diff --git a/alert-rules.md b/alert-rules.md
index 0b423464c3843..3af40be4bdb0b 100644
--- a/alert-rules.md
+++ b/alert-rules.md
@@ -217,7 +217,7 @@ This section gives the alert rules for the PD component.
 
 * Description:
 
-    etcd writes data to disk at a lower speed than normal. It might lead to PD leader timeout or failure to store TSO on disk in time, which will shut down the service of the entire cluster.
+    If the latency of the fsync operation exceeds 1 second, it indicates that etcd writes data to disk at a lower speed than normal. It might lead to PD leader timeout or failure to store TSO on disk in time, which will shut down the service of the entire cluster.
 
 * Solution:
 
@@ -233,7 +233,7 @@ This section gives the alert rules for the PD component.
 
 * Description:
 
-    The number of Region replicas is smaller than the value of `max-replicas`. When a TiKV machine is down and its downtime exceeds `max-down-time`, it usually leads to missing replicas for some Regions during a period of time. When a TiKV node is made offline, it might result in a small number of Regions with missing replicas.
+    The number of Region replicas is smaller than the value of `max-replicas`. When a TiKV machine is down and its downtime exceeds `max-down-time`, it usually leads to missing replicas for some Regions during a period of time. 
 
 * Solution: