* fix(supervisor): report rejected OCI working directories as WorkspaceValidationFailed
The RFC 0012 split dropped the supervisor exit status that drivers map to
the WorkspaceValidationFailed condition. An image WORKDIR the sandbox
identity cannot use then surfaced as ControlSupervisorStartFailed on
Docker and as a signal kill on Podman.
The supervisor now exits with the reserved status when the sandbox rejects
the image working directory. Docker maps that supervisor exit during
readiness and monitoring, and Podman maps the supervisor companion's exit
instead of the workload's.
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* feat(podman): honor OCI image working directories
Podman sandboxes now use a custom OCI image WORKDIR as the workspace and
as the working directory for agent commands, matching Docker. Empty, /,
and /sandbox values keep the managed /sandbox workspace volume.
A custom workspace stays in the image's container filesystem: no
workspace volume, no archive upload, and no root setup step. Resolve the
image ID, user, environment, and working directory from one pinned
inspection, validate the workdir with the shared OCI rules, and reject
Podman control-path overlaps and image volumes or driver mounts that
cover it. The runtime starts from / and passes the resolved path to agent
commands with --workdir.
Add podman_oci_identity to the Podman CI test list.
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* fix(drivers): preserve custom workspace failures across supervisor cleanup
Check the Podman custom workspace rather than the absent managed volume in the OCI identity E2E. Classify Bollard wait errors by exit code and record Docker workspace failures before removing the supervisor so readiness retains the specific failure reason.
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* fix(docker): use sandbox ID for recorded startup failures
Keep the workload container ID for Docker inspection and use the sandbox ID to retrieve the monitor failure. Test with distinct identifiers so the lookup cannot accidentally pass with a container ID.
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* refactor(podman): keep WORKDIR change focused on workspace behavior
Remove the new cross-driver workspace failure classification path and its monitor/readiness workaround. Preserve Podman WORKDIR resolution, mount safety, and workspace rejection without requiring a distinct condition reason.
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* refactor(podman): simplify inspected image metadata and test fixtures
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* code review
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
---------
Signed-off-by: Matthew Grossman <mgrossman@nvidia.com>
* feat(server): place supervisor sessions by consistent hash and hand them off on shutdown
Signed-off-by: divesh <dgude@nvidia.com>
* feat: route long-lived sandbox connections to the owning gateway replica
Signed-off-by: divesh <dgude@nvidia.com>
* feat(helm): add opt-in per-replica GRPCRoute routing
Signed-off-by: divesh <dgude@nvidia.com>
---------
Signed-off-by: divesh <dgude@nvidia.com>
* fix(ssh): persist sandbox host identities
Store each sandbox's Ed25519 host key in the gateway credential store and
deliver it only to the supervisor. Preserve identity across restarts,
delete owned credentials with the sandbox, and expose the public SHA256
fingerprint through sandbox and SSH-session APIs and client SDKs.
Cover credential ownership, cancellation, deletion retries, client
compatibility, and pinned SSH connections through lifecycle transitions.
Closes#3835
Signed-off-by: Mike Nguyen <miken@nvidia.com>
* fix(compute): clean up failed sandbox SSH identity creation
Signed-off-by: Mike Nguyen <miken@nvidia.com>
* test(ssh): wait for sandbox deletion before name reuse
Signed-off-by: Mike Nguyen <miken@nvidia.com>
---------
Signed-off-by: Mike Nguyen <miken@nvidia.com>
* feat(server): add gateway capacity metrics and optional HPA
Expose per-replica supervisor sessions, pending relay capacity, relay
rejections and claim latency, and outbound peer request outcomes and
latency. Use bounded labels and Prometheus histograms for the new latency
metrics while preserving existing summary metrics.
Add an optional Helm HPA with external-database and resource validation,
conservative scale-down defaults, and support for custom metrics. Keep
certificate hook pods outside gateway workload selectors.
Document per-pod scraping, scaling limits, upgrade behavior, and
PostgreSQL connection sizing.
Part of #3528
Signed-off-by: Emilien Macchi <emacchi@redhat.com>
* feat(server): unify routing metrics and clarify replica capacity
Combine local relay setup and outbound peer requests in one counter, labeled by operation, route, target, outcome, and gRPC status. Preserve peer latency metrics.
