This article details how to configure the best practice kernel parameters to ensure ClusterXL stability.
It is recommended to set ALL of those values on your cluster – also see sk92723 on Cluster flapping prevention.
Make sure the changes to the kernel parameters are performed on both members!
To test (will not survive a reboot):
fw ctl set int fwha_freeze_state_machine_timeout 200
fw ctl set int fwha_policy_update_timeout_factor 3
fw ctl set int fwha_pnote_timeout_mechanism_monitor_cpu 1
fw ctl set int fwha_pnote_timeout_mechanism_cpu_load_limit 80
fw ctl set int fwha_if_connectivity_tolerance 3
Once you are happy with the above you can enter the values into the $FWDIR/boot/modules/fwkern.conf file to make sure that the parameters stick after a reboot.
If the fwkern.conf doesn’t exist create it using the “touch” command:
[Expert@fw-trinity:0] # touch $FWDIR/boot/modules/fwkern.conf
Using the vi editor, insert the below commands into the fwkern.conf file:
fwha_freeze_state_machine_timeout = 200
fwha_policy_update_timeout_factor = 3
fwha_pnote_timeout_mechanism_monitor_cpu = 1
fwha_pnote_timeout_mechanism_cpu_load_limit = 80
fwha_if_connectivity_tolerance = 3