>I also remember something about racing with dnsmasq, at which point I'dsay
>that making cluster depend on availability of DNS is e-h-h-h unwise
Not my choice... Or at least I would deploy bind/unbound caching servers in the
same VLAN instead of dnsmasq.Also, Filesystem resource agent's read + write
check is quite usefull. We got a crazy clusterized environment where on several
occasions read-only FS did not cause a failover (it's not pacemaker) and I
prefer to not be awaken by the oncall when this happens with the Scale-out I'm
building (currently it's just the QA cluster, but soon coming the prod).
The bad thing is this one is the first pacemaker in the environment and I need
to make it completely 'killer' or I will be forced to use the old cluster
solution which is crap (due to the implementation , not due to the product)....
double fencing mechanisms, fencing when FS is dead or HANA is having troubles,
etc.
Anyway, I am hoping that such kind of constraints will be more easier to
implement in the future, as this one is quite complex and it will give me a
hard time to explain it to the colleagues.
Best Regards,Strahil Nikolov
_______________________________________________
Manage your subscription:
https://lists.clusterlabs.org/mailman/listinfo/users
ClusterLabs home: https://www.clusterlabs.org/