Hi, for heatbeat 2.1.4 there is IMHO no "out-of-the-box" solution for that problem.
I dont know, if the following method would be a valid method: Edit(!) the stonith script and add a sleep XX to the one of the nodes stonith script. This would cause one of the script to hang for some seconds. In consequence the resulting stonith actiond should not appear at the same time. Hopefully this does not work against haertbeat internal sleeps (did not test that so far). This method also causes, that teh cluster takeover action will run some seconds later (on the changed node), because the stonith action has to be fullfilled before other actions could be processed. @List: Would that be a valid work-arround? Regards Fabian Herschel [email protected] schrieb: > Betreff: > [Linux-HA] Configurating STONITH device (how to avoid reset each other) > Von: > Alessandra Giovanardi <[email protected]> > Datum: > Wed, 4 Mar 2009 18:07:32 +0100 (MET) > An: > [email protected] > > An: > [email protected] > > > Hi, > I'm using heartbeat on a cluster of 2 nodes and stonith to avoid split > brain with external/ipmi: > > heartbeat-stonith-2.1.4-0.11 > heartbeat-2.1.4-0.11 > > I'm using heartbeat with crm off (version 1-like). > > I've a question: If the nodes turn unavailable *each* *other*, how can > avoid that node-1 RESETS node-2 and node-2 RESETS node-1 at same time? > > Which is the same question of this post: > http://www.nabble.com/Configurating-STONITH-device-(reset-each-other)-td21672102.html > > > where the answer: > No, but it is extremely unlikely for this to happen. > > is for me not so exhaustive... > > Someone has solved this problem or evalutated the occurrence of this event? > > Thanks > A. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
