[Pacemaker] Fatal assert crm_peer_cache != NULL

[Andrew Beekhof]

On Tue, Mar 3, 2009 at 13:20, Andreas Vogler <av at geneon.de> wrote:
> Hi all,
>
> I have set up a two node cluster using pacemaker-1.0.0. This worked except for some strange behaviour with respect to handling a master/slave resource. So I upgraded one of the nodes to the openSuSE 10.3 RPM packages version 1.0.2 to make sure I had the most recent fixes before asking stupid questions ;-)
>
> But now I can't start the crm. It fails with the following assert:
>
> Mar  3 13:29:20 hideki crmd: [6293]: ERROR: crm_abort: crm_glib_handler: Forked child 6296 to record non-fatal assert at utils.c:438 : g_hash_table_lookup: assertion `hash_table != NULL' failed
> Mar  3 13:29:20 hideki crmd: [6293]: ERROR: crm_abort: crm_update_peer: Triggered fatal assert at membership.c:313 : crm_peer_cache != NULL
>
> Is it necessary to change my configuration after upgrading or is this a bug?

Looks like a bug to me.

Do you know how to use hb_report?  It will grab everything I need to
figure out what happened.