Skip to content

Commit 8eacb47

Browse files
feat(kubernetes): add sidecar supervisor topology (#2076)
* feat(kubernetes): add sidecar supervisor topology Add the Kubernetes sidecar supervisor topology, its Helm/Skaffold configuration, topology documentation, and sidecar e2e matrix coverage. Skip root-only sandbox identity rewriting when process enforcement is network-only so the low-permission sidecar process container can start successfully. Signed-off-by: Taylor Mutch <[email protected]> * fix(supervisor): avoid similar process id names Signed-off-by: Taylor Mutch <[email protected]> * fix(supervisor): avoid similar process id names Signed-off-by: Taylor Mutch <[email protected]> * fix(sandbox): avoid similar proxy id names Signed-off-by: Taylor Mutch <[email protected]> * docs(kubernetes): clarify sidecar topology limits Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): keep sidecar process leaf capless Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): refresh sidecar provider env snapshots Signed-off-by: Taylor Mutch <[email protected]> * test(supervisor): align hot-swap identity regression Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): stage sidecar mtls files before proxy chown Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): simplify sidecar supervisor topology Signed-off-by: Taylor Mutch <[email protected]> * chore(helm): reuse sidecar skaffold values Signed-off-by: Taylor Mutch <[email protected]> * fix(supervisor): avoid similar iptables helper names Signed-off-by: Taylor Mutch <[email protected]> * fix(e2e): harden kube gateway wrapper setup Signed-off-by: Taylor Mutch <[email protected]> * fix(supervisor): avoid nft batch rollback on OCP Run nftables setup as individual commands so optional conntrack and log expressions can fail without rolling back required table, chain, and reject rules. Signed-off-by: Seth Jennings <[email protected]> Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): preserve process identity in sidecar topology Render sidecar pods with a shared process namespace, keep binary-aware network policy enabled, and move Kubernetes sidecar settings under the nested sidecar config table. Also apply unprivileged Landlock/seccomp setup in NetworkOnly supervisor mode so sidecar topology keeps sandbox child hardening without privileged process setup. Signed-off-by: Taylor Mutch <[email protected]> * refactor(kubernetes): replace sidecar snapshots with control socket Coordinate sidecar policy and provider bootstrap over a local Unix socket so the process leaf no longer reads policy/provider snapshot files. Report entrypoint startup through the control channel and keep gateway credentials confined to the network sidecar. Signed-off-by: Taylor Mutch <[email protected]> * feat(kubernetes): support relaxed sidecar network identity Signed-off-by: Taylor Mutch <[email protected]> * fix(sandbox): satisfy sidecar clippy lint Signed-off-by: Taylor Mutch <[email protected]> * refactor(kubernetes): standardize topology naming Signed-off-by: Taylor Mutch <[email protected]> * fix(sandbox): satisfy linux clippy timeout import Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): support kata sidecar on ipv4 pods Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): satisfy linux clippy for sidecar fallback Signed-off-by: Taylor Mutch <[email protected]> * chore(kubernetes): remove stale supervisor topology references Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): enable sidecar binary policy inspection Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): harden sidecar control boundary Signed-off-by: Taylor Mutch <[email protected]> * fix(kubernetes): couple sidecar supervisor lifecycles Signed-off-by: Taylor Mutch <[email protected]> --------- Signed-off-by: Taylor Mutch <[email protected]> Signed-off-by: Seth Jennings <[email protected]> Co-authored-by: Seth Jennings <[email protected]>
1 parent bebf440 commit 8eacb47

54 files changed

Lines changed: 5463 additions & 560 deletions

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎.agents/skills/debug-openshell-cluster/SKILL.md‎

