Administering Agent Mesh
This section is the operator's day-two home — what you do once Solace Agent Mesh is running. First-time setup (install commands, initial configuration, choosing a deploy target) lives in Install and Deploy; the pages here cover operating, securing, observing, and troubleshooting a live deployment.
Before you move your Agent Mesh into production, we recommend that you review the sections that follow.
Production Checklist
-
Production Readiness Checklist—We strongly recommend that you use this pre-production checklist prior to moving to production.
Security and Access
-
Enabling Single Sign-On (SSO)—Configure OpenID Connect (OIDC) single sign-on: the provider catalog, callback URIs, and the login flow.
-
Managing API Keys—Issue, use, rotate, and revoke API keys so that pipelines can run
sam configandsam eval runas a system user without a browser login. -
Secure User-Delegated Tool Access—Let each user act on remote systems through their own OAuth credentials.
-
Managing Secrets—Where secrets are read from, the substitution syntax that wires them into YAML, and how to rotate each one without downtime.
-
Configuring TLS—TLS for the inbound listener, the event broker connection, and outbound OIDC, large language model (LLM), and Model Context Protocol (MCP) traffic.
-
Managing Audit and Compliance—The audit event types, the closed schema, and how to ship audit records to your log aggregator.
Observability and Feedback
-
Monitoring Your Agent Mesh—Operational logs, OpenTelemetry metrics,
traceIDcorrelation, and health endpoints. -
Collecting and Publishing User Feedback—Collect end-user feedback, publish it to the event mesh, and understand how long Agent Mesh keeps it.
-
Configuring Notifications (Early Access)—Turn the notification service on or off, choose its database and retention, and deliver notifications to Slack, Microsoft Teams, webhooks, and connectors.
-
Managing Model Spend with Budgets—Set monthly limits on model spend for all of Agent Mesh and for each user, assign users to budgets, and review cost and token usage.
Data and Backups
-
Managing Backups and Data Retention—What to back up, how log rotation and the data-retention sweep work, and what is delegated to the underlying storage.
The Event Broker Layer
-
The Event Mesh Communication Layer—The event broker layer Agent Mesh runs on: the runtime connection block, the namespace tenant fence, and the durable queues that appear on the event broker.
Runbooks and Troubleshooting
-
Operator Workflows—End-to-end walkthroughs: rotating an LLM credential without downtime, staging a pre-upgrade dry-run, and scaling the Agent-Workflow Executor.
-
Troubleshooting a Running Deployment—Live failures organized by what broke, each with symptoms, diagnostics, resolution, and prevention.