October 2026 Summaries
15 posts from Octopus Deploy
Filter
Month:
Year:
Post Summaries
Back to Blog
Octopus modernized GitHub Actions for five microsites by replacing twenty separately maintained workflow files with reusable workflows housed in a central microsite-deployment repository, alongside Terraform and deployment scripts. Each microsite now uses short calling workflows that provide project-specific inputs while shared logic handles building, testing, packaging, deployment, and secure behavior for pull requests from forks. The migration was intentionally incremental, preserving separate pull request and main-branch workflows where their structures differed while extracting shared Node and pnpm version management into a composite action. The approach eliminated cross-repository workflow drift and reduced changes from multiple pull requests to one, but introduced decisions and limitations around versioning, testing, visibility, and governance. The team chose floating @main references so all consumers receive updates immediately, accepting the risk of broad breakages, while noting that tags and commit SHAs offer more predictability but require consumers to update and are difficult to manage independently per workflow. Testing branch-level changes, particularly when reusable workflows call composite actions, remains awkward, and GitHub provides limited native visibility into workflow consumers or version adoption. Dependabot can propose updates and organization policies can block insecure versions, but neither offers a gradual, centralized migration path, making early planning for references, testing, and consumer management important for platform teams.
Oct 28, 2026
3,077 words in the original blog post.
Platform engineering teams face an ongoing choice between mandatory and optional adoption, with recent research suggesting that mandatory platforms can achieve stronger goals when they provide genuinely useful capabilities and are supported by developer feedback. Platforms have expanded substantially, from a typical four features to roughly 24, and broader platforms tend to perform better when they solve meaningful developer problems such as security and compliance rather than merely accumulating functionality. Making adoption compulsory can reduce the feedback signal that reveals platform shortcomings, so teams should continuously measure developer satisfaction, maintain direct contact with users, and treat platforms as internal products using practices such as lean product management. Successful platforms also need to demonstrate organizational value through relevant metrics, particularly as AI investments may compete for budgets. The most effective platforms may become nearly invisible because they remove friction from developers’ daily work, making clear evidence of their return on investment essential for preserving long-term funding.
Oct 27, 2026
714 words in the original blog post.
Octopus Deploy 2026.3 can combine subscriptions, Slack integration, webhook-triggered runbooks, variables, and OIDC authentication to automatically alert responsible team members when a deployment reaches a manual intervention, preventing approvals or validation checkpoints from being overlooked in the user interface. A basic subscription can send batched Slack notices, while a more tailored approach sends the subscription event as a webhook to a runbook, which extracts intervention, project, environment, task, and responsible-team details from the JSON payload. The runbook uses a read-only service account and a Generic OIDC account to securely query Octopus without storing API keys, maps team members’ email addresses to Slack user IDs through a library variable set, and posts a message containing direct user mentions and a link to the relevant deployment or runbook task. The approach can be standardized across spaces using Terraform, requires the Octopus CLI on the worker, and can be tested by capturing webhook payloads and evaluating variable expressions before publishing the runbook. The same event-driven pattern can also support notifications and automation for guided failures, failed or successful deployments, expiring certificates, and external ITSM approval workflows.
Oct 26, 2026
3,601 words in the original blog post.
Octopus’s Pipeline Explorer is a project overview tool designed to help users understand unfamiliar deployment configurations by presenting pipelines and deployment processes as workflow diagrams on a single page. It visualizes how releases are triggered, move through environments, and vary by channel, while also exposing related settings such as lifecycles, deployment targets, credentials, and approvals that may otherwise be spread across multiple areas. The view can flag configuration conditions that could affect upcoming deployments, but it does not replace existing process, variables, or settings pages; instead, it serves as a navigational map built on the same project data and is available from a project’s navigation menu.
Oct 23, 2026
339 words in the original blog post.
Octopus Server 2026.4.6909 introduces multi-node support for Polling Tentacles in self-hosted high-availability clusters, eliminating the previous requirement for each Tentacle to poll every server node. Using a shared Redis-based queue, any cluster node can retrieve a Tentacle’s pending work and route its response back to the originating node, allowing Tentacles to use a single address or an optional TCP load balancer without risking missed deployment tasks. Messages are encrypted with the Octopus Master Key, while larger files remain in the cluster shared directory, and connectivity pages now aggregate logs across active nodes. Adoption requires the updated Octopus Server version on all nodes, a reachable Redis instance configured as a short-lived queue, a shared read/write cluster directory, and optionally a load balancer; existing Tentacle configurations can be migrated gradually.
