Merge #1389

1389: fix(retire/rebuild): schedule rebuild termination async r=tiagolobocastro a=tiagolobocastro

When retiring don't wait for rebuild to terminate as it may get stuck if the device is not fully retired. It seems that IOs get stuck? For now we simply defer terminate to master reactor.

Co-authored-by: Tiago Castro <[email protected]>

.gitmodules

@@ -1,12 +1,12 @@
 [submodule "rpc/mayastor-api"]
 	path = rpc/mayastor-api
 	url = https://github.com/openebs/mayastor-api
-	branch = develop
+	branch = release/2.2
 [submodule "spdk-rs"]
 	path = spdk-rs
 	url = https://github.com/openebs/spdk-rs
-	branch = develop
+	branch = release/2.2
 [submodule "utils/io-engine-dependencies"]
 	path = utils/io-engine-dependencies
 	url = https://github.com/openebs/mayastor-dependencies.git
-	branch = develop
+	branch = release/2.2

io-engine/src/bdev/nexus/nexus_bdev_children.rs

@@ -955,8 +955,10 @@ impl<'n> Nexus<'n> {
             // Cancel rebuild job for this child, if any.
             if let Some(job) = child.rebuild_job() {
                 debug!("{self:?}: retire: stopping rebuild job...");
-                job.terminate().await.ok();
-                debug!("{self:?}: retire: rebuild job stopped");
+                let terminated = job.terminate();
+                Reactors::master().send_future(async move {
+                    terminated.await.ok();
+                });
             }
 
             let uri = child.uri();

io-engine/src/grpc/v0/nexus_grpc.rs

@@ -56,9 +56,8 @@ fn map_child_state(child: &NexusChild) -> (ChildState, ChildStateReason) {
         ChildStateClient::Closed => (Degraded, Closed),
         ChildStateClient::Faulted(r) => (
             match r {
-                FaultReason::AdminCommandFailed => Faulted,
-                FaultReason::IoError => Faulted,
-                s if s.is_recoverable() => Degraded,
+                FaultReason::NoSpace => Degraded,
+                FaultReason::Offline => Degraded,
                 _ => Faulted,
             },
             map_fault_reason(r),

io-engine/src/grpc/v1/nexus.rs

@@ -158,9 +158,8 @@ fn map_child_state(child: &NexusChild) -> (ChildState, ChildStateReason) {
         ChildStateClient::Closed => (Degraded, Closed),
         ChildStateClient::Faulted(r) => (
             match r {
-                FaultReason::AdminCommandFailed => Faulted,
-                FaultReason::IoError => Faulted,
-                s if s.is_recoverable() => Degraded,
+                FaultReason::NoSpace => Degraded,
+                FaultReason::Offline => Degraded,
                 _ => Faulted,
             },
             map_fault_reason(r),