This is a clone of issue RHEL-21879 to use for version rhel-9.7
–
Original description:
What were you trying to do that didn't work?
When setting up a cluster, pcs doesn't warn about too low values of knet ping_interval and ping_timeout. This leads to an error in corosync when starting the cluster. It is not possible to start a cluster with such a configuration.
Please provide the package NVR for which bug is seen:
pcs-0.11.6-6.el9.x86_64
How reproducible:
always, easily
Steps to reproduce
- pcs cluster setup corotest rh93-node1 rh93-node2 transport knet link ping_interval=50 ping_timeout=100
- pcs cluster start --all
Expected results
pcs reports an error in the cluster setup step informing the user about bad values of knet ping settings
Actual results
- pcs allows to create a cluster
- `pcs cluster start --all` fails
- errors in corosync.log:
warning [KNET ] link: host: 1 link: 0 interval: 50 too small (Success). interval lower than thread_timer_res (200 ms) has no effect error [KNET ] link: host: 1 link: 0 pong timeout: 100 too small (Invalid argument). timeout cannot be less than thread_timer_res (200 ms) error [TOTEM ] knet_link_set_ping_timers for nodeid 1, link 0 failed: Invalid argument (22) warning [KNET ] host: host: 2 has no active links warning [KNET ] link: host: 2 link: 0 interval: 50 too small (Success). interval lower than thread_timer_res (200 ms) has no effect error [KNET ] link: host: 2 link: 0 pong timeout: 100 too small (Invalid argument). timeout cannot be less than thread_timer_res (200 ms) error [TOTEM ] knet_link_set_ping_timers for nodeid 2, link 0 failed: Invalid argument (22) error [MAIN ] Corosync Cluster Engine exiting with status 8 at totemknet.c:881
- clones
-
RHEL-21879 pcs should validate knet ping_interval and ping_timeout better [rhel-10]
-
- Planning
-
- links to
