Novell Cluster Services for NetWare provides a Preventing Cascading Failovers option. Its behavior was controlled using the clstrlib /hmo=off option in ldncs.ncf. Is there an equivalent option for Linux?
Cascading failover prevention was not implemented for Linux because user space applications (cluster resources) typically cannot cause a kernel failure and cascade failure of other kernels due to failing over user space code, as compared to ring0 NLMs on NetWare.