Lines changed: 51 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -268,6 +268,57 @@ kubectl -n openshell get configmap openshell-config -o jsonpath='{.data.gateway\
268268
kubectl -n <sandbox-namespace> get sandbox <sandbox-name> -o jsonpath='{.spec.template.spec.serviceAccountName}{"\n"}'
269269
```
270270

271+
If `topology = "sidecar"` is rendered under `[openshell.drivers.kubernetes]`,
272+
sandbox pods should have an `openshell-network-init` init container running
273+
`--mode=network-init`, an `agent` container running
274+
`openshell-sandbox --mode=process`, and an `openshell-supervisor-network`
275+
container running `--mode=network`. The init container owns nftables setup and
276+
should be the only sidecar topology container with `NET_ADMIN`. It also needs
277+
`CHOWN`/`FOWNER` to hand shared emptyDir state to the effective sidecar UID. The
278+
default binary-aware network sidecar runs as UID 0 with primary GID
279+
`sandbox_gid` and adds `SYS_PTRACE` plus `DAC_READ_SEARCH`. When
280+
`process_binary_aware_network_policy = false`, it runs as the configured
281+
non-root `proxy_uid` without those inspection capabilities. The pod `fsGroup`
282+
is set to `sandbox_gid` in both modes.
283+
284+
In sidecar topology only the network sidecar should mount the gateway bootstrap
285+
credentials (`openshell-sa-token` and `openshell-client-tls`). The process
286+
container should not receive `OPENSHELL_ENDPOINT`, gateway TLS env vars, the
287+
sandbox token file, or those credential mounts. Instead, the network sidecar
288+
serves policy and provider environment state over the Unix control socket from
289+
`OPENSHELL_SIDECAR_CONTROL_SOCKET` (`/run/openshell-sidecar/control.sock` by
290+
default). The process supervisor must be the first and only client. After
291+
validating its peer UID, GID, and PID, the sidecar unlinks the listener. If the
292+
connection later closes, the network sidecar exits non-zero so Kubernetes can
293+
restart it with a fresh listener. If the process supervisor fails before
294+
launching the workload,
295+
inspect both containers for control-socket bind, connect, bootstrap, or update
296+
errors. If new SSH/exec sessions do not pick up refreshed provider environment,
297+
inspect the network sidecar settings-poll logs and the process container logs
298+
for provider environment update handling; the process container should consume
299+
newer provider-env revisions without receiving gateway credentials.
300+
301+
The process container reports the workload entrypoint PID over the same control
302+
socket, and the network sidecar uses that PID for binary-scoped policy
303+
decisions through `/proc`. If rules with `policy.binaries` are unexpectedly
304+
denied, inspect the sidecar control logs and confirm the pod has
305+
`shareProcessNamespace: true`.
306+
The shared state directory should preserve `sandbox_gid` inheritance
307+
(`02775`). Sidecar SSH uses the Linux abstract socket
308+
`@openshell-sidecar-ssh`; the network sidecar verifies its peer PID before
309+
bridging gateway relay requests. No `ssh.sock` file should appear in the shared
310+
state directory.
311+
Inspect all three when sandbox registration or egress enforcement fails:
312+
313+
```bash
314+
kubectl -n openshell get configmap openshell-config -o jsonpath='{.data.gateway\.toml}' | grep -E '^\[openshell\.drivers\.kubernetes\]|^topology\s*='
315+
kubectl -n <sandbox-namespace> get pod <sandbox-pod> -o jsonpath='{range .spec.initContainers[*]}{.name}{" "}{.command}{"\n"}{end}'
316+
kubectl -n <sandbox-namespace> get pod <sandbox-pod> -o jsonpath='{range .spec.containers[*]}{.name}{" "}{.command}{"\n"}{end}'
317+
kubectl -n <sandbox-namespace> logs <sandbox-pod> -c openshell-network-init --tail=200
318+
kubectl -n <sandbox-namespace> logs <sandbox-pod> -c openshell-supervisor-network --tail=200
319+
kubectl -n <sandbox-namespace> logs <sandbox-pod> -c agent --tail=200
320+
```
321+
271322
### Step 6: Check VM-Backed Gateways
272323

273324
Use the VM driver logs and host diagnostics available in the user's environment. Verify:

‎.agents/skills/helm-dev-environment/SKILL.md‎

Lines changed: 27 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -60,9 +60,25 @@ mise run helm:skaffold:dev
6060
mise run helm:skaffold:run
6161
```
6262

