Relax rwlock test (for CI) (kaspanet#408)

* relax rwlock test

* fix tn11 disk requirements based on actual network load

docs/testnet11.md

@@ -23,7 +23,7 @@ Since we want the test condition to be as close as possible to organic, we encou
 
 The venue for discussing and monitoring the experiment will be the \#testnet channel on Discord. We encourage participants to describe the experience in general, and also tell us what hardware they are using and how well it handles the load.
 
-The recommended hardware requirements are 16GB of RAM, preferably a CPU with at least 8 cores, and an SSD drive with at least 100GB of free space (the 150GB for a safety margin is preferable). It might also be possible to operate with 8GB of RAM, by passing the `--ram-scale=0.6` parameter to `kashd`.
+The recommended hardware requirements are 16GB of RAM, preferably a CPU with at least 8 cores, and an SSD drive with at least 250GB of free space (the 300GB for a safety margin is preferable). It might also be possible to operate with 8GB of RAM, by passing the `--ram-scale=0.6` parameter to `kashd`.
 
 ## Setup Instructions
 

utils/src/sync/rwlock.rs

@@ -122,22 +122,22 @@ mod tests {
 
     #[tokio::test]
     async fn test_writer_reentrance() {
-        for i in 0..64 {
+        for i in 0..16 {
             let l = Arc::new(RfRwLock::new());
             let (tx, rx) = oneshot::channel();
             let l_clone = l.clone();
             let h = std::thread::spawn(move || {
                 let mut write = l_clone.blocking_write();
                 tx.send(()).unwrap();
-                for _ in 0..5 {
+                for _ in 0..10 {
                     std::thread::sleep(Duration::from_millis(2));
                     write.blocking_yield();
                 }
             });
             rx.await.unwrap();
             // Make sure the reader acquires the lock during writer yields. We give the test a few chances to acquire
             // in order to make sure it passes also in slow CI environments where the OS thread-scheduler might take its time
-            let read = timeout(Duration::from_millis(5), l.read()).await.unwrap_or_else(|_| panic!("failed at iteration {i}"));
+            let read = timeout(Duration::from_millis(18), l.read()).await.unwrap_or_else(|_| panic!("failed at iteration {i}"));
             drop(read);
             timeout(Duration::from_millis(100), tokio::task::spawn_blocking(move || h.join())).await.unwrap().unwrap().unwrap();
         }