Twice I have received a Visual Fortran Run-Time error using BOINC. My work unit runs for less than 30 hours and appears to trickle. Thereafter, more work is requested and the work unit starts at zero hours. I am not sure if I get the Fortran error at each trickle. The error on screen is:

fortl:severe (38)error during write,unit 8, file c:\\
program files\\BOINC\\projects\\climatprediction.net\\
hadcm3lb_rb1p_05055947\\TMP\\pipe_dummy

There is more to the error perhaps irrelevant.

---

The TMP\\pipe_dummy can be a read or write error on the hard disk.
It may indicate that there are faulty sections on the hd.

The TMP\\pipe_dummy can be a read or write error on the hard disk.
It may indicate that there are faulty sections on the hd.

Thanks, I will check HD condition.

---

Just for reference, I/O error 38 is \'reached end of file\' (i.e., file was shorter than expected).

A pipe is usually not a real file, but an RPC mechanism between two processes, which uses file semantics (for example, the unix pipe for chaining the output of one command into the input of the next command). But since this is \'pipe_dummy\', it may imply it is a real file pretending to be a pipe :-).

In the context of a real pipe, an I/O error 38 indicates that the source process supplied insufficient information to the consuming process. But I\'m not sure what processing is happening here, so the problem could be just about anything (hardware, software, timing issues, antivirus, firewall ...)

I doubt any of that is actually useful information, but as background ...

If you recall, this was exactly the same problem that \'Bob\' on the BBC boards was having a month or so back. I don\'t think that was solved at the time.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

I too encountered this situation, once a few minutes ago, and again while writing this note. I\'ve inspected my hard drives and discovered no problems--neither with the format nor with a read scan.

I have a screen shot captured, but don\'t see a way to upload it. It is the same error except ...hadcm3lb_5kr7_05048159\\tmp\\pipe_dummy. The stack traceback is without symbols, but perhaps it might be of use:
hadcm3transum_5.0 006EC5BA
hadcm3transum_5.0 006E983C
hadcm3transum_5.0 0069AFCC
hadcm3transum_5.0 0069B47E
hadcm3transum_5.0 00683B61
hadcm3transum_5.0 004111A5
hadcm3transum_5.0 00407084
KERNEL32.dll 7C98989

Both my occurrences gave an identical message.

---

Which antivirus do you use? I recently found a couple of people who had this exact error while an antivirus scan was running.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

I\'ve had this occur. I changed to BOINC 5.4.6 and a dead model came back to life, and no more erro messages.

---