fdanapfel
commented
1 day ago I noticed tha there is another check that is referencing the same Red Hat KCS 4849731 but actually does check if the resource defaults are set: https://github.com/SAP/arp142-hana/blob/main/scripts/lib/check/7018_pacemaker_attributes.check
Therefore my recommendation would actually be to remove the checks for RHEL from https://github.com/SAP/arp142-hana/blob/main/scripts/lib/check/7015_pacemaker_version.check
https://github.com/SAP/arp142-hana/blob/main/scripts/lib/check/7015_pacemaker_version.check
The check references Red Hat KCS 4849731 "Why does pacemaker stop the Slave of a SAPHana resource when pengine reports "Pre-allocation failed"? ((https://access.redhat.com/solutions/4849731) and based on this reports the following when a lower pacemaker version is installed on RHEL 8:
[C] L7015 pacemaker must be updated (SAP Note #1552925,Red Hat KB #4849731) (is: 2.1.5-9.4.el8_8, should be: >=2.1.6-8.el8)
Since pacemaker version 2.1.6-8.el8 is only available for RHEL 8.10 and Red Hat has no plans to backport the fix for this issue to previous RHEL 8 minor releases it wil not be possible for customers who have locked their RHEL 8 release to a lower RHEL 8 minor release (for example 8.8) to implement the recommendation from the check.
The check apparently does not take into account that there is a workaround available for the issue documented in KCS 4849731: "As a workaround for this issue, setting a default resource-stickiness will also avoid the warnings."
The Red Hat documentation for the HA solutions for SAP actually states that customers must set up resource defaults for resource-stickiness and migration-threshold: HANA: https://docs.redhat.com/en/documentation/red_hat_enterprise_linux_for_sap_solutions/8/html/automating_sap_hana_scale-up_system_replication_using_the_rhel_ha_add-on/asmb_config_ha_cluster_automating-sap-hana-scale-up-system-replication NetWeaver / S4HANA app servers: https://docs.redhat.com/en/documentation/red_hat_enterprise_linux_for_sap_solutions/8/html/configuring_ha_clusters_to_manage_sap_netweaver_or_sap_s4hana_application_server_instances_using_the_rhel_ha_add-on/asmb_set_cluster_v8-configuring-clusters-to-manage#con_config_general_cluster_v8-configuring-clusters-to-manage
It would therefore be good if the check could be enhanced to verify if customers have set a resource default for resource-stickiness and if that is not the case then customers should be adviset to set the default value for resource stickiness according to the documentation, instead of just telling customers tha they have to update the pacemaker version.