rabbitmq-server ocf script missing timeout value

[Andrew Woodward]

in some cases, it might be possible for the default timeout value to be missing when the OCF script is called. In this case the script may terminate with leftover iptables rules blocking the rabbit port 5673

OCF_RESKEY_CRM_meta_timeout_default caused a divide by zero (unknown value) error

diff --git a/deployment/puppet/nova/files/ocf/rabbitmq b/deployment/puppet/nova/files/ocf/rabbitmq
index 69591e1..44d5724 100644
--- a/deployment/puppet/nova/files/ocf/rabbitmq
+++ b/deployment/puppet/nova/files/ocf/rabbitmq
@@ -23,7 +23,9 @@ OCF_RESKEY_pid_file_default=/var/run/rabbitmq/pid
 OCF_RESKEY_log_dir_default=/var/log/rabbitmq
 OCF_RESKEY_mnesia_base_default=/var/lib/rabbitmq/mnesia
 OCF_RESKEY_node_port_default=5672
+OCF_RESKEY_CRM_meta_timeout_default=600000

+: ${OCF_RESKEY_CRM_meta_timeout=${OCF_RESKEY_CRM_meta_timeout_default}}
 : ${HA_LOGTAG="lrmd"}
 : ${HA_LOGFACILITY="daemon"}
 : ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}}

was able to resolve the issue in the environment

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/143387

[Bogdan Dobrelya (bogdando)]

Then you set OCF_RESKEY_CRM_meta_timeout_default=600000 you just tell the resource agent to ignore all timeout options from operations metadata of pcs resource. That would be a dirty hack, the issue should be resolved in another way. For example, wrap all ocf_run/shell commands in timeout. Or increase timeout parameters for corresponding operations.

But we should never make RA inconsistent with its resource metadata by hardcoding anything

[OpenStack Infra (hudson-openstack)]

Change abandoned by Bart?omiej Piotrowski (<email address hidden>) on branch: master
Review: https://review.openstack.org/143387