reenable stalld now that it is in coreos

opening up realtime cpu usage

allow optionally disabling stalld via env

Updated the bash and configured 4.10 upgrade to 4.11 to disable stalld

fixed a bash syntax error and added some output to the if statement

swapping to 4.9 to 4.10 from 4.10 to 4.11

ci-operator/config/openshift/release/openshift-release-master__ci-4.10-upgrade-from-stable-4.9.yaml

@@ -74,6 +74,7 @@ tests:
   steps:
     cluster_profile: gcp-openshift-gce-devel-ci-2
     env:
+      STALLD_ENABLED: "false"
       TEST_TYPE: upgrade-conformance
     workflow: openshift-upgrade-gcp-ovn-rt
 - as: e2e-azure-ovn-upgrade

ci-operator/step-registry/optional-operators/configure-realtime-tuned-profile/optional-operators-configure-realtime-tuned-profile-commands.sh

@@ -7,6 +7,8 @@ set -o pipefail
 node_role=${APPLY_NODE_ROLE:=worker}
 max_cpu=8
 isolated_cpu=${COMPUTE_NODE_ISOLATED_CPU:-4}
+sched_rt_runtime_us=-1
+stalld_service="service.stalld=start,enable"
 gcp_pattern="[n|c|m|a]{1}[1-9]{1}d?-(standard|highcpu|highmem|highgpu){1}-([0-9]+)"
 
 # Currently RT is only supported on GCP
@@ -24,6 +26,13 @@ if [[ "$isolated_cpu" == "$max_cpu" ]]; then
   echo "max and isolated cpu are equal, setting isolated CPU to $isolated_cpu"
 fi
 
+if [ ${STALLD_ENABLED:="true"} != "true" ]
+then
+  echo "disabling stalld and setting default realtime timeout"
+  sched_rt_runtime_us=950000
+  stalld_service=""
+fi
+
 echo "Creating new realtime tuned profile on cluster"
 oc create -f - <<EOF
 apiVersion: tuned.openshift.io/v1
@@ -48,6 +57,9 @@ spec:
       energy_perf_bias=performance
       min_perf_pct=100
 
+      [service]
+      $stalld_service
+
       [vm]
       transparent_hugepages=never
 
@@ -63,7 +75,7 @@ spec:
       [sysctl]
       kernel.hung_task_timeout_secs = 600
       kernel.nmi_watchdog = 0
-      kernel.sched_rt_runtime_us = 950000
+      kernel.sched_rt_runtime_us = $sched_rt_runtime_us
       kernel.timer_migration = 0
       kernel.numa_balancing=0
       net.core.busy_read=50

ci-operator/step-registry/optional-operators/configure-realtime-tuned-profile/optional-operators-configure-realtime-tuned-profile-ref.yaml

@@ -21,5 +21,9 @@ ref:
       Isolated cores to use for RT configuration
       4 vCPU is the current default for workload isolation, this is mirrored in our configs for RT
       https://github.com/openshift/enhancements/blob/master/enhancements/workload-partitioning/management-workload-partitioning.md#goals
+  - name: STALLD_ENABLED
+    default: "true"
+    documentation: |-
+      To use stalld to prevent thread starvation
   documentation: |-
     The configure-realtime-tuned-profile step applies realtime tuned profile to cluster workers.