I'm using the xeyes program to play around with process control with a two node cluster and pacemaker/openais/corosync. I've written a RA to control xeyes and tested the script for conformity using ocf-verify and it passes all the tests. I've also verified by hand that the return codes for start, stop, and monitor are correct for the RA, and that xeyes starts and stops correctly. I've placed the RA in a directory under /usr/lib/ocf/resource.d/bbnd on both nodes, and specified the location in my resource definition in my cib.
I defined an "eyes" resource and associated the RA with it, and I've configured each node to have the same weight in my rsc_location tag for the "eyes" resource, assuming that it would choose one of the other cluster nodes to start the RA (and xeyes). When I look at the status using crm_resource -resource eyes -locate, it claims the resource is running on both nodes. In fact, it's not running on either node. Also, when I run crm_verify -L, it complains that the resource is running on too many nodes. Again, it's not running at all on either node in the cluster. (BTW, I've run xhost + on both nodes so xeyes can have the display from any user). I'm not sure what I'm doing wrong...any advice on how to debug this situation would be appreciated. -Thanks Frank
_______________________________________________ Openais mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/openais