Rename the rejection reason from global_capacity to replica_capacity to reflect the per-replica relay budget. Keep capacity limits unchanged.
Update tests and documentation.
Signed-off-by: divesh <dgude@nvidia.com>
* fix(server): align routed request metric labels and outcomes
Count a local relay as successful only when its supervisor claims it, the
same event that answers a peer relay on the owner, and rename the outcomes
to local_error and remote_error so they say where an attempt failed. Label
the peer latency histogram by operation, like the routed request counter,
and rename the target label to relay_kind so it does not read as the
Prometheus scrape target.
Rename RelayCapacity.global to per_replica, drop the per-sandbox relay
capacity gauge, which no per-sandbox series can pair with, describe the
latency histogram as peer-only, and restore the note that unavailable
spikes are expected during rollouts.
Part of #3528
Signed-off-by: Emilien Macchi <emacchi@redhat.com>
---------
Signed-off-by: Emilien Macchi <emacchi@redhat.com>
Signed-off-by: divesh <dgude@nvidia.com>
Co-authored-by: divesh <dgude@nvidia.com>
* feat(gateway): validate VM filesystem tools during preflight
Check required local VM tools through config preflight and share executable
resolution with VM image operations. Report selected paths and actionable
errors without creating gateway or sandbox state.
Bound probe output and execution time, and clean up probe descendants on
interruption. Preserve pure static validation and skip local tool checks
for remote driver endpoints and unrelated drivers.
Fixes#3951
Related to #3955
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(gateway): stabilize filesystem preflight checks
Combine identical filesystem-tool error arms and normalize rendered
diagnostics in command tests so terminal wrapping preserves assertions.
Describe driver TLS validation without depending on removed guest fields.
Signed-off-by: Shiju <shiju@nvidia.com>
* test(gateway): serialize preflight fixture paths as TOML
Keep temporary paths quoted and escaped through the TOML serializer
instead of relying on Rust Debug formatting.
Signed-off-by: Shiju <shiju@nvidia.com>
---------
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(vm): enforce the configured workload identity
Reject conflicting policy users and groups before VM image preparation
and before guest attach or process startup changes state. Validate
supervisor policy updates against the protected VM workload identity.
Preserve the gateway CA transport and capability-free sandbox launcher.
Signed-off-by: Shiju <shiju@nvidia.com>
* test(sandbox): clarify VM identity rejection fixtures
Name invalid user and group fixtures distinctly and move the final
workload identity into its group mismatch test.
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(supervisor): align VM identity startup with current APIs
Pass the optional rejection-log key for VM identity failures and keep
generic startup-write regressions free of VM identity constraints.
Repair the call sites after the branch rebase so the identity and cleanup
proposals compile against the current startup helpers.
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(vm): restore inactive sandbox workload identity
Recover the persisted overlay owner before publishing stopped and terminal
sandboxes. Keep resources manageable when identity metadata is invalid.
Clarify fixed MicroVM ownership in policy-generation guidance.
Signed-off-by: Shiju <shiju@nvidia.com>
* test(vm): flush identity fixture before restart
Persist the canonical identity file before readiness and report the observed
exec, canonical and file-owner identities before comparing them.
Signed-off-by: Shiju <shiju@nvidia.com>
---------
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(providers): restore supervisor-backed GCP metadata discovery
Relay the reserved metadata endpoint to the supervisor and restore project, account, and placeholder token responses from live provider state. Cover Google SDK discovery and repeated refresh with provider E2E tests.
Closes#3860
Signed-off-by: Philippe Martin <phmartin@redhat.com>
* fix(providers): preserve default metadata account without an email
Use the default account identifier when the optional service account email is missing or empty. Cover repeated SDK refresh for missing, empty, and configured email values with distinct providers for parallel E2E execution.
Signed-off-by: Philippe Martin <phmartin@redhat.com>
* test(sandbox): fix metadata relay lint and timeout
Signed-off-by: Philippe Martin <phmartin@redhat.com>
---------
Signed-off-by: Philippe Martin <phmartin@redhat.com>
Since #2726 the canonical main process's stdout and stderr are captured
in pipes that feed only the in-memory replay buffer used by sandbox
connect. Agent output therefore never reaches the container's own stdout
and stderr, so it is missing from kubectl logs, docker logs, and podman
logs and from anything that collects container logs. Before #2726 the
entrypoint inherited the container's descriptors and its output appeared
there.
