LHC@home

I bet you guys with 60 seconds spend more time writing checkpoints files than computing. This problem crops up at malariacontrol which has huge checkpoint files.

i'd rather loose upto 15 minutes of work, as I don't reboot all that often than waste half the day wearing out my disk drive.

My guess is from my 12MB for 4 hours it is generating 1MB per checkpoint, and if you write 60MB per hour that is 300MB for 5 hours per task. This might be your problem. Especially if your client pauese a lot of tasks to start new ones. Also this multiplies by every project you run.

Try either 300 for 5 minutes or 900 for 15 minutes. I think in the long run you'll find overall better performance with not that much loss, unless your client regularly crashes or you reboot multiple times a day.

Report back if that helps, then we will know.

Just a thought, do you guys have your "write to disk" interval at some small number, maybe it is checkpointing way too often and creating a large file.

I use 900 seconds which is 15 minutes.

Client state Compute error
Exit status -177 (0xffffffffffffff4f)

This is a client error, not project error.

.......

You need to check the boinc manager disk tab, how much space is available to boinc and how much is free ? How much is showing using for LHC ? maybe it is another project taking up all the space ?

I'm running 3 LHC 1.0 on one machine, all at 5 hours (51%) so far and the total used by LHC is 65MB. On a second machine it is running 8 at about 1 hour so far and only used 55MB. I've got a lot of projects attached and only used 5GB and still 200GB free for boinc.