63+
**Supervisor sidecar topology** (build once and leave running):
64+
```bash
65+
mise run helm:skaffold:run:sidecar
66+
```
67+
68+
**Supervisor sidecar topology with TLS/mTLS enabled** (build once and leave running):
69+
```bash
70+
mise run helm:skaffold:run:sidecar-mtls
71+
```
72+
6373
Both commands build the `gateway` and `supervisor` images and deploy the OpenShell Helm
64-
chart. The `pkiInitJob` hook (a pre-install Job that runs `openshell-gateway generate-certs`)
65-
generates mTLS secrets on first install. Envoy Gateway opt-in; see the Optional Add-ons section below.
74+
chart. The sidecar profile renders an `openshell-network-init` init container for
75+
nftables setup and an `openshell-supervisor-network` runtime sidecar for proxying.
76+
Binary-aware policy mode runs that sidecar as UID 0 with `SYS_PTRACE` and
77+
`DAC_READ_SEARCH`; relaxed mode can run it as the configured proxy UID. The
78+
sidecar-mTLS profile reuses `ci/values-sidecar.yaml` and restores
79+
`server.disableTls=false` inline for Skaffold. The `pkiInitJob` hook (a pre-install
80+
Job that runs `openshell-gateway generate-certs`) generates mTLS secrets on first
81+
install. Envoy Gateway opt-in; see the Optional Add-ons section below.
6682

6783
The gateway Service uses ClusterIP. Access is via Envoy Gateway (port `8080`) or `kubectl port-forward`.
6884

@@ -74,7 +90,8 @@ create the Secret named `openshell-ha-pg` with a `uri` key, then run
7490
### TLS behaviour
7591

7692
`ci/values-skaffold.yaml` sets `server.disableTls: true`, so Skaffold-based deploys run
77-
plaintext by default. To test with TLS enabled, comment out that line and redeploy.
93+
plaintext by default. To test sidecar topology with TLS enabled, use
94+
`mise run helm:skaffold:run:sidecar-mtls`.
7895

7996
| Mode | `server.disableTls` | Gateway scheme |
8097
|------|---------------------|----------------|
@@ -126,6 +143,12 @@ openshell sandbox list --gateway-endpoint https://localhost:8090
126143
mise run helm:skaffold:delete
127144
```
128145

146+
For a sidecar-profile deployment:
147+
148+
```bash
149+
mise run helm:skaffold:delete:sidecar
150+
```
151+
129152
### Delete the cluster entirely
130153

131154
```bash
@@ -250,6 +273,7 @@ for dependencies still declared in `Chart.yaml`.
250273
| `deploy/helm/openshell/ci/values-gateway.yaml` | Envoy Gateway GRPCRoute + Gateway overlay |
251274
| `deploy/helm/openshell/ci/values-high-availability.yaml` | HA test overlay (`replicaCount: 2` with external PostgreSQL Secret) |
252275
| `deploy/helm/openshell/ci/values-keycloak.yaml` | Keycloak OIDC overlay |
276+
| `deploy/helm/openshell/ci/values-sidecar.yaml` | Supervisor sidecar topology overlay for Kubernetes e2e/dev |
253277
| `deploy/helm/openshell/ci/values-spire.yaml` | SPIFFE/SPIRE provider token grant overlay |
254278
| `deploy/helm/openshell/ci/values-spire-stack.yaml` | SPIRE hardened chart values for local dev |
255279
| `deploy/helm/openshell/ci/values-tls-disabled.yaml` | Lint-only: TLS + auth disabled (reverse-proxy edge termination) |

‎.github/workflows/branch-e2e.yml‎

Lines changed: 10 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -121,16 +121,25 @@ jobs:
121121
include:
122122
- agent_sandbox_api: v1beta1
123123
agent_sandbox_version: v0.5.0
124+
topology: combined
125+
extra_helm_values: ""
124126
- agent_sandbox_api: v1alpha1
125127
agent_sandbox_version: v0.4.6
128+
topology: combined
129+
extra_helm_values: ""
130+
- agent_sandbox_api: v1beta1
131+
agent_sandbox_version: v0.5.0
132+
topology: sidecar
133+
extra_helm_values: deploy/helm/openshell/ci/values-sidecar.yaml
126134
permissions:
127135
contents: read
128136
packages: read
129137
uses: ./.github/workflows/e2e-kubernetes-test.yml
130138
with:
131139
image-tag: ${{ github.sha }}
132-
job-name: Kubernetes E2E (Rust smoke, Agent Sandbox ${{ matrix.agent_sandbox_api }})
140+
job-name: Kubernetes E2E (Rust smoke, ${{ matrix.topology }}, Agent Sandbox ${{ matrix.agent_sandbox_api }})
133141
agent-sandbox-version: ${{ matrix.agent_sandbox_version }}
142+
extra-helm-values: ${{ matrix.extra_helm_values }}
134143

