/data management software he has.

> > I looked up the source code for the latest error you got, shows it failed
> to
> > open (for output) "state.sah" for in the slot(slash)0 directory. And it
> was
> > while trying to take a checkpoint, which gets done every minute.
> >
> > Anything you can think of that would cause that open to fail?
>

Yes, as mentioned, there is no such directiory. C: is a very small boot partition. It is not used for any operating system or application purposes.

---

...don

> In a different message you said BOINC is on G:. Is that local or remote?
> That is, is it a hard drive partition on a drive physically attached to your
> hard drive controller? Or is it attached some other way, like on another PC
> and accessed over the network, or an externak USB/Firewire drive plugged into
> your system?
>

Yes, G: is a local drive formatted NTFS. BOINC is installed on D:, also a local NTFS drive.

---

...don

> > I looked up the source code for the latest error you got, shows it failed
> to
> > open (for output) "state.sah" for in the slot(slash)0 directory. And it
> was
> > while trying to take a checkpoint, which gets done every minute.
> >
> > Anything you can think of that would cause that open to fail?
>
>
> Maybe some BOINC add-ons open it for their stats.
>
> BOINC opens those files exclusive, it fails if any other program has them
> opened (even if it's just for reading) just at that moment.
>

---

...don

> BOINC opens those files exclusive, it fails if any other program has them
> opened (even if it's just for reading) just at that moment.
>

Possibility. There is a daily backup process that archives any changed files on the disk BOINC is installed on. There is also a weekly trash collection process.

BTW, that is one of the reasons that I requested an easy way to separate the application and data locations. My other applications write to a 'data' partition where special precations are taken to not disturb open files. OS, code, and configuration information, however, are backed up so that a failed system can be quickly recovered.

---

...don