Should I configure the storage multipath software to queue I/O when all paths have failed in my RHEL High Availability cluster ?
Issue
- I am setting up a cluster with
device-mapper-multipathmanaging the redundant storage links, and am not sure if I should configure the devices to queue I/O when all paths have failed, or to fail I/O back to the application when this happens. - Should I use
multipath'squeue_if_no_pathfeature in a cluster? - What should I set
no_path_retryto for mymultipathdevices when using GFS2? - Should I configure EMC PowerPath to block when all paths have failed on a cluster node?
Environment
- Red Hat Enterprise Linux (RHEL) 5, 6, or 7 with the High Availability or Resilient Storage Add On
- Some sort of multipath software managing redundant storage links on each server in the cluster
device-mapper-multipath- EMC PowerPath
- Hitachi Dynamic Link Manager
- Other 3rd party multipath solutions
- Shared storage accessed by applications or file systems managed by the cluster
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.