Oct 21, 2026
743 words in the original blog post.
Apache Ant, a Java-based build tool first released in 2000, remains actively used in enterprise environments and can be integrated into Octopus Deploy without preinstalling Ant or Java on deployment workers. The process uses a sample Ant application alongside portable Apache Ant and Azul Zulu OpenJDK ZIP packages uploaded to Octopus’s built-in feed under its required package naming format. During an Octopus script deployment step on Windows workers, PowerShell extracts the referenced packages into the working directory, configures JAVA_HOME and the session Path for the bundled JDK and Ant binaries, verifies the Ant executable, and runs Ant commands to compile, package, and execute the application. This dependency-based approach supports consistent, repeatable deployments across workers while avoiding machine-level tooling installation.
Oct 20, 2026
660 words in the original blog post.
High-performing software delivery depends less on stated corporate values than on daily interactions that determine whether employees can safely share risks, concerns, and bad news. Drawing on Douglas McGregor’s motivation theory and Ron Westrum’s model of pathological, bureaucratic, and generative cultures, the discussion argues that generative organizations protect messengers, encourage cross-functional collaboration, and use early information to prevent failures. By contrast, intimidation, rigid hierarchies, and pressure to meet fixed scope, cost, and deadline commitments suppress information flow without changing underlying constraints, often leading to delayed discoveries, defects, rework, and escalating delivery pressure. Feedback from users, automated tools, and team members is presented as essential operational information that improves both employee experience and organizational outcomes.
Oct 20, 2026
1,137 words in the original blog post.
Controlled Delivery is presented as a user-centered approach to software releases that focuses on delivering meaningful value rather than merely pushing technical changes across an ecosystem. It combines composable content, which lets users adopt only the capabilities they need, mindful rollout practices that limit disruption and evaluate safety, compatibility, resilience, and usefulness, and verifiable receipts that trace signed, deterministic changes from development through deployment. The approach uses established CI/CD practices such as feature flags, staged or ring deployments, rollback mechanisms, testing across legacy environments, shadow releases, blue/green deployments, and direct user feedback, but applies them with active attention to how people experience and use changes. It is not intended as a standalone compliance system, unattended automation, or justification for constant releases; instead, organizations are encouraged to assess their existing delivery processes, connect mature tools and rituals into a predictable flow from code to user benefit, and build or purchase any missing capabilities.
Oct 19, 2026
1,216 words in the original blog post.
Octopus Deploy modernized the delivery pipelines for five Astro-based public microsites after identifying accumulated friction in duplicated GitHub Actions workflows and Octopus deployment processes, slow S3 uploads, an underused shared staging environment, difficult preview workflows, and manually managed AWS infrastructure. Although each issue was manageable independently, together they reduced consistency, slowed feedback, discouraged improvements, and made it difficult for technical and nontechnical contributors to validate changes before production. The team replaced copied workflows with reusable GitHub workflows, consolidated deployment processes into versioned Octopus templates, migrated hosting to Azure Blob Storage with AzCopy to enable faster full deployments, and eliminated the image-pruning workaround that had prevented clean deployments. Every pull request can now optionally receive an ephemeral hosted preview environment with a URL, supported by automated provisioning and deprovisioning through Terraform-based infrastructure practices. The effort also prompted broader discussion of platform-team ownership, standardized guardrails, execution containers, and the balance between shared processes and team flexibility, with lessons intended to apply beyond static websites.
Oct 15, 2026
3,164 words in the original blog post.
Octopus has introduced a built-in remote Model Context Protocol server that lets AI agents connect to an Octopus instance through its `/mcp` HTTP endpoint, eliminating the need to install and maintain a local MCP server on each user’s machine. Initially available to Octopus Cloud customers and planned for Octopus Server 2026.3, the feature supports permission-controlled access to core capabilities including navigating spaces, projects, tenants, and tags; checking deployment and pull request status; creating, deploying, and promoting releases; running runbooks; investigating task failures and configuration issues; and querying audit events and deployment health. Octopus plans to expand the remote MCP’s functionality based on user feedback while developing broader agent-governance features, including scoped API keys, OAuth authentication, agent-aware policies, and enhanced audit-log filtering for agent activity.
Oct 13, 2026
518 words in the original blog post.
