Files
coder/codersdk/agentsdk/google.go
T
Michael Suchacz e5707a13d6 feat: support multiple agents with shared instance-identity auth (#24325)
> This PR was authored by Mux on behalf of Mike.

## Summary

Adds support for multiple peer root workspace agents sharing the same
`auth_instance_id`, so AWS, Azure, and GCP instance-identity auth can
issue the correct session token for a selected agent instead of assuming
a
single root agent per instance.

## Problem

When a Terraform template attaches two or more `coder_agent` resources
(with `auth = "aws-instance-identity"`) to a single compute instance,
every agent shares the same cloud instance ID. The existing singular
lookup picks whichever agent was created most recently, silently
ignoring
the others.

## Solution

Introduce an optional pre-auth agent selector (`CODER_AGENT_NAME`) and
make the server-side lookup ambiguity-aware.

**Database layer:**
- `GetWorkspaceAgentsByInstanceID` (`:many`): returns all matching root
  agents for an instance ID.
- `GetWorkspaceAgentByInstanceIDAndName` (`:one`): returns the named
root
  agent for disambiguation.

**SDK and CLI:**
- `agent_name` field added to AWS, Azure, and GCP request structs
  (`omitempty` for backward compatibility).
- `CODER_AGENT_NAME` env var and `--agent-name` flag wired into the
agent
  bootstrap before instance-identity auth runs.

**Server handler (`handleAuthInstanceID`):**
- When `agent_name` is present: direct lookup by (instance ID, name).
- When absent: legacy lookup, then resource-scoped ambiguity check.
  Returns 409 with available agent names if multiple root agents match.
- Whitespace-only names are trimmed and treated as unspecified.
- Sub-agents remain excluded (`parent_id IS NULL` filter).

**Verification template:**
- `examples/templates/aws-multi-agent/` provisions one EC2 instance with
  two agents (`main` and `dev`), both using instance-identity auth with
  `CODER_AGENT_NAME` set in the cloud-init user data.

## Backward compatibility

Existing single-agent deployments work unchanged. The `agent_name` field
is optional with `omitempty`, and the unnamed path preserves today's
behavior when only one root agent matches.
2026-04-16 13:59:09 +02:00

80 lines
2.7 KiB
Go

package agentsdk
import (
"context"
"encoding/json"
"fmt"
"net/http"
"cloud.google.com/go/compute/metadata"
"golang.org/x/xerrors"
"github.com/coder/coder/v2/codersdk"
)
type GoogleInstanceIdentityToken struct {
JSONWebToken string `json:"json_web_token" validate:"required"`
// AgentName optionally selects a specific agent when multiple
// agents share the same instance identity. An empty string is
// treated as unspecified.
AgentName string `json:"agent_name,omitempty"`
}
// GoogleSessionTokenExchanger exchanges a Google instance JWT document for a Coder session token.
// @typescript-ignore GoogleSessionTokenExchanger
type GoogleSessionTokenExchanger struct {
serviceAccount string
gcpClient *metadata.Client
client *codersdk.Client
agentName string
}
func WithGoogleInstanceIdentity(serviceAccount string, gcpClient *metadata.Client, opts ...InstanceIdentityOption) SessionTokenSetup {
cfg := applyInstanceIdentityOptions(opts)
return func(client *codersdk.Client) RefreshableSessionTokenProvider {
return &InstanceIdentitySessionTokenProvider{
TokenExchanger: &GoogleSessionTokenExchanger{
client: client,
gcpClient: gcpClient,
serviceAccount: serviceAccount,
agentName: cfg.AgentName,
},
}
}
}
// exchange uses the Google Compute Engine Metadata API to fetch a signed JWT, and exchange it for a session token for a
// workspace agent.
//
// The requesting instance must be registered as a resource in the latest history for a workspace.
func (g *GoogleSessionTokenExchanger) exchange(ctx context.Context) (AuthenticateResponse, error) {
if g.serviceAccount == "" {
// This is the default name specified by Google.
g.serviceAccount = "default"
}
gcpClient := metadata.NewClient(g.client.HTTPClient)
if g.gcpClient != nil {
gcpClient = g.gcpClient
}
// "format=full" is required, otherwise the responding payload will be missing "instance_id".
jwt, err := gcpClient.Get(fmt.Sprintf("instance/service-accounts/%s/identity?audience=coder&format=full", g.serviceAccount))
if err != nil {
return AuthenticateResponse{}, xerrors.Errorf("get metadata identity: %w", err)
}
// request without the token to avoid re-entering this function
res, err := g.client.RequestWithoutSessionToken(ctx, http.MethodPost, "/api/v2/workspaceagents/google-instance-identity", GoogleInstanceIdentityToken{
JSONWebToken: jwt,
AgentName: g.agentName,
})
if err != nil {
return AuthenticateResponse{}, err
}
defer res.Body.Close()
if res.StatusCode != http.StatusOK {
return AuthenticateResponse{}, codersdk.ReadBodyAsError(res)
}
var resp AuthenticateResponse
return resp, json.NewDecoder(res.Body).Decode(&resp)
}