Stops Folding...But

[Insidious]

Since our last protien change, I have noticed the following 3 or 4 times:

[SETUP]

WindowsXP Pro (fully updated)
DF installed as a service
Cable internet conection

service.cfg as follows:

service=1
useram=1
progress=15
uploadfreq=999

[Issue}

I find that DF is not producing. (CPU 0% usage shown by taskmanager)
There are no errors showing in Error.log. There are no System or Application events showing in Computer Management. dfGUI shows that DF is running with the display that would have been the last screen update just prior to an upload. VIAHM shows my CPU temperatures consistant with that of an idling state, which seems to indicate the computer has been in this state for a significant period of time.

Re-booting the machine returns all to normal.

-Sid

[cygnussphere]

does any of this look familiar?
http://www.free-dc.org/forum/showthread.php?s=&threadid=2674
I was seeing the same symptoms that you describe

[Insidious]

yeah, that does sound familiar...

I was troubleshooting a (come to think of it...) failing video card
a couple weeks ago and perhaps something got corrupted..

I am going to try the re-install now....

Thanks cygnusspher!

Sid

[Brian the Fist]

How about doing what I suggested and running it from the command line NOT as a service, to see if any error message shows up that way.

[Insidious]

Originally posted by Brian the Fist
How about doing what I suggested and running it from the command line NOT as a service, to see if any error message shows up that way.
I'll be glad to....
Which switches would you like me to use?
(if you could post the line for me to place in foldit.bat, that would help eliiminate any possibilities of mis-communication)

-Sid

(I thought you were suggesting a re-install in that thread, but this is fine as well.)

[Brian the Fist]

It doesn't matter what switches you use, just see if the thing works or not.

[Insidious]

Two Machines running CLI client, non-service since your (Howard) last post...
no abnormalities to date.

I'll give them one or two more days before going back to service installs, then you're on your own.

[Brian the Fist]

If you return to running it as a service and then have that problem again, do NOT erase the directory and re=install. Instead, stop the service (using a kill from task manager if necessary) and then run it from the command line. You should then see an error message of some sort indicating the problem, and if you do, send that to me and I can tell you what is happening. Thanks

[Insidious]

Will do

-Sid

[Insidious]

OK, I think I have this one better identified. It is a very different "bug" than I believed at the start of this thread.

(After returning DF to running as a service last night)

I have repeated this scenerio 3 times now:


DF is running as a service

I run dfGUI to stop the service (using the stop client button) to free resources for a game, or whatever.

The structures buffered indication of dfGUI immediately goes to zero (???) when the bar turns red and dfGUI tells me the client is apparently stopped (but I didn't attempt to upload!)

{at this point, Task Manager still shows CPU and RAM usage as normal for an operating client. But dfGUI reports the client is apparently stopped and no updates occur to its' display}

Normally, when I finished whatever task I was doing, I would come back to dfGUI and use the Start Client button to resume folding. This is where things went wrong. THIS is when DF stops folding as evidenced by 0 cpu usage (with no errors) If I were to reboot at this point, DF would fold normally on the re-start, but I would have TWO of the ......val.bz2 files in the directory indicating the buffered folds of last time around have been "lost".

I found that if I just ignored the dfGUI display and didn't try to start the client (which had never really stopped) all folding proceeded normally with no leftover .....val.bz2 files in the distribfold directory.

note: The first time I did the above, I tried to run the client in non-service/ASCII mode and look for errors as you asked and the DOS window came up for the looking for updates/handle found part, then the window closed itself with no errors reported on the screen or in error.log (obviously, since the df client was still running, but I didn't realize that) On a subsequent reboot, the service would fold normally, but with the extra ....val.bz2 file indicating the lost buffer.


Soooooooooo, it looks like something has changed in the interraction of dfGUI and the df client. This doesn't happen every time I use the dfGUI stop button, but it is not hard to repeat with a few tries.

looks like Geoff may be interested in this one.

I hope I have tried to explain this scenerio well enough to be helpfull.

Sid

[Digital Parasite]

Hi Sid,

This seems very strange. Can you tell me what version of dfGUI you are using and also post your dfGUI.ini file?

First of all, in dfGUI is the "Client installed as service" checkbox checked?

If yes, when you click on the "Stop" button, all the GUI does is delete the foldtrajlite.lock file which tells the DF client to stop. When dfGUI shows the red bar saying the client isn't running it does this by checking to see if there is a foldtrajlite.lock file in your DF client directory (which there shouldn't be because you just deleted it when you hit STOP).

dfGUI looks at the filelist.txt file to see how many structures you have buffered. If that is showing 0 that means it probably can't read the filelist.txt file.

So can you do a test for me? Get your client running however you did before, then when you hit STOP in dfGUI and it shows the red bar and says 0 structures are buffered, look in your DF client directory and tell me if you can find the foldtrajlite.lock file and the filelist.txt file. If you can find the filelist.txt file, please send it to me so I can see if I can reproduce the problem.

