Incorrect calculation - Client 1.18

[M Cluster]

The new client 1:18 produces only error in the calculation.
Does anyone have an idea that lies ahead?


Greetings

8474076 17621489 1 Dec 2010 10:43:37 UTC 1 Dec 2010 10:55:58 UTC error computing 1:13 0:00 --- eon Client v1.18

[nullcoding]

Same problem here.

The 727560067-xxxxx WUs seem to run indefinitely. One had run for two hours and the progress bars all read 0.000%. I've suspended the project because there's definitely something wrong with the new client.

OSX 10.6.5 on Intel i5, by the way.

[chill]

I'm looking in to this. Could everyone, who hasn't said already, please tell me their operating system? If you run linux the output of uname -a would be nice. I have setup a new build process for compiling the client for all four platforms, and I did a small beta test and everything seemed to work, but these problems might be related to that. The one error I see now is that any linux computer with a kernel older than 2.6.27 will error out. I have fixed this and have new linux binaries I'll push out soon. But I'm not sure about this long workunits that aren't finishing. I have to check and see whats going on.

[chill]

I'm looking in to this. Could everyone, who hasn't said already, please tell me their operating system? If you run linux the output of "uname -a" would be nice. I have setup a new build process for compiling the client for all four platforms, and I did a small beta test and everything seemed to work, but these problems might be related to that.

The one error I see now is that any linux computer with a kernel older than 2.6.27 will error out. I have fixed this and have new linux binaries I'll push out soon. But I'm not sure about these long workunits that aren't finishing. I have to check and see whats going on.

[chill]

It looks like the stuck workunits are a bug in our code (not a bad client). It current is possible for the code to get stuck in an infinite loop. I'll fix this at the same time that I update the linux binaries.

[upstatelabs]

I'm having the same problem as nullcoding with some of the 727560067-* WUs, the WU just keeps running but makes no apparent progress. One machine I've had this on is a win2000 SP4 running BOINC 6.10.43, another is a winXP SP3 running BOINC 6.6.20. I have several others but haven't checked them yet.

edit: wu #'s

Update: After some checking, I have found that several of these 727560067-* WU's have completed successfully on my machines. Just not all.

[Phil P]

I aborted 2 after an hour on each, and got 2 more.
I'll give them a few hours and see if they wake up.

[nullcoding]

Ah...infinite loop. Long timescale dynamics. Haha.

Well okay - glad it's not the client. Seems like there has been a new client version every couple of days recently. Good thing it's not an issue with having developed them too fast or something like that.

Thanks for the update.

[Robert Pick]

some of the 171 WU's are doing the same thing!
I'm running an i7, xp,sp3.

[chill]

I turned off the problematic simulation. We are currently working on coding up a fix for this type of simulation so that it cannot get stuck.

[Beyond]

All my machines WUs errored out too. Unfortunately eon left some of the machines in a state where they were unstable, causing other projects both CPU and GPU to error.

[matob]

I had cancelled a few erroring WUs maybe an hour ago...
http://eon.ices.utexas.edu/eon2/result. ... id=8755661