[openib-general] [Bug 351] New: Routing table problem in SLES10 when using port #2

[bugzilla-daemon at lists.openfabrics.org]

https://bugs.openfabrics.org/show_bug.cgi?id=351

           Summary: Routing table problem in SLES10 when using port #2
           Product: OpenFabrics Linux
           Version: 1.2
          Platform: All
        OS/Version: SLES 10
            Status: NEW
          Severity: major
          Priority: P1
         Component: IPoIB
        AssignedTo: bugzilla at openib.org
        ReportedBy: yohadd at mellanox.co.il


There is an issue with the routing table on SLES10 when using IB port #2.
After host reboot the routing table contain two entries for 12.X.X.X !!!
One of the entries is correct and point to ib1, the other one point to ib0.

Route output:

sw087:~ # route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
10.4.0.0         *               255.255.0.0     U     0      0        0 eth0   
12.4.0.0         *               255.255.0.0     U     0      0        0 ib1 
12.4.0.0         *               255.255.0.0     U     0      0        0 ib0  
link-local      *               255.255.0.0     U     0      0        0 eth0
11.4.0.0         *               255.255.0.0     U     0      0        0 ib0
loopback        *               255.0.0.0       U     0      0        0 lo
default         10.4.0.211      0.0.0.0         UG    0      0        0 eth0

The first entry for 12.4.0.0 point to I/F ib1, so in this configuration the
ipoib over 12.4.X.X will work fine.
After restarting the ib1 I/F with the ifconfig commands, the routing table
changed (the order between the two 12.4.0.0 entries changed) and ipoib over
12.4.X.X will not work any more.

Restarting ib1 I/F:

sw087:~ # ifdown ib1
    ib1       device: Mellanox Technologies MT23108 InfiniHost (rev a1)
sw087:~ # ifup ib1
    ib1       device: Mellanox Technologies MT23108 InfiniHost (rev a1)

Route output:
sw087:~ # route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
10.4.0.0         *               255.255.0.0     U     0      0        0 eth0 
12.4.0.0         *               255.255.0.0     U     0      0        0 ib0    
12.4.0.0         *               255.255.0.0     U     0      0        0 ib1  
link-local      *               255.255.0.0     U     0      0        0 eth0
11.4.0.0         *               255.255.0.0     U     0      0        0 ib0
loopback        *               255.0.0.0       U     0      0        0 lo
default         10.4.0.211      0.0.0.0         UG    0      0        0 eth0


Host info:

sw087:~ # hostinfo 
Name         =sw087
IP           =10.4.3.87
CpuNum       =4
CpuVendor    =GenuineIntel
CpuModel     =                  Intel(R) Xeon(TM) CPU 3.20GHz
CpuMhz       =3200.190
MemSizeKb    =4047700
MachType     =x86_64
KernelRev    =2.6.16.21-0.8-smp
ChipSet      =Intel Corporation E7520 Memory Controller Hub (rev 0c)
Os           =Welcome to SUSE Linux Enterprise Server 10 (x86_64) - Kernel \r
(\l).
IBDevsNum    =1
HCA0Name     =mthca0
HCA0Desc     =sw087 HCA-1
HCA0Type     =MT23108
HCA0FWVer    =3.5.0
HCA0PSID     =MT_0030000001
HCA0GUIDS    =NODE:0x0002c9871297bce0;SYS:0x0002c9871297bce3
HCA0Ports   
=1:0x0002c9871297bce1:0x0:11.4.3.87:DOWN;2:0x0002c9871297bce2:0x0:12.4.3.87:INIT
HCA1Name     =NONE
HCA1Desc     =NONE
HCA1Type     =NONE
HCA1FWVer    =NONE
HCA1PSID     =NONE
HCA1GUIDS    =NONE
HCA1Ports    =NONE
IBStack      =/usr/local/
IBStackType  =ofed
IBStackVer   =OFED-1.2-20070211-1558
IBMPI        =/usr/local//mpi
MST_BUILD    =4.3.6
IBADM_BUILD  =IBADM 2.1.0, 20060720-1410
WRITE_BW     =/usr/local/bin/ib_write_bw


-- 
Configure bugmail: https://bugs.openfabrics.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.