fix: Avoid controller crashes when running large number of workflows (argoproj#9691)

Signed-off-by: Yuan Tang <[email protected]>

Author: terrytangyuan

.spelling

@@ -148,6 +148,7 @@ k8s-jobs
 kube
 kubelet
 kubernetes
+liveness
 localhost
 memoization
 memoized

docs/environment-variables.md

@@ -33,12 +33,14 @@ most users. Environment variables may be removed at any time.
 | `GZIP_IMPLEMENTATION`                  | `string`            | `PGZip`                                                                                     | The implementation of compression/decompression. Currently only "`PGZip`" and "`GZip`" are supported.                                                                                                                                                                    |
 | `INFORMER_WRITE_BACK`                  | `bool`              | `true`                                                                                      | Whether to write back to informer instead of catching up.                                                                                                                                                                                                                |
 | `HEALTHZ_AGE`                          | `time.Duration`     | `5m`                                                                                        | How old a un-reconciled workflow is to report unhealthy.                                                                                                                                                                                                                 |
+| `HEALTHZ_LIST_LIMIT`                   | `int`               | `200`                                                                                       | The maximum number of responses to return for a list call on workflows for liveness check.                                                                                                                                                                               |
 | `INDEX_WORKFLOW_SEMAPHORE_KEYS`        | `bool`              | `true`                                                                                      | Whether or not to index semaphores.                                                                                                                                                                                                                                      |
 | `LEADER_ELECTION_IDENTITY`             | `string`            | Controller's `metadata.name`                                                                | The ID used for workflow controllers to elect a leader.                                                                                                                                                                                                                  |
 | `LEADER_ELECTION_DISABLE`              | `bool`              | `false`                                                                                     | Whether leader election should be disabled.                                                                                                                                                                                                                              |
 | `LEADER_ELECTION_LEASE_DURATION`       | `time.Duration`     | `15s`                                                                                       | The duration that non-leader candidates will wait to force acquire leadership.                                                                                                                                                                                           |
 | `LEADER_ELECTION_RENEW_DEADLINE`       | `time.Duration`     | `10s`                                                                                       | The duration that the acting master will retry refreshing leadership before giving up.                                                                                                                                                                                   |
 | `LEADER_ELECTION_RETRY_PERIOD`         | `time.Duration`     | `5s`                                                                                        | The duration that the leader election clients should wait between tries of actions.                                                                                                                                                                                      |
+| `LIST_LIMIT`                           | `int`               | `200`                                                                                       | The maximum number of responses to return for a list call on workflows for workflow informer.                                                                                                                                                                            |
 | `MAX_OPERATION_TIME`                   | `time.Duration`     | `30s`                                                                                       | The maximum time a workflow operation is allowed to run for before re-queuing the workflow onto the work queue.                                                                                                                                                          |
 | `OFFLOAD_NODE_STATUS_TTL`              | `time.Duration`     | `5m`                                                                                        | The TTL to delete the offloaded node status. Currently only used for testing.                                                                                                                                                                                            |
 | `POD_NAMES`                            | `string`            | `v2`                                                                                        | Whether to have pod names contain the template name (v2) or be the node id (v1) - should be set the same for Argo Server.                                                                                                                                                |

workflow/controller/controller.go

@@ -803,6 +803,7 @@ func (wfc *WorkflowController) tweakListOptions(options *metav1.ListOptions) {
 	labelSelector := labels.NewSelector().
 		Add(util.InstanceIDRequirement(wfc.Config.InstanceID))
 	options.LabelSelector = labelSelector.String()
+	options.Limit = int64(env.LookupEnvIntOr("LIST_LIMIT", 200))
 }
 
 func getWfPriority(obj interface{}) (int32, time.Time) {

workflow/controller/healthz.go

@@ -13,7 +13,8 @@ import (
 )
 
 var (
-	age = env.LookupEnvDurationOr("HEALTHZ_AGE", 5*time.Minute)
+	age   = env.LookupEnvDurationOr("HEALTHZ_AGE", 5*time.Minute)
+	limit = int64(env.LookupEnvIntOr("HEALTHZ_LIST_LIMIT", 200))
 )
 
 // https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-a-liveness-http-request
@@ -30,7 +31,7 @@ func (wfc *WorkflowController) Healthz(w http.ResponseWriter, r *http.Request) {
 	labelSelector := "!" + common.LabelKeyPhase + "," + instanceIDSelector
 	err := func() error {
 		// avoid problems with informers, but directly querying the API
-		list, err := wfc.wfclientset.ArgoprojV1alpha1().Workflows(wfc.managedNamespace).List(ctx, metav1.ListOptions{LabelSelector: labelSelector})
+		list, err := wfc.wfclientset.ArgoprojV1alpha1().Workflows(wfc.managedNamespace).List(ctx, metav1.ListOptions{LabelSelector: labelSelector, Limit: limit})
 		if err != nil {
 			return err
 		}