I have a small local project that I'm testing. Yesterday I adjusted the FPops estimate and bound values in the input template and since then the vast majority of my work units are getting cancelled by the client with a "finish file present too long" message.

How can I fix this error?

You probably have to get your application to clean up all its working files and perform whatever post-processing is necessary, before you call boinc_finish().

I think the "too long" message refers to any delay between signalling that the app has finished, and the process finally quitting.

Thanks for your reply.

I was able to fix it in a trial-and-error fashion, by increasing the number of total results. My workunits did not fail anymore.

However I'm a little intrigued because now I can't know for sure what caused the problem.

We have a number of these reports at WCG, which the techs consider a 'race condition'. Some of the recent BOINC development work regarding cleaning slot directories when a job completes may resolve this. https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,36995

These reports are very sparse, so much so that no action from the side of the WCG techs was to be undertaken after investigation:

I tested the workunit standalone and found no issues. In looking through the client code it looks like this condition occurs when the client finds that the boinc finish file has been written to disk but the science application process is still running. Since the finish file was written then there must be a hang in boinc_finish somewhere. Or it could be a bug or race condition in the client causing a false positive.

Have you upgraded your client version recently? Let me know if you continue to see these errors.

Thanks,
armstrdj

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt