Bug 164535

That message:
 SM: 01000001 sm_stop: SG still joined

bothers me. It might be the case that cman has shut down (thus freeing its list
of nodes) before you've called kcl_get_node_addresses().

There's no loop in get_node_by_nodeid, it simply indexes an array (with bounds
checking) so this seems the most likely scenario. In fact, looking at the code,
there is an OBO error that allows a request for nodeid==0 even when the cluster
is down. This is the real bug I think.

Here are a couple more panics seen in cman membership on cman-1.0.4-0 at 
bootup:

> --------- Kernel BUG at membership:279 invalid operand: 0000 [1] SMP
> 
> Entering kdb (current=0x00000100e63c27f0, pid 7400) on processor 0 
> Oops: <NULL> due to oops @ 0xffffffffa01d70dd
>      r15 = 0x000001007d7b5dc8      r14 = 0x00000100e78b6400
>      r13 = 0x0000000000000003      r12 = 0x0000000000000003
>      rbp = 0x000001007d7b5e38      rbx = 0x000001007d7b5da8
>      r11 = 0xffffffff8011bac4      r10 = 0x0000000100000000
>       r9 = 0x000001007d7b5da8       r8 = 0xffffffff803dc9c8
>      rax = 0x000000000000001e      rcx = 0xffffffff803dc9c8
>      rdx = 0xffffffff803dc9c8      rsi = 0x0000000000000246
>      rdi = 0xffffffff803dc9c0 orig_rax = 0xffffffffffffffff
>      rip = 0xffffffffa01d70dd       cs = 0x0000000000000010
>   eflags = 0x0000000000010212      rsp = 0x000001007d7b5d30
>       ss = 0x000001007d7b4000 &regs = 0x000001007d7b5c98 [0]kdb> bt 
> Stack traceback for pid 7400
> 0x00000100e63c27f0     7400        1  1    0   R  
> 0x00000100e63c2bf0 *cman_memb
> RSP           RIP                Function (args)
> 0x1007d7b5d30 0xffffffffa01d70dd
> [cman]dispatch_messages+0x11ce (0x0, 0x100e5c78a80, 0x20100001e, 
> 0x900020104, 0x0)
> 0x1007d7b5e48 0xffffffffa01d8180 [cman]membership_kthread+0xbd9
> 0x1007d7b5f58 0xffffffff8010fdeb child_rip+0x8 [0]kdb>
> 
> 
> lizard03
> ----------- [cut here ] --------- [please bite here ]
> --------- Kernel BUG at membership:3150 invalid operand: 0000 [1] SMP
> 
> Entering kdb (current=0x000001007cda1030, pid 7494) on processor 0 
> Oops: <NULL> due to oops @ 0xffffffffa01d4131
>      r15 = 0x0000000000000003      r14 = 0xffffffffa01ee5e0
>      r13 = 0x00000100e2961a78      r12 = 0xffffffffffffffff
>      rbp = 0xffffffffa01ee750      rbx = 0xffffffffa01ee5e0
>      r11 = 0xffffffff80000000      r10 = 0xffffffff80000000
>       r9 = 0x0000000000000001       r8 = 0x0000000000000004
>      rax = 0x0000000000000000      rcx = 0x0000000000000080
>      rdx = 0x0000000000000080      rsi = 0x0000000000000080
>      rdi = 0x000001007c851e08 orig_rax = 0xffffffffffffffff
>      rip = 0xffffffffa01d4131       cs = 0x0000000000000010
>   eflags = 0x0000000000010246      rsp = 0x000001007c851df0
>       ss = 0x000001007c850000 &regs = 0x000001007c851d58 [0]kdb> bt 
> Stack traceback for pid 7494
> 0x000001007cda1030     7494        1  1    0   R  
> 0x000001007cda1430 *cman_memb
> RSP           RIP                Function (args)
> 0x1007c851df0 0xffffffffa01d4131 [cman]elect_master+0x3a 
> (0x100e322a180, 0x27c850712)
> 0x1007c851e08 0xffffffffa01d5373 [cman]a_node_just_died+0x186 
> (0xffffffffa01ee640, 0x1007cda1030)
> 0x1007c851e28 0xffffffffa01d56a6
> [cman]process_dead_nodes+0x5e (0x100005efa120000, 0x23d10d00000039, 
> 0x2014b001e, 0x1400020104, 0xd00000039010000)
> 0x1007c851e48 0xffffffffa01d8158 [cman]membership_kthread+0xbb1
> 0x1007c851f58 0xffffffff8010fdeb child_rip+0x8 [0]kdb>

I've seen this on the mailing list too. cman is trying to elect a master node
but there are no active nodes in the cluster (it thinks), so it can carry on.

