diff options
author | Håkon Hallingstad <hakon@verizonmedia.com> | 2021-03-19 21:55:42 +0100 |
---|---|---|
committer | Håkon Hallingstad <hakon@verizonmedia.com> | 2021-03-19 21:55:42 +0100 |
commit | 2d16332e15d66e23dcfb51dabe329c82a02e1eff (patch) | |
tree | 481efef4682a89fef2e509279eb62c29055d34d8 /clustercontroller-core/src | |
parent | 1ff6fa617f45cef77e97ba76750c476de48271c1 (diff) |
Increase the minimum time before first cluster state broadcast [run-systemtest]
Diffstat (limited to 'clustercontroller-core/src')
-rw-r--r-- | clustercontroller-core/src/main/java/com/yahoo/vespa/clustercontroller/core/FleetControllerOptions.java | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/clustercontroller-core/src/main/java/com/yahoo/vespa/clustercontroller/core/FleetControllerOptions.java b/clustercontroller-core/src/main/java/com/yahoo/vespa/clustercontroller/core/FleetControllerOptions.java index 528a9d79a7b..7e67cf27cba 100644 --- a/clustercontroller-core/src/main/java/com/yahoo/vespa/clustercontroller/core/FleetControllerOptions.java +++ b/clustercontroller-core/src/main/java/com/yahoo/vespa/clustercontroller/core/FleetControllerOptions.java @@ -94,7 +94,9 @@ public class FleetControllerOptions implements Cloneable { * Minimum time to pass (in milliseconds) before broadcasting our first systemstate. Set small in unit tests, * but should be a few seconds in a real system to prevent new nodes taking over from disturbing the system by * putting out a different systemstate just because all nodes don't answer witihin a single cycle. - * If all nodes have reported before this time, the min time is ignored and system state is broadcasted. + * The cluster state is allowed to be broadcasted before this time if all nodes have successfully + * reported their state in Slobrok and getnodestate. This value should typically be at least + * maxSlobrokDisconnectGracePeriod and nodeStateRequestTimeoutMS. */ public long minTimeBeforeFirstSystemStateBroadcast = 0; |