Handle NotInitialized::ShuttingDown error in shard split (#8506)

There is a race condition between timeline shutdown and the split task.
Timeline shutdown first shuts down the upload queue, and only then fires
the cancellation token. A parallel running timeline split operation
might thus encounter a cancelled upload queue before the cancellation
token is fired, and print a noisy error.

Fix this by mapping `anyhow::Error{ NotInitialized::ShuttingDown }) to
`FlushLayerError::Cancelled` instead of `FlushLayerError::Other(_)`.


Fixes #8496

pageserver/src/tenant/timeline.rs

@@ -137,7 +137,7 @@ use self::layer_manager::LayerManager;
 use self::logical_size::LogicalSize;
 use self::walreceiver::{WalReceiver, WalReceiverConf};
 
-use super::config::TenantConf;
+use super::{config::TenantConf, upload_queue::NotInitialized};
 use super::{debug_assert_current_span_has_tenant_and_timeline_id, AttachedTenantConf};
 use super::{remote_timeline_client::index::IndexPart, storage_layer::LayerFringe};
 use super::{remote_timeline_client::RemoteTimelineClient, storage_layer::ReadableLayer};
@@ -642,7 +642,13 @@ impl FlushLayerError {
     // When crossing from generic anyhow errors to this error type, we explicitly check
     // for timeline cancellation to avoid logging inoffensive shutdown errors as warn/err.
     fn from_anyhow(timeline: &Timeline, err: anyhow::Error) -> Self {
-        if timeline.cancel.is_cancelled() {
+        let cancelled = timeline.cancel.is_cancelled()
+            // The upload queue might have been shut down before the official cancellation of the timeline.
+            || err
+                .downcast_ref::<NotInitialized>()
+                .map(NotInitialized::is_stopping)
+                .unwrap_or_default();
+        if cancelled {
             Self::Cancelled
         } else {
             Self::Other(Arc::new(err))