Are there any other cman-related messages on the other nodes and earlier on in
the syslog of this node ?

Here's an attempt to fix this

-rSTABLE:
Checking in membership.c;
/cvs/cluster/cluster/cman-kernel/src/Attic/membership.c,v  <--  membership.c
new revision: 1.44.2.18.6.4; previous revision: 1.44.2.18.6.3
done

-rRHEL4:
Checking in membership.c;
/cvs/cluster/cluster/cman-kernel/src/Attic/membership.c,v  <--  membership.c
new revision: 1.44.2.22; previous revision: 1.44.2.21
done

Here's are a few cman messages that QA saw before the panic. Once we get the new
cman build we'll verify it's fixed.


CMAN: removing node link-07 from the cluster : No response to messages
CMAN: Attempt to re-add node with id 1
CMAN: existing node is link-01
CMAN: new node is link-08
----------- [cut here ] --------- [please bite here ] ---------
Kernel BUG at membership:279
invalid operand: 0000 [1] SMP
CPU 1
Modules linked in: gnbd(U) lock_nolock(U) gfs(U) lock_harness(U) dlm(U) cman(U)
md5 ipv6 parport_pc lp parport autofs4 i2c_dev i2c_core sunrpc ds yenta_socket
pcmcia_core button battery ac ohci_hcd hw_random tg3 floppy dm_snapshot dm_zero
dm_mirror ext3 jbd dm_mod qla2300 qla2xxx scsi_transport_fc mptscsih mptsas
mptspi mptfc mptscsi mptbase sd_mod scsi_mod
Pid: 4055, comm: cman_memb Not tainted 2.6.9-34.ELsmp
RIP: 0010:[<ffffffffa02150d5>] <ffffffffa02150d5>{:cman:dispatch_messages+4558}
RSP: 0018:0000010039f2dd48  EFLAGS: 00010212
RAX: 000000000000001d RBX: 0000010039f2dda8 RCX: 0000000000000246
RDX: 0000000000004b39 RSI: 0000000000000246 RDI: ffffffff803d9e60
RBP: 0000010039f2de38 R08: 000000000000000d R09: 0000010039f2dda8
R10: 0000000100000000 R11: 0000000000000000 R12: 0000000000000001
R13: 0000000000000001 R14: 000001003d26fe00 R15: 0000010039f2ddc8
FS:  0000002a95574b00(0000) GS:ffffffff804d7b80(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000002a9556c000 CR3: 0000000037e22000 CR4: 00000000000006e0
Process cman_memb (pid: 4055, threadinfo 0000010039f2c000, task 0000010039a91030)
Stack: 0000010039a91030 0000010020712a60 0000000000000000 ffffffff80304a85
       0000010039f2de38 ffffffff80304add 000000001ff89030 0000010039a92a00
       0000000000000000 0000000000000074
Call Trace:<ffffffff80304a85>{thread_return+0} <ffffffff80304add>{thread_return+88}
       <ffffffffa0216178>{:cman:membership_kthread+3033}
<ffffffff801333c8>{default_wake_function+0}
       <ffffffff801333c8>{default_wake_function+0}
<ffffffff801333c8>{default_wake_function+0}
       <ffffffff8013212e>{schedule_tail+55} <ffffffff80110e17>{child_rip+8}
       <ffffffffa021559f>{:cman:membership_kthread+0}
<ffffffff80110e0f>{child_rip+0}


Code: 0f 0b e3 de 21 a0 ff ff ff ff 17 01 48 c7 c7 60 d2 22 a0 e8
RIP <ffffffffa02150d5>{:cman:dispatch_messages+4558} RSP <0000010039f2dd48>
 <0>Kernel panic - not syncing: Oops

*** Bug 190230 has been marked as a duplicate of this bug. ***

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2006-0559.html