diff --git a/src/main/docbkx/troubleshooting.xml b/src/main/docbkx/troubleshooting.xml
index 7a9caa4..14e8e58 100644
--- a/src/main/docbkx/troubleshooting.xml
+++ b/src/main/docbkx/troubleshooting.xml
@@ -664,6 +664,18 @@ You may also need to download the
+
Secure Client Connect ([Caused by GSSException: No valid credentials provided (Mechanism level: Request is a replay (34) V PROCESS_TGS)])
+
+This issue is caused by the Kerberos bugs for its replay_cache component, #1201 and #5924, whom were talking about that the old version of krb5-server would false-positively blocks subsequent requests sent from a Principal.
+
+
+In this case, which means that the krb5-server would sometimes block the connections sent from one Client (one HTable instance with multi-threading connection instances for each regionserver); You would see such message 'Request is a replay (34)' in your client log if you hit this bug. You can just ignore this message due to the implementation of HTable will retry 5 * 10 for each failed connection by default. The HTable will throw IOException if any connection to regionserver was failed after the retries, therefore, the user client code for HTable instance still can handle it further.
+
+
+Otherwise, you can pick a new version of krb5-server to solve this issue gracefully. We tested and mitigated this issue in the environment: hbase-0.94.16 with krb5-server-1.10.3 on RHEL-6.4_GA-x86_64-10-Hourly2 on AWS, compared to the old environment: HBase-0.94.16 with krb5-server-1.6.1 on CentOS-5.3. Please refer to JIRA HBASE-10379 for more details.
+
+