Copy the main process's output to the launcher's stdout and stderr in
addition to the replay buffer, restoring the earlier behavior:
- Output is copied byte for byte to the matching stream from a
forwarder thread per stream, after it is published to the replay
buffer. When the container runtime falls behind on a stream, that
stream's reader waits instead of dropping output, so backpressure
reaches the agent as it did with inherited descriptors, while the
other stream and attachments keep receiving output.
- Before the main process's exit is published, the output readers
finish and queued output is drained to the container log, so an
agent's final lines are not lost at shutdown. A 30 second deadline
covers both; when it expires, readers waiting on the container log
are released and drain the pipes into the replay buffer only, so a
stalled container log cannot block exit reporting.
- PTY-mode processes are not copied. The terminal stream carries escape
sequences and echoed input, and terminal commands never reached the
container log before #2726.
- Exec, SSH, and SFTP sessions are not copied.
Launcher log lines keep their existing format and remain in the
container's stderr. They are written as whole lines, and a newline is
inserted first when the agent left stderr mid-line, so launcher and
agent lines do not merge.
The Docker and VM drivers appended the tail of the workload's output to
failure messages: Docker the workload container's log, and the VM driver
the guest console, which carries the launcher's stdout and stderr. Those
messages land in the sandbox's Ready condition and in platform events
that the gateway republishes to the sandbox event stream. With agent
output in that log, those messages would carry arbitrary agent output,
including anything sensitive the agent prints, into gateway status and
events. The supervisor starts its health endpoint only after the agent
starts, so every Docker failure path could include agent output, and the
VM driver reports one whenever the VM or host supervisor exits. Forward
only the supervisor's log tail, matching the Podman driver, which reads
the workload log solely to match fixed launcher markers and never
forwards raw workload output. The workload's output remains available
through docker logs and the VM's rootfs-console.log.
Document where main process output appears in the logging docs and the
cluster debugging skill.
Closes#3928
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* fix(supervisor): restore canonical stdin after connection loss
Probe idle SSH peers and enforce a receive deadline during transport I/O,
including writes blocked by a stalled relay. Release the dead attachment's
stdin lease through existing handler cleanup.
Retry denied write intent on later ordinary input without displacing a
healthy owner. Preserve explicit read-only, EOF and detach behavior, and
discard control bytes retained while input ownership was denied.
Cover half-open forwarding, blocked writes, healthy idle peers and competing
reconnects through the production supervisor frame bridge and real SSH.
Fixes#3648
Signed-off-by: Shiju <shiju@nvidia.com>
* docs(skills): describe read-only reconnect input retry
Explain what an openshell-cli user sees when automatic recovery reattaches before the supervisor closes the dead connection: the attachment reports read-only, later ordinary input retries stdin acquisition and prints `input enabled`, input typed while read-only is discarded, exit keys still detach, and an explicitly read-only viewer or a healthy owner is never affected.
Signed-off-by: Shiju <shiju@nvidia.com>
---------
Signed-off-by: Shiju <shiju@nvidia.com>
* feat(server): write gateway OCSF events to JSONL
Previously, gateway security activity was available only in diagnostic output,
and events not associated with a sandbox, such as TLS certificate reloads, had
no independent structured record.
Now, configuring `openshell.gateway.ocsf_log` writes every gateway-produced
OCSF record to a bounded JSONL destination independently of `RUST_LOG`. The
destination supports daily or disabled rotation, retention limits, queue bounds,
and optional schema downgrade to OCSF 1.1 or 1.3.
Additionally, existing gateway emitters (TLS reloads, service routing, and
policy approval and auto-approval audits) emit structured events, so they reach
the JSONL destination, console shorthand, and the affected sandbox's log
stream. Records identify the gateway by its configured name in `device.uid` and
`device.name`, shared across replicas, with `device.hostname` identifying the
replica and `device.os` the gateway's operating system. Metrics and warnings
expose known best-effort losses.
