D2OL Woes

[magnav0x]

Not sure how long I'll participate in this project. I'm having serious problems keeping them crunching without baby sitting the clients. It seems that most of the machines stop crunching once they finish the WU buffer and don't even upload them. When I woke up today, 5 of my 6 machines were idle doing nothing at all. I just can't handle having to restart each one every time the buffered WU's are completed.

[rshepard]

are they crunching through the buffered units in offline mode and then refusing to upload when you put the client back into online mode? How large of a cache are you using?

[PCZ]

I find the D2OL client to be one of the most reliable.

I run online and every 6 hrs it connects to the mothership and uploads work and downloads WU's to replenish the cache.

magnav0x

How are you clients managing to run out of work ?
You can cache 2000 WU's thats enough to keep a fast PC busy for 3 weeks.

[StarDog]

The clients have been very stable for me also. Only problems I found was when running as a service, however it was pointed out in another thread that this problem could be resolved with a configuration change in the lax file.

If the client is having problems connecting for some reason, it usually gives an error message of some sort in the Status box from the GUI.

Are you running it as a service, or as a normal application? And are you using any command line options? Have you tried backing up the node.prp file and reinstalling? Just some suggestions...

[Bok]

Yeah,

Magnav0x, it does sound as if it's in offline mode.

if you are in controller:cli mode

type - List

and check the value of Online

if it's false, do

set Online=true

HTH

Bok

[Fozzie]

but only when runing 2 instances on an old dual box under Red Hat 9.

Never really got it sorted so i ran one instance of D2OL and one of LM.

[MerePeer]

I'm having a related prob. Fresh WHITEBOX linux install (which is same as redhat AS 3). Fresh java 1.4.2 install. Fresh communitytsc install.

Symptom, through observing ps -ef, is that "GridWin.exe" completes its job then the status says Docking Conformer #1 BUT it doesn't do this, and it doesnt proceed to #2 nor #20 and after a while it grabs another work unit (total queued decremented) and returns to running GridWin.exe. I'm running controller:cli. I'm seeing a consistent 40% of the CPU going to D2OL and the other 50%+ to GridWin.

Suspecting a newbie-linux-install issue I went into the res/data/bin dir and typed in ./DockWin.exe
and I get
./DockWin.exe: error while loading shared libraries: libstdc++-libc6.2.-2.so.3 cannot open shared object file: No such file or directory.

So (1) this should be easy to fix if someone knows please reply meanwhile I'll search d2ol forum. (2) magnovox could this be your problem too? This behavior of slowly reducing the queue but not really producing work -- are you seeing it sit in DockWin.exe at all/for any length of time?

[Bok]

nope, this is the same as I got.

I just took this library from another machine and copied it into /usr/lib

that fixes it.

Bok

[MerePeer]

Yes I can't seem to find that file. THe rpmfind.net seems to have some stale redhat links. If I get the right rpm, is an rpm like a zip, where you can pull 1 file out of it?

[IronBits]

Originally posted by Bok
nope, this is the same as I got.

I just took this library from another machine and copied it into /usr/lib

that fixes it.

Bok .tar or .zip it up and post it :crazy: :)

[Bok]

here you go

file (http://stats.open-dc.org/libstdc++-libc6.2-2.so.3)

Bok

[magnav0x]

While checking on my machines today I noticed one wouldn't process the WU's at all. I'd have it manualy download say 100 WU's, as soon as it was down downloading them it would say they were complete and want me to upload them :rolleyes: They were all working for a while :bang: Damn java and it's insuperior code! I may just go ahead and throw all my machines back on LM soon.

[MerePeer]

Bok - thank you that worked like a charm.

magnav0x - did you do an "ldd" of DockWin.exe and GridWin.exe?

[Bok]

java is only running the front end...

the programs themselves are in C++.

This sounds like the same problem as above

run

ldd Dockwin.exe

it's in the res/data/bin dir

Bok

[matrix_fan]

i seem to be having a simliar problem now.. Althought it's been very reliable project in he past. It'll startup, and then 1 minutes later it will leave the tasktray, and it doesn't even download it's tasks... It's not in offline mode or anything.... Anybody have any ideas?BTW the box taht's having the problem is a Win XP Pro