Hello, I am trying to get FCoE working, at least getting a initiator to exist at first (and eventually a target). I am having some trouble keeping a link up once lldpad has started.
If I boot up and ifup the adapter for IP, but not start lldpad or fcoe yet, the interface is functional and I can ping other devices. However, once I start lldpad, the interface link cycles on/off in 20-30 second intervals. /var/log/messages simply has: > Aug 14 14:49:23 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:49:27 spv-20 kernel: ixgbe 0000:20:00.1: eth9: NIC Link is Up 10 > Gbps, Flow Control: RX/TX > Aug 14 14:49:42 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:49:44 spv-20 kernel: ixgbe 0000:20:00.1: eth9: NIC Link is Up 10 > Gbps, Flow Control: RX/TX > Aug 14 14:50:22 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:50:22 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:50:26 spv-20 kernel: ixgbe 0000:20:00.1: eth9: NIC Link is Up 10 > Gbps, Flow Control: RX/TX > Aug 14 14:50:41 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:50:41 spv-20 kernel: ixgbe 0000:20:00.1: eth9: NIC Link is Up 10 > Gbps, Flow Control: RX/TX The log for the switch (a BNT G8124) has the following once lldpad is started: > Aug 14 14:49:50 10.0.1.243 NOTICE dcbx: Not able to detect DCBX peer on port > 8 > Aug 14 14:49:51 10.0.1.243 NOTICE link: link down on port 8 > Aug 14 14:49:57 10.0.1.243 NOTICE link: link up on port 8 > Aug 14 14:50:10 10.0.1.243 NOTICE link: link down on port 8 > Aug 14 14:50:15 10.0.1.243 NOTICE link: link up on port 8 > Aug 14 14:50:45 10.0.1.243 ALERT stg: STG 1, topology change detected > Aug 14 14:50:45 10.0.1.243 ALERT stg: STG 106, topology change detected > Aug 14 14:50:50 10.0.1.243 NOTICE link: link down on port 8 > Aug 14 14:50:57 10.0.1.243 NOTICE link: link up on port 8 > Aug 14 14:51:10 10.0.1.243 NOTICE link: link down on port 8 > Aug 14 14:51:12 10.0.1.243 NOTICE link: link up on port 8 It's not really clear to me what is bringing down the link: The host or the BNT switch. If I start fcoe, it appears to start the process of creating a fc host, but does not get very far, I suspect because the link is going down/up: > Aug 14 14:56:49 spv-20 fcoemon: /etc/fcoe/cfg-eth9: FCOE_ENABLE = "yes" > Aug 14 14:56:49 spv-20 fcoemon: /etc/fcoe/cfg-eth9: DCB_REQUIRED = "yes" > Aug 14 14:56:49 spv-20 fcoemon: /etc/fcoe/cfg-eth9: AUTO_VLAN = "yes" > Aug 14 14:56:49 spv-20 fcoemon: connected to lldpad > Aug 14 14:56:49 spv-20 fcoemon: Successfully created socket, socket file and > binding > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 10049 index 1 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 2 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth0 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 3 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth1 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 4 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth6 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 5 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth7 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 6 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth8 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 11043 index 7 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth9 > Aug 14 14:56:49 spv-20 fcoemon: eth9, Monitoring port eth9 > Aug 14 14:56:49 spv-20 fcoemon: eth9, INIT -> GET_DCB_STATE > Aug 14 14:56:49 spv-20 fcoemon: eth9, IEEE_INIT -> IEEE_GET_STATE > Aug 14 14:56:49 spv-20 fcoemon: eth9, GET_DCB_STATE -> GET_DCB_STATE > Aug 14 14:56:49 spv-20 fcoemon: eth9, IEEE_GET_STATE -> IEEE_GET_STATE > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 11043 index 8 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth2 > Aug 14 14:56:49 spv-20 fcoemon: sent 'C301010004eth9', rc=14 bytes succeeded > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 9 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth3 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 10 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth4 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 11 > Aug 14 14:56:49 spv-20 fcoemon: ifname eth5 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 01002 index 12 > Aug 14 14:56:49 spv-20 fcoemon: ifname usb0 > Aug 14 14:56:49 spv-20 fcoemon: Link event: 16 flags 11043 index 13 > Aug 14 14:56:49 spv-20 fcoemon: ifname br2 > Aug 14 14:56:49 spv-20 fcoemon: fcm_process_ieee_msg: eth9: FCoE pri mask = > 0x00 > Aug 14 14:56:49 spv-20 fcoemon: received len 18 buf 'R00C301010004eth90' > Aug 14 14:56:49 spv-20 fcoemon: eth9, GET_DCB_STATE -> DONE > Aug 14 14:56:49 spv-20 fcoemon: eth9, DONE -> INIT > Aug 14 14:56:49 spv-20 fcoemon: fcm_process_ieee_msg: eth9: FCoE pri mask = > 0x00 > Aug 14 14:56:49 spv-20 fcoemon: eth9, IEEE_GET_STATE -> IEEE_DONE > Aug 14 14:56:59 spv-20 fcoemon: Link event: 16 flags 01003 index 7 > Aug 14 14:56:59 spv-20 fcoemon: ifname eth9 > Aug 14 14:56:59 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:57:00 spv-20 fcoemon: received len 15 buf 'E5204eth9050101' > Aug 14 14:57:00 spv-20 kernel: ixgbe 0000:20:00.1: eth9: detected SFP+: 4 > Aug 14 14:57:00 spv-20 kernel: ixgbe 0000:20:00.1: eth9: NIC Link is Up 10 > Gbps, Flow Control: RX/TX > Aug 14 14:57:00 spv-20 fcoemon: Link event: 16 flags 11003 index 7 > Aug 14 14:57:00 spv-20 fcoemon: ifname eth9 I have configured the switch for FCoE (CEE, DCB, etc). At this point I cannot tell where the problem is. Any ideas? Also, is it possible to test FCoE with two hosts directly connected to each other? FYI, the ixgbe driver is 3.10.16 with default module options. Thanks, -Andrew _______________________________________________ devel mailing list [email protected] https://lists.open-fcoe.org/mailman/listinfo/devel
