This is an automated email from the ASF dual-hosted git repository.
bmahler pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/mesos.git
commit 543c01ab5c59695795fc6c7e6fc5bc02c3524121
Author: Benjamin Mahler <bmah...@apache.org>
AuthorDate: Mon Apr 15 14:44:36 2024 -0400
Mitigate a case where the agent gets stuck sending TASK_DROPPED.
Per MESOS-7187, there is a case where the master holds a stale resource
UUID for the agent's resources, and all subsequent task launches result
in the agent sending TASK_DROPPED due to "Task assumes outdated resource
state".
While this patch doesn't fix the general issue of MESOS-7187, it does
mitigate a known problematic case due to the introduction of the agent
having its own resource UUID.
---
src/master/master.cpp | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
diff --git a/src/master/master.cpp b/src/master/master.cpp
index 4a601c97e..e4f40104a 100644
--- a/src/master/master.cpp
+++ b/src/master/master.cpp
@@ -7746,6 +7746,22 @@ void Master::updateSlave(UpdateSlaveMessage&& message)
// providers as well.
}
+ // We don't expect the agent's resource version to change, but above we
+ // do have a check to see if it's changed and therefore set `updated`
+ // to true, so we might as well assign the new value here rather than
+ // ignore it.
+ //
+ // Now that we have resource versions in the agent, the lack of the
+ // update to the version here was causing MESOS-7187 to be triggered
+ // when the master receives a re-registration message from the old run
+ // of an agent and then ignores the new re-registration message from a
+ // new run of the agent. When the new agent sends the update message,
+ // the master was seeing the different resource uuid and setting
+ // `updated`, but wasn't actually setting the new version.
+ if (message.has_resource_version_uuid()) {
+ slave->resourceVersion = message.resource_version_uuid();
+ }
+
ReconcileOperationsMessage reconcile;
// Reconcile operations on agent-default resources.
