[Pacemaker] start of pacemaker fails

[Andrew Beekhof]

On 10/06/2013, at 4:56 PM, Kazunori INOUE <inouekazu at intellilink.co.jp> wrote:

> Hi,
> I'm using pacemaker-1.1 (8807e990c7. the latest devel) with corosync-2.3.0.
> 
> After this commit, start of pacemaker fails.
> https://github.com/ClusterLabs/pacemaker/commit/17237616a12e37e2c073b3bff7dded3d66bc8201
> 
> I have not set nodelist{} in corosync.conf.
> 
> $ service corosync start
> Starting Corosync Cluster Engine (corosync):               [  OK  ]
> $ service pacemaker start
> Starting Pacemaker Cluster Manager:                        [FAILED]
> $ grep stonith-ng /var/log/ha-log
> Jun 10 14:39:54 dev1 pacemakerd[5398]:     info: start_child: Forked child 5401 for process stonith-ng
> Jun 10 14:39:55 dev1 stonith-ng[5401]:   notice: crm_add_logfile: Additional logging available in /var/log/ha-debug
> Jun 10 14:39:55 dev1 stonith-ng[5401]:    debug: crm_update_callsites: Enabling callsites based on priority=7, files=(null), functions=(null), formats=(null), tags=(null)
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: crm_log_init: Changed active directory to /var/lib/heartbeat/cores/root
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: get_cluster_type: Verifying cluster type: 'corosync'
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: get_cluster_type: Assuming an active 'corosync' cluster
> Jun 10 14:39:55 dev1 stonith-ng[5401]:   notice: crm_cluster_connect: Connecting to cluster infrastructure: corosync
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: crm_get_peer: Created entry e96819ac-1936-4332-9d42-089a45d80fbe/0xe004f0 for node (null)/3232261525 (1 total)
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: crm_get_peer: Node 3232261525 has uuid 3232261525
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: crm_update_peer_proc: init_cpg_connection: Node (null)[3232261525] - corosync-cpg is now online
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: init_cs_connection_once: Connection to 'corosync': established
> Jun 10 14:39:55 dev1 pacemakerd[5398]:   notice: stop_child: Stopping stonith-ng: Sent -15 to process 5401
> Jun 10 14:39:55 dev1 stonith-ng[5401]:   notice: corosync_node_name: Unable to get node name for nodeid 3232261525
> Jun 10 14:39:55 dev1 stonith-ng[5401]:   notice: get_node_name: Could not obtain a node name for corosync nodeid 0
> Jun 10 14:39:55 dev1 stonith-ng[5401]:    error: init_cs_connection_once: Could not establish local node name
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     crit: main: Cannot sign in to the cluster... terminating
> Jun 10 14:39:55 dev1 stonith-ng[5401]:     info: crm_xml_cleanup: Cleaning up memory from libxml2
> Jun 10 14:39:55 dev1 pacemakerd[5398]:    error: pcmk_child_exit: Child process stonith-ng (5401) exited: Network is down (100)
> Jun 10 14:39:55 dev1 pacemakerd[5398]:  warning: pcmk_child_exit: Pacemaker child process stonith-ng no longer wishes to be respawned. Shutting ourselves down.
> 
> The start was successful when adding the following change.
> In the case of corosync, is it specifications not to set TRUE to do_uname?

Looking back at the code, I think that all references to do_uname can be removed.
Thanks for testing!

> 
> $ git diff --patch-with-stat
> lib/cluster/cluster.c |    1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/lib/cluster/cluster.c b/lib/cluster/cluster.c
> index bee991c..efb3b42 100644
> --- a/lib/cluster/cluster.c
> +++ b/lib/cluster/cluster.c
> @@ -318,6 +318,7 @@ get_node_name(uint32_t nodeid)
> #  if SUPPORT_COROSYNC
>         case pcmk_cluster_corosync:
>             name = corosync_node_name(0, nodeid);
> +            do_uname = TRUE;
>             break;
> #  endif
> #endif
> $
> 
> Best Regards,
> Kazunori INOUE
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org