Hi all: i have a glusterfs cluster with 4 replicated nodes. The client mount this one on a distribute volume. I use a script to test performance, i'm creating 10.000 files distributed like this 32k.file 40% 500k.file 30% 5m.file 15% 256k.file 10% 50m.file 5% I launch file ccopy script on a client (distributed & replicated config) and the performance are very very poor, more or less: 300 kbytes/sec (2.500 kbits/sec) Nodes are on 1Gb linked lan. Centos 5.4 64 bits (already updated) & glusterfs 2.0.8 I think that this isn't normal performance. Can anybody help me? Regards Andoni Ayala but i don't know why.