LHC@home

Hi all,


i got back to work after the weekend and found most of my PC's out of work except one. The Boinc 4.09 was stopped with an error message
SIGPIPE: write on a pipe with no reader
Exiting..

i can easily restart the project, but it ends up after 4-5 minutes with the same error message. I also allready got over this error and made the client to continue. I susect that this error is somehow associated with poor internet line.

Machine is on linux, for details see: http://lhcathome.cern.ch/show_host_detail.php?hostid=7207

Rings a bell to anybody?

thanks,

marshall

That error means one process is trying to write to a pipe that has no process at the other end to read what the first process is writing. If BOINC uses pipes to communicate with the science applications (haven't looked at that part of the code) then I would immagine that either BOINC or the LHC science application is crashing or in some way breaking the pipe and then this causes the error you are seeing. Are there any other lines of interest from any of your logs? Might keep a close eye on your syslog and (if you aren't doing so already) capture everything that BOINC prints to a log file so you can search through it more easily and see if anything else is amiss. You aren't running this over an NFS mounted filesystem, are you? I have heard there are major problems with that although I'm not sure what the details of it are. You might also try upgrading to BOINC 4.11 (currently officially in alpha but works for me). Get it at http://setiboincdata.ssl.berkeley.edu/boinc_alpha/download.php
Also you might want to take a look at this: http://forums.pcper.com/showthread.php?t=349185

Good luck! :)


--------------------------------------
A member of The Knights Who Say Ni!
My BOINC stats site

it is running on ext3, no NFS. i really suspect the poor internet connection (via radio). there is nothing more in my logs, actually now the system is working fine, got over it (uploaded file) and continues crunching.