Compute Engine

The startup script for the managed instance group that routes external traffic to Apigee X now writes net.ipv4.ip_forward to /etc/sysctl.d/, so an instance no longer stops forwarding traffic when a reboot or OS update resets that setting; unrelated iptables NAT rules are unchanged. Existing instances must be recreated or restarted with the updated startup script to pick up this fix, since it is not applied retroactively to already-running instances.

Matters for: Teams running Apigee X on Google Cloud where a Compute Engine managed instance group handles external traffic routing.

The latest Compute Engine Agent Platform Workbench image (20261011.00_p0) installs the newest upstream packages and fixes a bug where gcloud and gsutil failed to load with an AttributeError in OpenSSL.crypto caused by a version mismatch between pyOpenSSL and cryptography. Engineers running this image no longer need to work around the crash to use gcloud or gsutil for managing their Compute Engine resources.

Matters for: Teams running Agent Platform Workbench notebook instances on Compute Engine that rely on gcloud and gsutil for resource management.

Compute Engine now generally supports syncing a VM's clock with its host server's clock using chrony and ptp_kvm, with accuracy designed to be within 1 ms. This accurate time synchronization is available in all standard public Google Cloud regions and zones.

Matters for: Relevant to teams running latency-sensitive or time-critical workloads such as distributed databases, financial trading systems, or event ordering pipelines that depend on precise clock synchronization across VMs.

Compute Engine now generally supports sharing a single Hyperdisk volume across multiple instances on additional machine series: C3 bare metal instances support multi-writer mode for Hyperdisk Balanced volumes, and C4 instances created after May 1, 2026 support multi-writer mode for Hyperdisk Balanced, Hyperdisk Extreme, and Hyperdisk Balanced High Availability volumes. This is a non-breaking, additive capability, so no existing configurations need to change.

Matters for: Teams running clustered or high-availability workloads on C3 bare metal and C4 Compute Engine instances that need shared block storage access across multiple VMs.

Google rolled out a bulk security patch that remediates Critical and High severity CVEs in the custom container images used by Workbench instances running on Compute Engine VMs, and installed the latest upstream dependency packages for those images.

Matters for: Teams running notebook or Workbench-based development environments on Compute Engine VMs that rely on Google-maintained container images for their workloads.

Google Cloud has released the X5 series memory-optimized machine type, generally available, which raises the memory ceiling from the previous 32TB limit on X4 to 48TB configurations on a single node.

Matters for: Teams running large-scale SAP HANA and RISE with SAP deployments that need to scale mission-critical databases vertically on a single node.

This release note covers updates to Apigee hybrid, BigQuery, Bigtable, Cloud Logging, Cloud Monitoring, and Cloud SQL for PostgreSQL, but it contains no entries describing changes to Compute Engine. As a result, no new Compute Engine capability can be summarized from this source text.

Matters for: This applies to anyone tracking Compute Engine release notes, since no relevant changes are documented in this particular update.