You then said that after you click on START, that is when the DF client seems to stop. How long did you wait after clicking on STOP before you hit the START button again? If the DF client is currently working on a structure I don't think it checks to see if the foldtrajlite.lock file has been deleted until it finishes the structure so it may take a bit of time. Try waiting for a minute and then hitting START again to see if the same thing happens.

PS: Its Jeff not Geoff. ;)

Jeff.

[Insidious]

PS: Its Jeff not Geoff
OOPS :bang:

sorry bout that.

yeah, I'll give it a try. It'll probably be tomorrow evening.

-Sid

[Digital Parasite]

I tried reproducing your bug in the GUI but I can't seem to get it to do what you describe. Any time I hit STOP, within 20 seconds the client stops as expected and the buffered structures remain valid. Then hitting Start again uploads the structures and continues on.

I will really need to see your dfGUI.ini file to get a better idea.

Jeff.

[Insidious]

Howard and Jeff,

I have captured one of these failures.

I wrote the email and attached a rather large .zip file (13MB) that I believe will let you understand fully what I am seeing. (It contains screenies and distribfold folder contents during and after the problem.)

because of the attachment size I did not try to send it through the profile links and I really don't want to re-type the whole thing anyway.

If you will PM me with your addresses I will mail it immediately.
If the file is too large for your email, let me know and I will PM you a URL to retrieve the file.

-Sid

[Insidious]

I went ahead and sent you two e-mails with a URL to the file of the saved info.

-Sid

[Brian the Fist]

Hmm, seems to work fine for me. It sound slike the problem is when you hit the stop button in dfGUI, it is not exiting like it should. This is a problem. I am not sure what dfGUI does when you hit the stop button, maybe Jeff can tell me. But it sounds like its not exiting, thats the problem that needs to be fixed. It should disappear totally from task manager

Instead of stopping with the dfGUI stop button, try going to the folder with the program in it and deleting 'foldtrajlite.lock'. This also stops it. If you can reproduce the problem this way, the bug is likely in my software and not Jeff's.

[Insidious]

The problem is that I am finding the DF client stopped (when it should not be) in the first place.

All the rest is just trying not to loose the folds.

The next time I find it in this state, I will first try to find/delete the .lock file. Unfortunately, if you look in my Pre-boot distribfold directory (the files I gave you) you will see there is NO .lock file

Thanks

-Sid

(1. Is it possible that dfGUI and DF are both trying to access a file at the same time and the clients get "confused"
2. Is it possible that on-line gaming is (Specifically, BF1942) is messing up the attempted file updates?
3. I was kind of hoping that those pre- and post- directories I gave you might indicate something I have wrong in my setup.... no?)

Did the e-mail, files and screen shots I gave you help? Was there more information I should have given?

Anyway, I don't want this whole thing to get blown out of proportion, but if there is anymore I can do to help I will be happy to.

[Brian the Fist]

Originally posted by Insidious
The problem is that I am finding the DF client stopped (when it should not be) in the first place.

All the rest is just trying not to loose the folds.

From what I understand, it is running happpily with DFGUI, as a service, and you say 'Ok, game time,' and you click 'stop' in DFGUI. And while CPU usage goes to zero, it does not really stop. So why do you say the client is stopped when it shouldn't be?? Don't you mean it is NOT stopped when in fact it SHOULD be??? Its very easy to confuse me with stuff like this :confused:

[Insidious]

OK, I see the miscommunication now. As I continued to closely watch DF, I got the problem better defined.

The e-mail I sent is the best (most current) explanation/observations:

I check my computer and find that CPU usage for DF is Zero. That is the first symptom I see. During gaming and such, I normally do not stop DF.

That first screen shot I sent you is how I find my machine. At that point I have not tried to manipulate DF in any way. (It was supposed to be running)

The stuff with the dfGUI stop button is all after the fact. I had hit the stop button hoping to be able to either up-load or re-start the service (with a subsequent hit of the start button)

-Sid

I think the very first post I made here, and (even more so) the e-mail are accurate. The stuff I reported in the middle of this thread.... well, all I can figure is that I fooled myself by playing with the dfGUI controls when DF was in the abnormal state and I didn't recognize it.

[Digital Parasite]

Originally posted by Brian the Fist
Hmm, seems to work fine for me. It sound slike the problem is when you hit the stop button in dfGUI, it is not exiting like it should. This is a problem. I am not sure what dfGUI does when you hit the stop button, maybe Jeff can tell me. But it sounds like its not exiting, thats the problem that needs to be fixed. It should disappear totally from task manager

Sorry, I have been away and haven't had a chance to look at Sid's files that he sent me yet.

When you click on the STOP button, all the GUI does is delete the foldtrajlite.lock file so it lets the DF client itself shutdown (hopefully gracefully).

Jeff.