Peter Waltman
2008-Nov-06 21:09 UTC
[R] Looking for suggestions on how to debug pvm/snow proc's
Hi All - I'm running a faily long script that uses rpvm & snowFT to spawn off multiple processes with the 'clusterApplyFT' function. Specifically, what happens is that the head node generates a number of seed clusters that are then spawned off to the pvm cluster (in this case, nodes on a 4 dual-core machine) using clusterApplyFT to apply a function to each seed cluster that optimizes each seed cluster and returns the new, optimized clusters. In some cases, however, clusterApplyFT will return try-error's for some of the clusters. Since it's a try-error, I know there's no way to get a traceback (I've tried using clusterApplyFT to call traceback with no luck), but even when I switch to snow instead of snowFT, I still can't figure out a way to get a traceback for the failed clusters. Any suggestions? Thanks! Peter [[alternative HTML version deleted]]