Anand Avati <anand.avati@xxxxxxxxx> wrote: > Do you have something like strace in NetBSD to measure the syscall times on > the backend filesystem? It will be helpful to eliminate the backend > variable (by measuring time taken) during the slow runs.. Yes, here is a few seconds of kernel trace: http://ftp.espci.fr/shadow/manu/ktrace.txt -- Emmanuel Dreyfus http://hcpnet.free.fr/pubz manu@xxxxxxxxxx