Announcements
- There have been no announcements in the last month.
Recent Updates
- 2025.11.25.1
Better VM Control, Improved Volumes Management, and Platform Fixes
This release adds smarter volume management and more resilient load balancer probe handling. It introduces automatic job-queue retries and includes fixes for RBD initialization and IPv6 auto-extend behavior. Gateway instances are now removed automatically when no longer needed, and VM resource limits can be set arbitrarily in the portal.
- fixed
ISO Attachments for VMs
ISO downloads, for VM attachments, now work within environments where the discovery service is running at another provider.
- improvement
Clean Orphaned External Volumes
Cycle will now auto-clean orphaned volumes, as long as 'Delete' is passed during volume delete.
- improvement
Handle Early HTTP Client-Exits (Probes) Within the LB
If an HTTP client exits, prior to TLS negotiation, the LB no longer considers the request an initialization timeout.
- improvement
Auto-Retry for Job Queuing
If a job is being queued during an election, the API will auto-retry every 3 seconds, for 3 attempts.
- fixed
IP Pool Auto-Extend
Fixed invalid gateway CIDR for IPv6 ranges that are auto-extended as usage
- fixed
RBD Kernel Module
Automatically initialize the RBD kernel module for Ceph if needed, and not already initialized
- improvement
Auto Removal of Gateway Instances
When a server no longer needs gateway instances, they're now automatically removed.
- improvement
Virtual Machine Resource Limits
Virtual machines can now be set to arbitrary resource limits in the portal.
- 2025.11.12.1
Bring Your Own Storage: Mount SANs, Ceph Pools, and More to Containers & VMs
Full external volume support for containers and VMs has just landed in this update. Cycle now lets you plug in SAN or Ceph storage as easily as local disks, bringing flexibility to how you manage persistent data. Windows VMs are now supported as well, bringing more variety to the workloads you can run.
- added
External Volumes - A New Cycle Primitive
Cycle now supports connecting storage backends such as SAN or Ceph clusters, allowing you to create block or filesystem volumes that can be mounted into containers and VMs.
- added
Windows VMs Now Supported
Windows VMs are now supported on Cycle. We've added the ability to specify what the VMs host OS is, and Cycle will make some adjustments to the VM to facilitate it. For windows, this means mounting the virtio drivers to support the virtualized hardware, including disk and network.
- improvement
Ability to Customize VM CPU Model, Machine Type, and More
It's now possible to customize the CPU model and CPU feature set that is exposed to a VM. This improves compatibility with a variety of different operating systems, including Windows.
- improvement
VM 'configuring' State
A new VM state, 'configuring', was introduced. This state indicates that the VM hypervisor is preparing the operating system, and isn't yet running.
- improvement
VPN Service Customization
You can now add additional directives to the VPN service container. New directives will be appended to the config on start.
- improvement
Optimized Traffic Routing for VMs via Gateway Service
We've reconfigured some SNAT/DNAT routing rules within the gateway service to remove an extra hop that previously affected virtual machines.
- added
'Use Disk' Option For Larger Image Builds
The new 'use disk' option tells Cycle to use the hard disk instead of RAM when generating a container image during a build. This is especially useful for extremely large images, which may not fit into RAM and would otherwise error out when building.
- improvement
Dozens and Dozens of Portal Improvements
We've made a ton of improvements to the Portal, fixing UI quirks, issues, and flows. Everything from improved TLS certificate messaging with wildcards and user certificates to dark mode improvements.
- 2025.09.26.1
Improved Volume Capabilities and Host Device Injection
We've made it easier than ever to manage your most demanding workloads. SAN volumes can now be detached and re-attached across resources, while device injection adds new hardware flexibility. Also in this release is an easier-to-manage virtual provider experience, more customizable service logging, and more granular control over API and pipeline key IP restrictions.
- added
Host Device Injection
Host devices can now be exposed directly to a container via the container config, adding support for more sophisticated applications which might require encryption devices, USB mounts, etc.
- improvement
SAN Volume Modification
VM/Containers that are SAN mounted can now be modified after creation, enabling a user to detach a LUN from one VM/container and attach it to another.
- improvement
SAN ISCSI Initiator Name
Cycle now generates unique initiator names on startup for all SAN-enabled servers.
