[Pacemaker] Question about fix for bug 2477

[Andrew Beekhof]

On Wed, Nov 10, 2010 at 3:09 AM, Bob Schatz <bschatz at yahoo.com> wrote:
> I am using 1.0.9.1 of Pacemaker.
>
> I have applied the fix for bug 2477 and it is not working for me.
> I started with this:  # crm_mon -n -1 ============ Last updated: Mon Nov  8
> 09:49:07 2010 Stack: Heartbeat Current DC: mgraid-mkp00009010repk-0
> (f4e5e15c-d06b-4e37-89b9-4621af05128f) - partition with quorum Version:
> 1.0.9-89bd754939df5150de7cd76835f98fe90851b677 2 Nodes configured, unknown
> expected votes 4 Resources configured. ============  Node
> mgraid-mkp00009010repk-0 (f4e5e15c-d06b-4e37-89b9-4621af05128f): online
> SSMKP00009010REPK:0     (ocf::omneon:ss) Master         icms:0  (lsb:S53icms)
> Started         mgraid-stonith:0        (stonith:external/mgpstonith) Started
> omserver:0      (lsb:S49omserver) Started Node mgraid-mkp00009010repk-1
> (856c1f72-7cd1-4906-8183-8be87eef96f2): online         omserver:1
> (lsb:S49omserver) Started         SSMKP00009010REPK:1     (ocf::omneon:ss) Slave
> icms:1  (lsb:S53icms) Started         mgraid-stonith:1
> (stonith:external/mgpstonith) Started  This is the output I received:  #
> ./crm_resource -r ms-SSMKP00009010REPK -W resource ms-SSMKP00009010REPK is
> running on: mgraid-mkp00009010repk-0 resource ms-SSMKP00009010REPK is running
> on: mgraid-mkp00009010repk-1
> The bug fix adds this check:
>     if((the_rsc->variant == pe_native) && (the_rsc->role == RSC_ROLE_MASTER)) {
> state = "Master";      }      fprintf(stdout, "resource %s is running on: %s
> %s\n",                  rsc, node->details->uname, state);
>
> When I dump the_rsc with the debugger I see that the_rsc->variant is pe_master
> and not pe_native.
>
> Also, the_rsc->role is RSC_ROLE_STOPPED.  This is even if I use the original
> crm_resource.c.  The complete dump of the the_rsc structure is:
>
> (gdb) print *the_rsc
> $2 = {id = 0x64d260 "ms-SSMKP00009010REPK", clone_name = 0x0,
>  long_name = 0x64d280 "ms-SSMKP00009010REPK", xml = 0x634ca0, ops_xml = 0x0,
> parent = 0x0,
>  variant_opaque = 0x64d6a0, variant = pe_master, fns = 0x7f8496b67f00, cmds =
> 0x0,
>  recovery_type = recovery_stop_start, restart_type = pe_restart_ignore,
> priority = 0, stickiness = 0,
>  sort_index = 0, failure_timeout = 0, effective_priority = 0,
> migration_threshold = 1000000,
>  flags = 262418, rsc_cons_lhs = 0x0, rsc_cons = 0x0, rsc_location = 0x0,
> actions = 0x0,
>  allocated_to = 0x0, running_on = 0x658060, known_on = 0x0, allowed_nodes =
> 0x60e2c0,
>  role = RSC_ROLE_STOPPED, next_role = RSC_ROLE_MASTER, meta = 0x648990,
> parameters = 0x648940,
>  children = 0x610280}
>
>
> Any idea why this can happen?
>
> Is there another fix I need for 1.0.9.1 to make this change work?

Quite possibly.
Its not at all recommended to backport individual patches.
Sometimes it works, but often it wont.

I'd recommend building the latest 1.0 code from mercurial instead.