Skip to main content

Seqera Cloud v27.1.0_cycle78

Breaking changes​

Private Studios now admit only the users on their allowed users list, the creator included. A creator who is not on the list can still start, stop, edit, and delete the Studio, but cannot connect to it or use its checkpoints. Existing private Studios with no allowed users have their creator added; a Studio already shared with another user keeps its list, and its creator must be added to it to connect.

Feature updates and improvements​

Studios​

  • Removed the limit of one additional allowed user on a private Studio, through the API.
  • Added stopping a private Studio, and extending its lifespan from inside the session, for its allowed users.
  • Added checkpoint access for a private Studio's allowed users, covering reading a checkpoint, renaming it, and creating a Studio from it.
  • Added mounting a data link in a Studio when it is in one of the compute environment's allowed S3 buckets, even if it uses other credentials. Applies to AWS Batch, AWS Cloud, and Seqera Compute compute environments.

Compute environments​

  • Added an Allowed S3 buckets field to the Seqera Compute compute environment form for granting access to S3 buckets in your own AWS account.
  • Added a Bucket policy section to the Seqera Compute compute environment details page with the policy to apply to each allowed bucket.
  • Redesigned the compute environments list as a table with columns for name, ID, region, enabled features, last activity, and status.

Monitoring​

  • Changed the run timeline to sort processes in pipeline order by default.
  • Added a Sort processes menu to the run timeline, with Earliest started and Latest started options.
  • Added an Attempt label to retried task rows in the run timeline, and a retry count to each process row.

Co-Scientist​

  • Changed the label below a conversation title to name the organization instead of the workspace.

Agents​

  • Added agents: named, reusable sets of instructions that Co-Scientist runs on your behalf in a workspace. Anyone in the workspace can run an agent without rewriting its prompt.
  • Added a service account as the identity an agent runs under. An agent's work is attributed to that account rather than to the person who created it.
  • Added binding a workspace GitHub App credential to an agent for clone, commit, and push operations in place of the creator's personal access token. Each installation token the agent mints is recorded in the audit log.
  • Added a read-only view of a workspace-owned agent thread in the Co-Scientist drawer, with a Fork button that copies the thread into a personal chat.
  • Added stopping a running agent session.
  • Added automatic disabling of an agent when its service account leaves the workspace.

Actions​

  • Added three event sources to actions: Bucket event, which fires when a marker file arrives in or is removed from cloud storage, Schedule, which fires on a recurring cadence, and Pipeline run event, which fires when a run of a watched pipeline reaches a terminal state.
  • Added an agent Target to the three new event sources, which hands the event to an agent instead of launching a pipeline.
  • Added Daily, Weekly, and Custom schedule options for a scheduled action. A saved action keeps the timezone it was created in.
  • Added Details and Trigger history tabs to the action details page. Trigger history lists each firing of the action and the event payload behind it.
  • Added a View target control to an agent row in Trigger history that opens the agent's conversation in the Co-Scientist drawer.
  • Added editing a saved action's name, labels, marker file, and schedule.
  • Added a label to the runs an agent action starts.
  • Added an optional pipeline and version to the Target of a pipeline run event action.
  • Changed a chained action to skip the run it caused itself.

Service accounts​

  • Added editing a service account's name, description, and workspace access from a dialog on its details page.
  • Blocked removing or deleting a service account that is bound to an active agent.

Access control​

  • Added single sign-out for applications that sign in with Seqera. Signing out of one now ends your Seqera session too.

General​

  • Improved screen reader output by hiding decorative icons and naming the icons that carry meaning.

Bug fixes​

Compute environments​

  • Fixed a compute environment deletion blocked by running workflows or active Studio sessions reporting a generic error instead of naming them.
  • Fixed Azure Batch Forge autoscaling adding one node per queued task instead of filling each node's task slots. Compute environments created earlier keep the previous formula.

Credentials​

  • Fixed the Azure credentials forms accepting Batch and Blob Storage account names that Azure rejects.
  • Fixed a Git credential with a repository base URL applying to other accounts whose names begin with the same text.
  • Fixed a Git credential with a repository base URL being ignored for a pipeline referenced by a short owner/repo name.

Datasets​

  • Fixed the dataset details page showing no avatar for a creator whose avatar comes from their identity provider.

Monitoring​

  • Fixed the run Parameters Groovy view producing text that is not valid Nextflow configuration. The view is now Nextflow config and downloads as a .config file.
  • Fixed the run Parameters JSON view showing a tree with unquoted keys instead of the JSON that Copy and Download produce.

Pipelines​

  • Fixed the Launchpad Recently updated sort ordering only the pipelines on the current page.
  • Fixed a launch form staying invalid with no explanation when the pipeline schema contained a pattern the browser cannot compile.

Projects​

  • Fixed the resource dialog clipping its pagination controls.

General​

  • Fixed icons in disabled buttons rendering at low contrast instead of following the button's color.
On this Page