Hello, I am testing out lustre, and I am running into a problem. If I am running read tests on three (or more) clients concurrently, I am getting an I/O error (EIO) from the read() system call. Here is the error from one of the OSS nodes: LustreError: 4235:0:(events.c:320:server_bulk_callback()) event type 8, status 19, desc d8b58000 LustreError: 4549:0:(ost_handler.c:540:ost_brw_read()) @@@ timeout on bulk PUT req@e03b9200 x279/t0 o3-><?>@<?>:-1 lens 328/280 ref 0 fl Interpret:/0/0 rc 0/0 LustreError: 4549:0:(ost_handler.c:600:ost_brw_read()) bulk IO comms error: evicting 5c83d_lov3_e62e363f5d@NET_0xa0a652f_UUID id 10.10.101.47-12345 LustreError: 4529:0:(ldlm_lib.c:1251:target_send_reply_msg()) @@@ processing error (-107) req@ed118000 x319/t0 o400-><?>@<?>:-1 lens 64/0 ref 0 fl Interpret:/0/0 rc -107/0 Lustre: 4322:0:(socknal.c:1408:ksocknal_close_stale_conns_locked()) Closing stale conn nid:0xa0a652f ip:0a0a6534/1021 incarnation:0x40641bb2eb921(0x40641bae6ab1d) Lustre: 4322:0:(socknal.c:1408:ksocknal_close_stale_conns_locked()) previously skipped 152 similar messages Version of lustre is 1.4.5, OS is SUSE 2.6.5-7.191 (both were compiled from source.) Configuration is 2 OSS nodes, with one of these nodes is also used to host the MDS. I don''t know if this could be a source of the problem? Anyway, here is my config: # OSS lmc -o config3.xml --add net --node ib-fe05x --nid ib-fe05b --nettype tcp --hostaddr ib-fe05b lmc -m config3.xml --add mds --node ib-fe05x --mds mds3 --fstype ext3 --dev /dev/vgtest1/lv_mds lmc -m config3.xml --add net --node ib-fe06x --nid ib-fe06b --nettype tcp --hostaddr ib-fe06b # clients lmc -m config3.xml --add node --node ib-be033 lmc -m config3.xml --add net --node ib-be033 --nid ib-be033b --nettype tcp --hostaddr ib-be033b # back-end client 2 lmc -m config3.xml --add node --node ib-be034 lmc -m config3.xml --add net --node ib-be034 --nid ib-be034b --nettype tcp --hostaddr ib-be034b lmc -m config3.xml --add net --node ib-be038 --nid ib-be033b --nettype tcp --hostaddr ib-be038b # # To create an lov made up of two partitions # lmc -m config3.xml --add lov --lov lov3 --mds mds3 --stripe_cnt 2 lmc -m config3.xml --add ost --node ib-fe05x --lov lov3 --ost ost3a --fstype ext3 --dev /dev/vgtest1/lv_a lmc -m config3.xml --add ost --node ib-fe06x --lov lov3 --ost ost3b --fstype ext3 --dev /dev/vgtest2/lv_b # mountpoints lmc -m config3.xml --add mtpt --node ib-fe05x --path /lustre3 --mds mds3 --lov lov3 lmc -m config3.xml --add mtpt --node ib-fe06x --path /lustre3 --mds mds3 --lov lov3 # lmc -m config3.xml --add mtpt --node ib-be033 --path /lustre3 --mds mds3 --lov lov3 lmc -m config3.xml --add mtpt --node ib-be034 --path /lustre3 --mds mds3 --lov lov3 lmc -m config3.xml --add mtpt --node ib-be038 --path /lustre3 --mds mds3 --lov lov3 Did anybody ran into a similar problem? Any help of how to solve this is appreciated. Thank You, Andras ------------------------ - Andras Frankel - - Environment Canada - - (514) 421-4721 - ------------------------