Refs #2762
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* fix(mxc): attribute ETW events to gateway
Signed-off-by: Evan Lezar <elezar@nvidia.com>
---------
Signed-off-by: Kris Hicks <khicks@nvidia.com>
Signed-off-by: Evan Lezar <elezar@nvidia.com>
Co-authored-by: Evan Lezar <elezar@nvidia.com>
* fix(snap): require mTLS for the snap gateway
Replace the installer opt-in with an authenticated snap gateway. The wrapper
no longer forces plaintext, so the gateway serves TLS from the bundle it
already generates in $SNAP_COMMON/tls. The install hook writes a config that
enables mTLS user auth instead of unauthenticated access, and a new
post-refresh hook migrates the exact legacy default on existing installs.
install.sh waits for the gateway, detects whether it serves TLS, copies the
client bundle into the target user's snap state directory, and registers the
gateway over HTTPS. Older plaintext snap revisions still register over HTTP
with a warning. The release canary asserts mTLS auth and HTTPS registration.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(snap): pass config preflight and detect the mTLS gateway reliably
An explicit [openshell.gateway.mtls_auth] table fails config preflight,
which validates mTLS auth before the local TLS bundle supplies the client
CA. Write a default that pins the Docker driver instead; with the wrapper's
TLS bundle the gateway requires client certificates and enables mTLS user
auth automatically, as the native packages do.
The mTLS gateway rejects TLS handshakes without a client certificate, and
it still answers plaintext loopback HTTP for sandbox service routing, so the
installer could misdetect it as a legacy plaintext gateway. Probe HTTPS with
the root-owned client bundle, and treat a gateway as legacy only when a
plaintext gRPC Health call succeeds.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* chore(snap): simplify install hook comment
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(snap): migrate insecure gateway configs on refresh
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* refactor(snap): simplify mTLS detection and config migration
Detect the mTLS snap from the installed revision's post-refresh hook instead
of probing plaintext gRPC, and drop the scheme global. Remove the installer's
pre-hook config fallback, which is dead now that every channel ships the
install hook and which wrote the insecure default. Give the install hook a
single write path with a simple backup name, and shorten the manual client
certificate steps.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(snap): stop keeping a copy of replaced insecure configs
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* feat(install): make the snap an opt-in install method
Stop selecting the OpenShell snap just because the snap command exists. Linux
installs default to the Debian or RPM package; OPENSHELL_INSTALL_METHOD=snap
(or deb, rpm) selects the package explicitly. Hosts that already have the
OpenShell snap keep refreshing it rather than gaining a second gateway on the
same port. The release canary and snap repro script opt in explicitly.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(snap): let the gateway auto-detect its compute driver
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(snap): restart the gateway after refresh
Published revisions use refresh-mode: endure, and snapd honors the old
revision's setting during a refresh, so the plaintext gateway kept running
with the migrated config unused until a manual restart. Restart the gateway
from the post-refresh hook so the mTLS config takes effect immediately.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* docs(snap): drop refresh notes from the snap description
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* docs(snap): trim snap refresh notes from installation docs
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
---------
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
The /_ws_tunnel endpoint pipes a WebSocket into the full gRPC service. On a
plaintext loopback gateway any web page could open it, since browsers do not
apply CORS to WebSocket upgrades. Mount the tunnel only when
enable_websocket_tunnel is set (config file, --enable-websocket-tunnel, or
OPENSHELL_ENABLE_WEBSOCKET_TUNNEL; server.enableWebsocketTunnel in Helm).
BREAKING CHANGE: gateways behind an authenticating edge proxy must set
enable_websocket_tunnel = true for CLI edge-tunnel connections.
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(examples): add quickstart rule with policy update and show OCSF logs
The quickstart applied policy.yaml with `openshell policy set`, which replaces
the whole policy. The file omitted /bin from the restrictive default, so the
live filesystem additivity check could reject it. Add the rule with
`openshell policy update` instead, and keep policy.yaml as a complete policy
for `sandbox create --policy` that covers the default read-only paths.
The demo and README filtered logs with `--level warn`, but the server ranks
OCSF events as INFO, which hid the policy decisions the demo shows. Query
`--source sandbox` without a level filter and match the OCSF shorthand
(DENIED/ALLOWED) instead of the retired key=value format.
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
* docs(skills): align generate-sandbox-policy with proxy behavior
Remove the `protocol: sql` validation check and the SQL `command` matcher,
which the published policy docs no longer describe.
Correct the private IP guidance: exact user-declared hostnames may reach
private addresses without allowed_ips. Wildcard, hostless, and
advisor-proposed endpoints still need allowed_ips, and loopback, link-local,
unspecified, and cloud metadata addresses stay blocked.
Stop describing an omitted protocol as pure L4 or uninspected. The proxy
still terminates TLS, parses HTTP strictly, and enforces request authority;
it only skips method and path rules.
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
* fix(providers): drop pip script paths from pypi profile binaries
OpenShell identifies a process by /proc/<pid>/exe, so a pip script runs as
its Python interpreter and the .venv/bin/pip entries could never match. The
venv interpreters that run those scripts are already listed, so remove the
script paths and explain in the header that users must list interpreters.
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
* docs(skills): fix openshell-cli policy iteration steps
Monitoring denials with `--level warn` hides them, because the server ranks
OCSF policy events as INFO. Drop the level filter and describe the OCSF
shorthand DENIED lines instead of the retired `action: deny` format.
`policy get --full > file` produced input that `policy set` cannot parse: the
output starts with revision details before the `---` separator, and --full
adds provider-composed rules. Export `--base` and keep only the YAML after the
separator. Also stop recommending full replacement for filesystem, Landlock,
or process changes, which require recreating the sandbox, and drop the SQL
mention that generate-sandbox-policy no longer covers.
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
* docs(skills): qualify authority checks for omitted-protocol endpoints
An endpoint without `protocol` only receives authority checks on HTTP requests
the proxy parses after default TLS handling. Without an L7 route or required
middleware, other CONNECT payloads such as HTTP/2 prior knowledge can use the
raw relay, and `tls: skip` bypasses termination and parsing. Stop describing
omitted-protocol endpoints as always authority-checked.
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
---------
Signed-off-by: Johnny Greco <jogreco@nvidia.com>
* fix(podman): restore host gateway alias mediation
Signed-off-by: Gordon Sim <gsim@redhat.com>
* fix(podman-e2e-tests): enable broader test podman e2e coverage
Signed-off-by: Gordon Sim <gsim@redhat.com>
* fix(tests): make test more reliable
Signed-off-by: Gordon Sim <gsim@redhat.com>
* fix(podman): fix macos linting error
Signed-off-by: Gordon Sim <gsim@redhat.com>
---------
Signed-off-by: Gordon Sim <gsim@redhat.com>
* fix(driver-kubernetes-secrets)!: store provider credentials in one namespace
The Kubernetes Secrets credential driver now stores every credential in its
configured namespace in all workspace modes and rejects handles that reference
any other namespace before contacting the Kubernetes API. The gateway reaches
credential Secrets through the Role in that namespace; this allows removing the
Secret rules from the ClusterRole.
- Remove the workspace_mode, gateway_id, and allow_reference_namespace driver
settings and stop rendering them from Helm. Configurations that set them fail
at startup. Existing credential state is not migrated.
- Add server.credentialDrivers.kubernetesSecrets.createNamespace to provision a
dedicated credential namespace. The namespace is kept on uninstall, adopted
by a reinstall of the same release, and left untouched when something else owns
it.
- Update the gateway config reference, Kubernetes setup docs, 0.1.0
upgrade guide, compute-runtime architecture, and cluster debugging skill.
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* fix(helm): reduce gateway Secret permissions
Remove the gateway's Secret list permission in every workspace mode and
grant source Secret reads through a Role in the sandbox namespace.
Bootstrap Secret cleanup deletes Secrets by exact name instead of listing
them.
- Grant get on the copied client TLS and image-pull Secrets through a Role in
the sandbox namespace. The ClusterRole keeps get and patch on those names
for the ownership check and server-side apply into workspace namespaces.
- Delete sandbox and supervisor bootstrap Secrets by exact name, derived from
the runtime generation recorded on the Sandbox and, on restart, the target
generation, tolerating 404. The generation annotation is cleared only after
cleanup succeeds, and each bootstrap Secret has a Pod owner reference, so
garbage collection removes any generation the driver does not name.
- Drop Secret list from the ClusterRole and the shared-mode sandbox Role.
- Extend the managed e2e RBAC checks to Secret list.
- Update the Kubernetes setup and sandbox runtime docs, compute-runtime
architecture, and cluster debugging skill.
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* fix(driver-kubernetes): stage workspace Secrets per runtime generation
Every Secret the Kubernetes driver writes into a workspace namespace is
now scoped to one sandbox runtime generation, immutable, and created
with create only, so the gateway never reads, patches, or adopts an
existing Secret there. This removes the gateway's cluster-wide get and
patch on the copied client TLS and image-pull Secret names.
- In managed mode, create an immutable copy of each configured
image-pull Secret per generation, named os-pull-<id>-<generation>-<n>
and owned by the generation's workload and supervisor Pods. Pods and
the restarted Sandbox template reference those names, and generation
cleanup deletes them by name. A Secret already holding a generation
name fails the create.
- Outside shared mode, stage the gateway client TLS material into the
supervisor bootstrap Secret instead of copying the client TLS Secret
into the workspace namespace.
- Remove the fixed-name TLS and image-pull copies, the target ownership
read, and the ClusterRole get and patch rule on the copied names.
Source reads stay in the sandbox-namespace Role.
- Update the managed e2e to expect generation image-pull Secrets and
client TLS material in the supervisor bootstrap Secret, and to check
that the gateway cannot read the copied names in workspace namespaces.
- Update the gateway config and compute driver references, Kubernetes
setup and sandbox runtime docs, compute-runtime architecture, driver
README, and cluster debugging skill.
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* fix(helm)!: grant operator-mode Secret permissions through the workspace chart
The operator-mode gateway ClusterRole grants no Secret permissions.
The openshell-workspace chart Role, installed in each operator-managed
namespace, grants the gateway create and delete on Secrets for sandbox
runtime generations. Operator-managed namespaces require the workspace
chart.
- Fail the chart tests on any ClusterRole rule that includes Secrets in
operator and shared modes.
- Install the workspace chart when the operator e2e provisions a
namespace, and assert that the gateway has no Secret permissions in a
namespace without it.
- Update the Kubernetes setup docs, 0.1.0 upgrade guide, compute-runtime
architecture, and cluster debugging skill.
Signed-off-by: Kris Hicks <khicks@nvidia.com>
---------
Signed-off-by: Kris Hicks <khicks@nvidia.com>
* feat(kubernetes): support corporate proxy CA bundle
The Kubernetes driver had no way to supply a CA bundle for the corporate
egress proxy, so an `https://` proxy with a private CA, or a TLS-intercepting
proxy, could not be used. Podman and VM already expose `proxy_ca_bundle`.
Add `proxy_ca_bundle` to `[openshell.drivers.kubernetes]` as a path the
gateway Pod reads. The gateway stages the PEM into the existing
per-generation supervisor bootstrap Secret and passes
`--upstream-proxy-ca-bundle` on the supervisor argv. That Secret is already
immutable, owner-referenced and garbage-collected, and its volume mounts
every key at /.openshell/supervisor with no items filter, so this needs no
new object kind, volume, mount, or RBAC verb, and works in shared, managed
and operator workspace modes.
The bundle is deliberately read from the gateway's filesystem rather than
referenced as an object in the sandbox namespace. It becomes a trust anchor
for every upstream the sandbox reaches, so it must stay in the gateway's
trust domain; the immutable staging Secret also keeps the anchor from
changing underneath a running sandbox.
Bound the staged bundle at 256 KiB. The shared reader's limit is exactly the
apiserver's own Secret limit and the bootstrap Secret carries four other
keys, so a bundle between the two would pass gateway startup and then fail
every sandbox create with an opaque `data: Too long`.
Delegate the URL, no_proxy, connect_by_hostname and ca_bundle rules to the
shared validate_upstream_proxy_settings, keeping the Secret-specific
credential block local: this driver accepts an explicit
`proxy_auth_allow_insecure = false` without credentials, which the shared
rules reject. This also fixes the acknowledgement being demanded for an
`https://` proxy, where the credential travels inside the verified TLS
session. Add auth_setting_label so the inline-credential diagnostic names
the Secret keys instead of proxy_auth_file, which this driver rejects as an
unknown key.
Document that the bundle should carry only the CA that signs the proxy's
certificate, or that an intercepting proxy re-signs upstream certificates
with. Public roots already reach the sandbox through the supervisor image and
its TLS stack, and the bundle is concatenated with that system store into a
single boundary control frame, so a full merged trust bundle spends the frame
budget on duplicated roots. The frame, not the apiserver Secret limit, is the
tighter of the two ceilings in practice; raising the staging bound requires
checking it.