135144
kubernetes-ha-e2e:
136145
needs: [pr_metadata, build-gateway, build-supervisor]

‎Cargo.lock‎

Lines changed: 3 additions & 0 deletions
Some generated files are not rendered by default. Learn more about customizing how changed files appear on GitHub.

‎architecture/build.md‎

Lines changed: 5 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -91,10 +91,11 @@ Runtime layout:
9191
as a release artifact. Linux GNU VM driver binaries must not reference
9292
`GLIBC_*` symbols newer than `GLIBC_2.28`; release workflows verify this
9393
before publishing artifacts.
94-
- **Supervisor**: `scratch` base, static musl binary at `/openshell-sandbox`.
95-
Static linkage is required because the image is mounted/extracted into
96-
sandbox environments (Docker extraction, Podman image volumes, Kubernetes
97-
init-container copy-self) and cannot rely on a dynamic loader.
94+
- **Supervisor**: Alpine base with `nftables`, static musl binary at
95+
`/openshell-sandbox`. Static linkage keeps the binary usable when the image
96+
is mounted/extracted into sandbox environments (Docker extraction, Podman
97+
image volumes, Kubernetes init-container copy-self), while `nftables` supports
98+
Kubernetes supervisor sidecar egress enforcement.
9899

99100
Gateway image builds bake the corresponding supervisor image tag into the
100101
gateway binary so Docker sandboxes do not depend on `:latest` by default.

‎architecture/compute-runtimes.md‎

Lines changed: 39 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -81,14 +81,52 @@ The supervisor must be available inside each sandbox workload:
8181
|---|---|
8282
| Docker | Bind-mounted local supervisor binary, or a binary extracted from the configured supervisor image. |
8383
| Podman | Read-only OCI image volume containing the supervisor binary. |
84-
| Kubernetes | Sandbox pod image or pod template configuration. |
84+
| Kubernetes | Supervisor image side-loaded into the sandbox pod by image volume or init container. |
8585
| VM | Embedded in the guest rootfs bundle. |
8686
| Extension | Defined by the out-of-tree driver. |
8787

8888
Driver-controlled environment variables must override sandbox image or template
8989
values for sandbox ID, sandbox name, gateway endpoint, relay socket path, TLS
9090
paths, and command metadata.
9191

92+
Kubernetes can run the supervisor in the default combined topology or in a
93+
sidecar topology. Combined mode keeps network and process supervision in the
94+
agent container. Sidecar mode runs network enforcement, the proxy, and gateway
95+
session in a dedicated sidecar, while the agent container runs only the
96+
process-supervision leaf and launches the user workload after the sidecar
97+
serves bootstrap state over a local control socket. The network sidecar owns
98+
gateway credentials and sends policy plus workload-facing provider environment
99+
state to the process leaf over that socket. It also streams provider
100+
environment updates after settings polls so future process sessions see
101+
updated provider env without giving the process leaf gateway access. The
102+
pre-workload process supervisor is the only accepted control client: the
103+
network sidecar verifies its UID, GID, and PID with peer credentials, removes
104+
the listener after accepting it, and ignores workload-supplied relay targets.
105+
SSH relays use a Linux abstract socket and verify its peer PID against that
106+
authenticated process-supervisor connection, so workload filesystem access
107+
cannot replace the relay endpoint. Either supervisor exits when this control
108+
connection closes. This couples their restart lifecycle and prevents a workload
109+
that survives an isolated network-sidecar restart from becoming the next
110+
authoritative control client. In sidecar mode, an init container performs the
111+
privileged pod-network nftables setup with
112+
`NET_ADMIN`. The default binary-aware network sidecar runs as UID 0 without
113+
`NET_ADMIN` and adds `SYS_PTRACE` plus `DAC_READ_SEARCH` so it can resolve
114+
cross-UID workload process/binary identity through shared `/proc`. Operators
115+
can set the sidecar `process_binary_aware_network_policy` flag false to run the
116+
sidecar as the configured non-root proxy UID, omit both inspection capabilities,
117+
and downgrade network policy to endpoint/L7 matching without `policy.binaries`.
118+
The init path applies nftables as individual commands so optional conntrack and
119+
log expressions can fail without rolling back the required table, chain, and
120+
reject rules.
121+
The agent container runs as the resolved sandbox UID/GID with no added Linux
122+
capabilities. Sidecar mode preserves gateway session and SSH behavior, but
123+
treats the process leaf as network-only: Landlock filesystem policy and child
124+
seccomp still apply where supported, while process privilege dropping and
125+
supervisor identity mount isolation do not run because the agent container is
126+
already unprivileged. Sidecar pods use a shared process namespace so the
127+
network sidecar can resolve workload process and binary identity through
128+
`/proc/<entrypoint-pid>`.
129+
92130
## Images
93131

