2011/6/6 Charles G Waldman <[log in to unmask]>: > > unless your job scripts have some sort of retry logic in them. > > We do, but there's no evidence that the retry is occurring. Either > the log files are misleading, or perhaps the segfault in xrdcp happens > after the file is transferred successfully (perhaps in some destructor > called at program exit?) That's a good point, but on segfault a failure should still be returned to shell. Is the failure happening on upload or download? > > Is there any chance to get the core files? > > I will adjust the 'ulimit' settings and see if I can get a useful > core dump. Thanks! Cheers, Lukasz