Closes#3443
Signed-off-by: Philippe Martin <phmartin@redhat.com>
* fix(helm): quote proxy CA ConfigMap references
Signed-off-by: Philippe Martin <phmartin@redhat.com>
---------
Signed-off-by: Philippe Martin <phmartin@redhat.com>
* feat(sandbox): default to official Alpine sandbox image
default_sandbox_image() now returns docker.io/library/alpine:3.22, a generic
version-qualified official image, so a fresh install no longer depends on the
community sandbox image catalog. All compute drivers (docker, podman,
kubernetes, vm) inherit this fallback.
Part of #3116.
Signed-off-by: Akram
Signed-off-by: Akram <akram.benaissi@gmail.com>
* feat(deploy): default deployment configs to the official Alpine sandbox image
Update the shared gateway default_image, Helm chart values, the standalone
Kubernetes manifest, and the dev gateway task scripts to use
docker.io/library/alpine:3.22 instead of the community base image, consistent
with default_sandbox_image(). GPU e2e image-build base is left unchanged (CUDA
needs a glibc base).
Part of #3116.
Signed-off-by: Akram
Signed-off-by: Akram <akram.benaissi@gmail.com>
* feat(driver): default to numeric non-root identity for USER-less images
With the default sandbox image now Alpine, images that declare no OCI USER
must start instead of being rejected. When the image declares no USER and
the policy requests none, the Podman and Docker drivers now supply a numeric
non-root identity (DEFAULT_SANDBOX_UID/GID = 1000) instead of rejecting,
matching the numeric-identity behavior of the Kubernetes and VM drivers. The
supervisor's resolved-identity path runs the sandbox as a synthesized
non-root account without the account existing in the image. Images that
declare a USER keep the OCI resolution path unchanged.
Part of #3116.
Signed-off-by: Akram <akram.benaissi@gmail.com>
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* test(conformance): use Alpine workload image
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* refactor(policy): drop community image /app path from default policy
The restrictive default policy granted read-only access to /app, a directory
that only existed in the community base image. A generic Alpine default has no
/app, so remove it. Landlock best-effort already ignores absent paths; this
just stops advertising a community-specific layout in the default.
Part of #3116.
Signed-off-by: Akram
Signed-off-by: Akram <akram.benaissi@gmail.com>
* docs(config): document Alpine default images
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* fix(podman): report early sandbox termination
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* fix(podman): initialize rootless workspace ownership
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* fix(sandbox): qualify NVIDIA Ubuntu default
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(podman): initialize rootful default workspace
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* feat(sftp): add native sandbox adapter
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(sftp): gate runtime helper support to Linux
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(sftp): support standard OpenSSH file operations
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(sftp): harden rename and special file handling
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* refactor(runtime): remove community image dependencies
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* test(e2e): build provider readiness tool fixture
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* fix(e2e): use a dedicated Noble fixture for Docker tests
Signed-off-by: Evan Lezar <elezar@nvidia.com>
---------
Signed-off-by: Akram
Signed-off-by: Akram <akram.benaissi@gmail.com>
Signed-off-by: Evan Lezar <elezar@nvidia.com>
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
Co-authored-by: Evan Lezar <elezar@nvidia.com>
Co-authored-by: Drew Newberry <anewberry@nvidia.com>
The opt-in glibc-static supervisor variant from #2682 is gone: the Nix
release builds (#2977) removed it from CI and the supervisor Dockerfile,
and the RFC 0012 sandbox split (#2942) removed it from the staging script.
The split also moved the binary that runs inside workload images to
openshell-sandbox; the supervisor now runs from its own image and is
dynamically linked. A few places still describe the old model:
- skills/debug-openshell-cluster told operators to check that
supervisor_image contains a static /openshell-supervisor. Ask instead for
a supervisor from the matching release whose loader and shared libraries
are available inside that image, and give a `--version` check that shows
both. The static requirement for /openshell-sandbox is unchanged.
- verify-static-binary.sh justified its check with the supervisor and the
glibc-static variant. Describe the property it enforces instead, with
the musl sandbox runtime as the main example.
- stage-prebuilt-binaries.sh kept an unreachable gnu-static case in
target_triple.
No behavior change: resolve_component only ever selects gnu or musl.
