CloudHSM: SDK v5 Failover Issue

We have the following configurations in our testing environment:

• 2 HSMs in the CloudHSM cluster
• 2 EC2 instances installed with SDK v5 and PKCS11 application
• EC2#1 is connected to HSM#1 and EC2#2 is connected to HSM#2

Both PKCS#11 apps in EC2#1 and EC2#2 are running as normal initially. However, when one of HSM is removed from the cluster, we are starting to see errors in both EC2 instances. During the errors occurred, there is stilll one active HSM in the cluster. The same errors are shown in both EC2 instances: java.io.IOException: PKCS11Exception: CKR_DATA_LEN_RANGE

Anyone has any idea why SDK v5 fails in this case?

Thank you.