A standard Windows Server AMI on g4dn or g5 may initially expose only Microsoft Basic Display Adapter, with no working nvidia-smi. After installing the driver, another gap appears: the Windows CloudWatch Agent cannot use the Linux nvidia_gpu receiver.

This guide connects both steps—installing an AWS GRID driver and publishing parsed nvidia-smi values as custom CloudWatch metrics.

Amazon EBS supports increasing a volume size, but it does not support shrinking one directly. For a data volume, create a smaller volume and copy the files to it. A Windows root volume, or a data volume containing continuously written files such as SQL Server databases, should be handled offline after stopping the instance.

This example uses Windows Server 2012 R2. The original root and data volumes are both 100 GB gp3 volumes, and the target size is 50 GB each.

Create an AMI before starting, test the procedure in a non-production environment, and schedule the root-volume migration during a maintenance window. Stopping an instance can lose instance-store data and an automatically assigned public IP address (an Elastic IP is not affected).

Windows instances fail Sysprep when creating an AMI, reporting error 0x80073cf2. The cause is an AppX package registered by Notepad++ 8.9+ that remains even after uninstalling the desktop version.

After an EC2 Windows instance boots, the system time is 8 hours behind. W32Time then applies a one-time 8-hour jump on sync, causing log timestamp chaos and Kerberos failures. The root cause is confusion between RTC local time and UTC.

A Mac mini mounts Windows PC drives (C/D/E) and a NAS via SMB. The core pain point: after macOS sleep, reboot, or network disconnection, SMB mounts drop silently and never auto-recover — requiring manual Finder reconnection every time. This article documents a complete auto-reconnect solution.

Two EC2 Windows instances in the same VPC fail to access a shared folder via SMB. Security groups and NACLs are open, Windows Firewall is disabled, but net use reports error 1792, and packet capture shows STATUS_NETLOGON_NOT_STARTED. The root cause is not VPC networking — security software is intercepting the SMB authentication flow.

EC2 instances launched from public Windows AMIs cannot be directly exported as VMDK via VM Import/Export because they contain AWS-licensed software. The workaround is to perform a block-level read of the EBS root volume after stopping the instance, convert it with qemu-img, and offline-enable Windows generic storage drivers to avoid blue screens in VMware.

On a Windows Server 2008 EC2 instance, the CloudWatch Agent service status was normal, but monitoring metrics could never be reported to CloudWatch. The logs repeatedly showed x509: certificate signed by unknown authority, and the root cause was ultimately traced to outdated system root certificates and insufficient TLS support.

When an EC2 Windows instance passes status checks, responds to ping, but RDP connections fail, and Stop & Start resolves the issue — don't only look at the network. Windows internal resource exhaustion can also cause critical processes like Winlogon to crash.

Enabling BitLocker system drive encryption on EC2 Windows instances is feasible, but the risks are higher than on physical machines. The key issue is that EC2 typically lacks a traditional TPM, so after encrypting the system drive, a password is required during the boot phase, and the standard console may not provide a reliable input channel.