Salesforce DX enables source-driven Salesforce development through Git-based workflows, CI/CD pipelines, and the Salesforce CLI, while Octopus Deploy can orchestrate deployments and manage associated variables and secrets. The process described uses a sample application uploaded as a package, a Salesforce Developer Edition account, and an SFDX authentication URL stored securely in Octopus as a secret variable, although JWT authentication is noted as a more automation-friendly alternative. An Octopus script step runs within a custom Node-based Docker container that installs the Salesforce CLI, configures environment variables for CI reliability, authenticates to Salesforce using the stored URL, and deploys the project with the `sf project deploy start` command. After deployment, the sample Lightning component can be added to a newly created Lightning app through Salesforce’s Lightning App Builder. Together, the Salesforce CLI handles platform interactions while Octopus Deploy provides deployment management, secret handling, and a foundation for automating Salesforce CD processes.
Oct 10, 2026
672 words in the original blog post.
Steve Fenton’s Continuous Delivery Office Hours discussion argues that large, “big-bang” rewrites of legacy software are usually risky because teams underestimate the institutional knowledge, edge cases, costs, and time embedded in long-running systems. Rather than viewing such systems as obsolete “legacy” code, the discussion recommends treating them as valuable “heritage” software that reflects years of domain learning and operational refinement. A replacement often appears cleaner only because it initially handles fewer scenarios, while rediscovered requirements eventually restore much of the original complexity and leave the organization behind competitors. As an alternative, the article advocates gradually replacing well-defined components through the strangler fig pattern, extracting functions from a monolith into independently maintainable services while the existing system continues operating. Domain-driven design and team topologies can help identify meaningful boundaries and assign teams ownership of specific domains, enabling more independent delivery, improved architecture, and lower-risk modernization.
Oct 06, 2026
883 words in the original blog post.
Octopus Deploy can use Auth0 as a generic OpenID Connect provider to authenticate users, requiring an Auth0 Regular Web Application configured with Octopus callback and optional logout URLs, followed by entering Auth0’s domain, client ID, and client secret in Octopus’s OpenID Connect settings. The configuration can optionally enable automatic user creation and uses the name claim for usernames. Because Auth0 does not automatically include roles in OIDC tokens, administrators must create Auth0 roles, assign users to them, and add a post-login Auth0 Action that inserts role memberships as a namespaced custom claim in the ID token. Octopus Deploy is then configured to use that custom claim as its role claim type, allowing Auth0 roles to be mapped to Octopus Teams and their associated permissions without adding individual users. For troubleshooting, the post recommends temporarily using jwt.io as an allowed callback URL and initiating an Auth0 authorization request to inspect the decoded token and verify that the expected roles claim is present.
Oct 05, 2026
1,155 words in the original blog post.
Octopus Deploy’s llms.txt feature provides a consolidated Markdown reference for its REST API, enabling AI coding agents to understand authentication, space scoping, endpoints, request types, deployment-step properties, and supported action settings without navigating extensive documentation. Using Claude Code with a local Octopus instance, the author demonstrated how an agent used this file to create realistic multi-tenant test data for a fictional retailer platform, including projects, deployment processes, releases, targets, variables, runbooks, and teams. The agent then built custom dashboards such as a tenant rollout board that identifies releases behind schedule, missing variables, and pending approvals, and a team health board that prioritizes failures and approval bottlenecks while showing deployment, target, and freeze status. It also produced a Sankey-style release-flow visualization connecting releases, environments, and tenants to reveal deployment patterns and anomalies. The post recommends starting with read-only API tasks, validating results in the Octopus portal, using API errors to refine requests, and protecting credentials by keeping API keys server-side behind a proxy rather than exposing them in browser-based dashboards.
Oct 02, 2026
1,707 words in the original blog post.
Octopus Deploy’s 2026 Future of Platform Engineering report, based on a survey of 379 platform practitioners and a literature review, describes platform engineering as a maturing mainstream discipline whose outcomes depend largely on organizational choices rather than technology alone. Organizations most commonly adopt platforms to improve automation and efficiency, making it important to define success measures that align with those motivations. Respondents identified unclear strategy as a more frequent barrier than technical complexity, particularly among teams building most features from scratch. The report found that mandatory adoption was associated with a higher rate of goal achievement than optional adoption, although this relationship also clustered with strong leadership engagement, job satisfaction, and budget confidence. It further suggests that AI improves software delivery speed and stability most for mature platforms with advanced capabilities such as ephemeral environments, cost controls, and code coverage, reinforcing the idea that AI amplifies existing platform foundations.
Oct 01, 2026
871 words in the original blog post.