I'm upgrading a cluster to CentOS-6.2 running an OFED-1.5.4.1 IB
stack. Every time a node tries to join the fabric, opensmd comes back
with this:
Oct 11 12:09:42 777493 [41F7700] 0x01 -> state_mgr_light_sweep_start:
ERR 3315: Unknown remote side for node 0x0008f10500108bfa (Voltaire
4036 # p3r17i1) port 15. Adding to light sweep sampling list
Oct 11 12:09:42 777532 [41F7700] 0x01 -> Directed Path Dump of 3 hop
path: Path = 0,1,23,5Oct 11 12:09:43 578014 [37F6700] 0x01 ->
log_send_error: ERR 5411: DR SMP Send completed with error
(IB_TIMEOUT) -- dropping
Method 0x1, Attr 0x15, TID 0x14a2
Oct 11 12:09:43 578050 [37F6700] 0x01 -> Received SMP on a 4 hop path:
Initial path = 0,1,23,5,15, Return path = 0,0,0,0,0
Oct 11 12:09:43 578065 [37F6700] 0x01 -> sm_mad_ctrl_send_err_cb: ERR
3113: MAD completed in error (IB_TIMEOUT): SubnGet(PortInfo), attr_mod
0x0, TID 0x14a2
These nodes work just fine on an older stack (CentOS-5.5,
OFED-1.5.3.1), and I've been running the same stack that I'm trying to
upgrade to (CentOS-6.2, OFED-1.5.4.1 with opensm 3.1.3.14) in
production for months on other clusters. I've tried multiple versions
of opensm already (both old and new). This cluster has slightly
different hardware (including the HCAs), but why isn't the SM able to
reach these nodes?
ibv_devinfo (on the old stack) shows:
hca_id: mlx4_0
transport: InfiniBand (0)
fw_ver: 2.7.9294
node_guid: 78e7:d103:0021:6984
sys_image_guid: 78e7:d103:0021:6987
vendor_id: 0x02c9
vendor_part_id: 26438
hw_ver: 0xB0
board_id: HP_0200000003
phys_port_cnt: 2
port: 1
state: PORT_ACTIVE (4)
max_mtu: 2048 (4)
active_mtu: 2048 (4)
sm_lid: 10
port_lid: 306
port_lmc: 0x00
link_layer: IB
port: 2
state: PORT_DOWN (1)
max_mtu: 2048 (4)
active_mtu: 256 (1)
sm_lid: 0
port_lid: 0
port_lmc: 0x00
link_layer: Ethernet
Let me know and I can provide any information necessary to help debug.
-JE
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
