We are happy to announce that the platform now supports automatic backups! Through a collaborative integration with Backblaze.com, stateful instances can now write backups on a user defined schedule & trigger restores through the portal and API. Backup settings are handled per container through an easy to use integration form, located in container configuration. A brand new "Backups" page - which boast a full log of events for each backup and additional meta information, is now available in the main container modal.
Stateful instances can be backed up and restored using the new container integration backups form.
A backups page where users can find a full log of events pertaining to their backups as well as other useful information. Restoring backups is also managed on this page.
Hubs can now set a reusable integration for object storage and backups. This integration can be used by any stateful container and is implemented through the container integrations backup form.
From time to time, a portal page refresh is needed when network drops occur between Cycle and a users browsers. The refresh would rectify container or instance states that had become out of sync. Improvements have been made so that the portal will be more resilient against situations like this and container/instance state should appear correctly more often after network drops that result in missed notifications.
Cycle has added support for Scoped Variables (Beta)! This new feature gives users the option of creating distributed runtime environment variables. These variables can have their values encrypted and can be either raw or url type. Also added to the platform is internal and public API support for instance telemetry. Both endpoints will now return more granular network information about the instance network usage. Other updates to the platform include an upgrade to RunC, bringing it to the most recent release and the ability for users to mark container filesystems as read-only. In portal news, the frontend team has made an effort to improve UI through more descriptive error messages and fixed a bug that caused certain graphs to show improper values intermittently.
The internal API now support the server/instances/telemetry endpoint which will return the latest cpu, ram, and network usage from every instance on that server.
The instance telemetry endpoint in the public API now includes information about network interface usage amounts.
Scoped Variables (Beta) have been released! Users can now select Scoped Variables from the environment navigation and enter either "Raw" or "URL" variables to be used by containers in that environment.
Users now have the ability to specify the container filesystem as read-only. Enabling this option will set the containers filesystem to readonly, but volumes associated with the container will not be affected.
There were a few places, such as container deployment strategy, container / instance states, and graphs, where data would become out of sync in the portal - this has been fixed.
The platform will no longer attempt to change the state of an instance if the instance is already in that state, even if the user submits subsequent requests that indicate an action is necessary.
When navigating the portal, users will notice more descriptive error messages - especially on pages which show a ghost, as well as improved skeleton loading for the activity list and instance list in the container modal.
The underlying RunC component was upgraded to the latest version.
In focus on this release was improvements and fixes for the Cycle portal. Lead by a range improvements and standardization to the instance modal. The update also addresses two minor routing bugs.
Fixed a bug that prevented the container modal from showing when clicking the associated link from a DNS linked record.
Sharing a link to a container modal and clicking on another nav menu item within the modal no longer causes the modal to disappear.
A wide range of improvements were made to portal UI including - standardization for git auth/ref form for image sources and stacks, instance migration revert panel formatting, and instance state badges for migrations and purge now show correctly via tooltip.
Cycle now supports Vultr's High Frequency VM's! These VM's carry 3GHz+ processors and NVMe storage . Vultr's internal testing shows that CPU processing, memory speed, and storage throughput deliver significant increases across the board when compare to the general compute machines.
Cycle now supports the "High Frequency" servers from the provider Vultr. These servers are packed with faster processors and NVMe storage options.
With the release of this new portal, users will immediately have access to faster page load times, a significant reduction in API requests originating from portal sessions, and a plethora of UI/UX updates and standardizations. Be on the lookout for a second set of updates to the new portal coming very soon!
The time in which content is available to view after loading shows about 30% sooner due to the use of skeleton loading.
The new portals data management was engineered to use roughly 45% less API requests.
The asset creation wizards were reworked to provide a more intuitive user experience.
In this release, the team has made fixes to two bugs. First, there was an issue where children of wildcard records wouldn't always receive a certificate ID update upon wildcard cert renewal. Secondly, there was a very rare case where restarts originating from health check failures would not fully restart the instance. Both of these issue have been solved and neither issue will require any changes from the user end to rectify effected assets!
Fixed an issue where children of wildcard records didn’t always get their certificate ID updated after the wildcard certificate was renewed.
A race condition that could prevent restarts from occurring during a health check failure was fixed.
The internal API now supports more granular image management options.