Signed-off-by: Emilien Macchi <emacchi@redhat.com>
* feat(kubernetes): support HA gateway rebalancing
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
* perf(server): cache peer connections, tokens, and owner lookups
Every forwarded relay rebuilt its setup from scratch: an owner lookup, a
blocking read of the peer token, a TLS connect to the owning replica, and
a TokenReview plus Pod GET on the receiving side. Sandbox service routing
does this per HTTP request, so the apiserver calls scaled with traffic.
Cache all of it on ServerState:
- peer channels pooled per endpoint, so relays multiplex over one
connection instead of redialing
- peer tokens keyed by SHA-256, expiring at min(ttl, token exp) so a hit
cannot accept an expired token
- owner records for 3s against a 45s ownership TTL, still freshness
checked before use
Entries are evicted when a relay fails. Also raise HTTP/2
max_concurrent_streams to 1024, since pooling funnels every relay between
two replicas onto one connection and hyper's default of 200 sits below
the 256 pending-relay budget.
Signed-off-by: divesh <dgude@nvidia.com>
* perf(server): pool upstream connections for sandbox services
Each HTTP request to a sandbox service opened its own supervisor relay,
paying a new TCP connection and HTTP/1 handshake every time. Worse, it
counted against the 32 in-flight relay cap, so a service handling more
than 32 concurrent requests failed outright.
Pool idle upstreams per endpoint and port, up to 8 each for 15s. Reuse is
safe because the pool only returns a connection hyper reports as ready,
and HTTP/1 cannot start a request until the previous body has drained.
Upgrades are never pooled since they take the connection over, and a
failed send evicts that endpoint. Pruning is bounded per key, with the
full sweep limited to once per 30s.
Signed-off-by: divesh <dgude@nvidia.com>
* fix(server): address HA gateway review findings (#3449)
- Let a gateway own supervisor sessions without a peer endpoint. Requiring
one whenever the store is PostgreSQL broke every single-instance
PostgreSQL deployment, because no sandbox supervisor could connect.
A cross-replica request to an owner that advertises no endpoint now fails
immediately naming the cause, instead of retrying until the wait timeout.
- Close a supervisor session on heartbeat only when another replica owns it,
or after renewals fail for the ownership TTL. A database error no longer
drops every session heartbeating during an outage.
- Clamp owner record ages at zero so a skewed or corrupt stored timestamp
cannot produce a negative age.
- Bound the cross-object advisory lock with a lock timeout, so a stuck holder
fails instead of blocking every mutation in the fleet.
- Refuse to start when a peer endpoint is configured on a multi-replica
backend but peer authentication is unavailable, and warn when a
multi-replica backend has no peer endpoint at all.
- Reject a plaintext peer endpoint when the gateway serves TLS.
- Skip the sandbox watch poller on single-replica backends, where the local
update bus already sees every write.
- Rate-limit the peer owner cache sweep so an insert no longer scans the
whole map under the lock.
- Retry GET and HEAD on a pooled upstream the sandbox closed, instead of
returning 502, and drop an emptied endpoint from the pool right away.
- Document the gateway peer environment variables and the post-rollout
ownership skew operators should expect.
Signed-off-by: divesh <dgude@nvidia.com>
* fix(server): harden HA supervisor ownership
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
---------
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
Signed-off-by: divesh <dgude@nvidia.com>
Co-authored-by: Drew Newberry <anewberry@nvidia.com>
Co-authored-by: divesh <dgude@nvidia.com>
Co-authored-by: Divesh Chowdary <47188680+FrostGod@users.noreply.github.com>
* feat(cli): promote profile commands to top level
Add profile discovery and management commands with shared handlers for
the existing provider entry points. List a flat catalog across scopes
and follow continuation tokens through full and short pages.
Describe metadata, credentials, endpoints, TLS inspection, and MCP access
settings while preserving complete JSON/YAML definitions. Cover parser
equivalence, scope forwarding, pagination, and inspection settings with
focused unit and compiled-CLI integration tests.
Update docs, public skills, examples, and E2E command invocations.
Refs #2588
Signed-off-by: Shiju <shiju@nvidia.com>
* fix(cli): remove redundant workspace selector qualification
Use the imported WorkspaceSelector in the provider integration helper so
the target passes Clippy with warnings denied.
Signed-off-by: Shiju <shiju@nvidia.com>
---------
Signed-off-by: Shiju <shiju@nvidia.com>
John T. Myers