94132
The gateway image and Helm chart are built from this repository. Sandbox images

‎crates/openshell-core/src/grpc_client.rs‎

Lines changed: 6 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -167,9 +167,14 @@ async fn build_plain_channel(endpoint: &str) -> Result<Channel> {
167167
.into_diagnostic()
168168
.wrap_err_with(|| format!("failed to read client key from {key_path}"))?;
169169

170-
let tls_config = ClientTlsConfig::new()
170+
let mut tls_config = ClientTlsConfig::new()
171171
.ca_certificate(Certificate::from_pem(ca_pem))
172172
.identity(Identity::from_pem(cert_pem, key_pem));
173+
if let Ok(server_name) = std::env::var(sandbox_env::GATEWAY_TLS_SERVER_NAME)
174+
&& !server_name.is_empty()
175+
{
176+
tls_config = tls_config.domain_name(server_name);
177+
}
173178

174179
ep = ep
175180
.tls_config(tls_config)

‎crates/openshell-core/src/provider_credentials.rs‎

Lines changed: 90 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -63,6 +63,63 @@ impl ProviderCredentialState {
6363
}
6464
}
6565

66+
/// Build a static provider state from an already-prepared child
67+
/// environment snapshot.
68+
///
69+
/// Kubernetes sidecar topology uses this in the process-only supervisor:
70+
/// the network sidecar owns provider credential resolvers and sends the
71+
/// workload-facing env map over a local control channel. The process leaf
72+
/// must inject that map into child processes without re-placeholderizing it
73+
/// or holding the gateway-side resolver material.
74+
pub fn from_child_env_snapshot(revision: u64, child_env: HashMap<String, String>) -> Self {
75+
let snapshot = Arc::new(ProviderCredentialSnapshot {
76+
revision,
77+
child_env,
78+
dynamic_credentials: HashMap::new(),
79+
});
80+
81+
Self {
82+
inner: Arc::new(RwLock::new(ProviderCredentialStateInner {
83+
current: snapshot,
84+
generations: VecDeque::new(),
85+
current_resolver: None,
86+
combined_resolver: None,
87+
suppressed_keys: HashSet::new(),
88+
})),
89+
}
90+
}
91+
92+
/// Install an already-prepared child environment snapshot.
93+
///
94+
/// This is intentionally narrower than [`Self::install_environment`]: it
95+
/// updates only the workload-facing env map and clears resolver state so a
96+
/// process that does not own gateway/provider resolver material can still
97+
/// pick up refreshed provider env for future child processes.
98+
pub fn install_child_env_snapshot(
99+
&self,
100+
revision: u64,
101+
mut child_env: HashMap<String, String>,
102+
) -> usize {
103+
let mut inner = self
104+
.inner
105+
.write()
106+
.expect("provider credential state poisoned");
107+
108+
for key in &inner.suppressed_keys {
109+
child_env.remove(key);
110+
}
111+
112+
inner.current = Arc::new(ProviderCredentialSnapshot {
113+
revision,
114+
child_env,
115+
dynamic_credentials: HashMap::new(),
116+
});
117+
inner.generations.clear();
118+
inner.current_resolver = None;
119+
inner.combined_resolver = None;
120+
inner.current.child_env.len()
121+
}
122+
66123
pub fn snapshot(&self) -> Arc<ProviderCredentialSnapshot> {
67124
self.inner
68125
.read()
@@ -594,6 +651,39 @@ mod tests {
594651
);
595652
}
596653