Container instance starts and stops should be noticeably faster due to optimizations made to how the portal handles these events.
The Dockerfile image source now has an optional registry credentials list, which is used to authenticate multi-stage image builds.
The internal API limit has been increased to 2,000 requests, per container, per 30 minutes.
Attempts made to create a TLS certificate can now be filtered by domain.
Initial functionality was added to support different versions of Cycle Environment Services.
Deploying an existing stack build through pipelines now shows more verbose output wile choosing images to (re)deploy.
We identified and fixed a race condition in one of Cycle’s core services that could’ve prevented the service from fully recovering after a network outage.
Added the ability to override the current tar.gz url for an image source, currently available for images created via the API.
Refactored how network routes are cleaned up once no longer needed. Additionally, Cycle now iterates through private network IPs with a grace period on previously utilized addresses, preventing potential conflicts during high-frequency scaling events.
Added the choice to omit container volumes during container create events.
Cycle team members who are added to hubs will now show a badge next to their name and will not be counted towards the total number of hub members.
Fixed an issue that caused container activity feeds to show incorrect information after certain pipeline reimage events.
Refactored how VPN routes are built to utilize less of the routing table while reducing the likelihood of routing conflicts.
Users can now serve Dockerfile Image Source's from a URL in the tar.gz format, adding even more flexibility to an already revolutionary feature. Along with this addition, there are some useful improvements to DNS, the agent, and and update for Cycle environment load balancers.
Users can now generate container images from a tar.gz, via a URL, using the docker-file image origin. All file management and image creation will be handled by the platform.
The Cycle Load Balancer service has upgraded the underlying HA Proxy package. This will resolve the segmentation fault error that some users have experienced recently.
Wildcard TLS certs for a hosted DNS Zone will now return an error on attempted delete if any other record (child) depends on that certificate.
Container DNS responses from the discovery service no longer include instances pending deletion.
Refactored agent sync logic to better handle network issues during automatic updates.
Resolved a race condition during the authorization process of Cycle’s communication mesh that could potentially deadlock and limit notification broadcasts.
Several UI components have been improved in pipelines, namely better validation measures on 'Enable Reference' form field in steps.
Calls made to the Servers and Containers endpoints will now return faster due to optimizations made to the API.
Image sources of type Dockerfile can now specify a branch, commit, or tag to be used when creating/importing new images from the source.
The platform now waits for a longer time period before severing the connection on incoming image builds. This makes it less likely for slower networks or large images to cause a build error.
Several UI elements have been improved in the portal, such as: stack create form inputs, pipeline editor styling / inputs, and image source list updates.
Build logs are essential to image debugging, but previously images created via Cycle Stacks wouldn't carry a build log of their own. Now, all images originating from stack builds to have access to build logs of their own, which should help users track down issues in their images much faster.
Test Hero
From simple tasks like importing an image, to complex sequences that affect entire clusters, Deployment Pipelines make it easy to automate nearly any task within Cycle.
You've spoken and we've listened! One of the most requested features on Cycle was simpler image management that didn't require re-entering information for subsequent image creation events. Now, users can simply connect an Image Source to Cycle and repeatedly create image artifacts from that source, greatly simplifying the update and reimage process as well as saving massive amounts of time on local > live debugging. More than that, you'll find that we've now included the ability for users to use a git hosted Dockerfile as an image source, completely removing the need to build and push images to a registry.
Image sources and their origins are now standardized throughout the platform. This includes the stack file spec as well as the portal GUI and public API.
Stacks can now build images from all supported image sources. This includes DockerHub, registries, or repositories containing Dockerfiles.
Images can now be built, created and imported using a git repo as the image source, completely removing the need to use an intermediary registry.
Cycles factory service has been rebuilt from the ground up to support any OCI-compliant image while also utilizing less RAM during build and increasing performance by over 30%.
Previously, a few different SFTP clients would fail to get directory listings due to the way we were encoding results on our end. With this refactor, users should see better performance along with a proper experience across more, if not all, SFTP clients.
A new API key capability - "capabilities.all" - has been added to the API key create and manage screens. This will allow a key to have access to all capabilities, making full access keys easier to manage.
A bug that would allow a user to submit a base volume reconfiguration larger than the maximum disk space on a server has been fixed.
We use cookies to enhance your experience. You can manage your preferences below.