- improvement
Virtual Provider Modal Refactor
The virtual provider modal has been rebuilt to utilize the 'resource modal' enabling easier management of Virtual Provider integrations, especially those with dozens of ISOs.
- added
Environment Service Log Draining
Services can now be excluded from the log drain.
- fixed
API Key / Pipeline Keys IP Modification
IP restrictions can now be removed entirely for keys that previously had keys attached.
- 2025.09.03.1
SAN (Beta), Faster Migrations, Smarter Deployments, and More
We’re excited to introduce Storage Area Network (SAN) support (beta), making it easier to store large amounts of data reliably for both containers and virtual machines. This update also brings more flexible IP pool options, much faster migration speeds (up to 80% quicker), and smarter deployment strategies for edge, distribution, and HA setups. On top of that, we’ve fixed issues with instance migration and public IP filtering to make your environments more stable and dependable.
- added
SAN Support (BETA)
Servers, deployed via Virtual Providers, can now utilize SANs enabling large, persistent, data storage for both containers and virtual machines.
- improvement
Discovery Via TCP
The discovery service will now default to TCP resolution when resolving external DNS domains, enabling larger DNS payloads.
- improvement
Distribution, Edge, and HA Deployment Strategies
These orchestrators have been completely rebuilt to better account for existing instances, yielding a more calculated deployment of new instances.
- fixed
Instance Migration Race Condition
When moving a stateful container that utilizes a large (>1GB) volume across cloud providers, a timeout could occur which would trigger another migration attempt before the first was complete.
- improvement
LACP Bonds on Virtual Provider
The previous way we handled LACP negotiation didn't work for all NIC vendors, our new process supports a much larger range of NICs, if not all.
- improvement
Migration Speed
We've increased the block size for streaming migrations, decreasing the time for migrations by as much as 80% -- depending on link speed.
- added
IP Pool Options
Self-managed IP pools, added via Virtual Providers, can now be customized further to enable features like ARP proxying.
- fixed
Public IP Filtering
The traffic rule introduced to prevent two publicly accessible containers that existed within different environments from communicating was overly strict and, in rare cases, for single-node setups, blocked traffic that should have been valid. This has been fixed.
- 2025.07.24.1
IPv6 Support for Virtual Providers, Serial Console Access, and a SDN API Change
This update enhances platform flexibility with IPv6 support for Virtual Providers and improves overall system stability with critical fixes. It also refines API behavior to align with broader update patterns for a more consistent experience.
- improvement
Dual Stack Virtual Providers
Added the ability to configure IPv6 for virtual provider servers via the portal / API.
- fixed
Environment Service Updates
Previously, automatic updates could occasionally break the ability to SSH into a service container.
- changed
SDN L2 API
The updating of an L2 configuration has been moved from a PATCH to a reconfiguration job task to align with how other update calls are implemented in the API. This is a breaking change, but the base functionality was introduced in the last build.
- fixed
Cycle Kernel with UEFI Console
A recent kernel build broke console output on some bare metal models, that is now fixed.
- 2025.07.15.1
Incremental Improvements to VM Provisioning, Deployments, and Networking
This incremental release brings improvements and fixes to virtual machine provisioning, container deployments, and networking.
- fixed
Volume Provision for VMs
A race condition existed where an empty base disk could've been mounted as a cd-rom during VM provisioning, this has been fixed.
- changed
Container Deployment Strategy Updates
The platform now allows for a container's deployment strategy to be changed after creation IF it was never specified
- improvement
Firewall NAT Rule
We've changed the rule that prevented egress containers from talking to each other over their NAT IP. Previously, this blocked any egress traffic to any 10.x address, but now we're specifically only limiting traffic from 10.10.x.x to another 10.10.x.x (NAT egress).
- improvement
Traffic Drain Notifications
We now push traffic drain notifications down the internal api's notification pipeline, enabling containers to listen for drain events for other containers.
- added
Per-Deployment Stack Variables
Via the 'Deploy Stack' step within pipelines, users can now define variables that will override a stack's build-time variables at the time of deployment. This is beneficial for teams looking to deploy a single stack to multiple environments, but with unique variables per environment.
- added
Bare Metal H100 + A100 @ Vultr
The platform now supports provisioning of A100/H100 bare metal at Vultr.
- fixed
Container Deprecate Restart
Previously, marking a container as deprecated would cause the container to restart. This issue has been resolved.