654+
#[test]
655+
fn child_env_snapshot_install_updates_env_without_resolver_material() {
656+
let state = ProviderCredentialState::from_child_env_snapshot(
657+
1,
658+
HashMap::from([
659+
("GITHUB_TOKEN".to_string(), "old".to_string()),
660+
("GCE_METADATA_HOST".to_string(), "marker".to_string()),
661+
]),
662+
);
663+
state.remove_env_key("GCE_METADATA_HOST");
664+
665+
let env_count = state.install_child_env_snapshot(
666+
2,
667+
HashMap::from([
668+
("GITHUB_TOKEN".to_string(), "new".to_string()),
669+
("GCE_METADATA_HOST".to_string(), "marker".to_string()),
670+
]),
671+
);
672+
673+
let snapshot = state.snapshot();
674+
assert_eq!(snapshot.revision, 2);
675+
assert_eq!(env_count, 1);
676+
assert_eq!(
677+
snapshot.child_env.get("GITHUB_TOKEN").map(String::as_str),
678+
Some("new")
679+
);
680+
assert!(!snapshot.child_env.contains_key("GCE_METADATA_HOST"));
681+
assert!(
682+
state.resolver().is_none(),
683+
"child-env snapshots must not install provider resolver material"
684+
);
685+
}
686+
597687
#[test]
598688
fn stale_generation_falls_back_to_current_credential_after_retention_window() {
599689
let state = ProviderCredentialState::from_environment(

‎crates/openshell-core/src/sandbox_env.rs‎

Lines changed: 28 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -29,6 +29,34 @@ pub const SANDBOX_COMMAND: &str = "OPENSHELL_SANDBOX_COMMAND";
2929
/// Deployment-controlled telemetry toggle propagated to the sandbox supervisor.
3030
pub const TELEMETRY_ENABLED: &str = "OPENSHELL_TELEMETRY_ENABLED";
3131

32+
/// Supervisor pod/runtime topology. Kubernetes sidecar mode sets this to
33+
/// `"sidecar"`; the default combined supervisor path omits it.
34+
pub const SUPERVISOR_TOPOLOGY: &str = "OPENSHELL_SUPERVISOR_TOPOLOGY";
35+
36+
/// Network enforcement backend selected by the compute driver.
37+
pub const NETWORK_ENFORCEMENT_MODE: &str = "OPENSHELL_NETWORK_ENFORCEMENT_MODE";
38+
39+
/// Whether network policy evaluation must bind requests to the peer binary.
40+
///
41+
/// The default when unset is `"required"`. Kubernetes sidecar experiments may
42+
/// set this to `"relaxed"` to enforce endpoint and L7 policy without per-binary
43+
/// `/proc` identity binding.
44+
pub const NETWORK_BINARY_IDENTITY: &str = "OPENSHELL_NETWORK_BINARY_IDENTITY";
45+
46+
/// Unix socket used by Kubernetes sidecar topology for local coordination.
47+
///
48+
/// The network sidecar owns gateway credentials and serves policy/provider
49+
/// state over this socket instead of exposing gateway credentials to the agent
50+
/// container.
51+
pub const SIDECAR_CONTROL_SOCKET: &str = "OPENSHELL_SIDECAR_CONTROL_SOCKET";
52+
53+
/// Optional TLS server name override used when connecting to the gateway.
54+
pub const GATEWAY_TLS_SERVER_NAME: &str = "OPENSHELL_GATEWAY_TLS_SERVER_NAME";
55+
56+
/// Directory where the network supervisor writes the proxy CA files consumed
57+
/// by workload child processes.
58+
pub const PROXY_TLS_DIR: &str = "OPENSHELL_PROXY_TLS_DIR";
59+
3260
/// Path to the CA certificate for mTLS communication with the gateway.
3361
pub const TLS_CA: &str = "OPENSHELL_TLS_CA";
3462

0 commit comments

Comments
 (0)