The 600 s clamp was unreachable in practice: the controller's stop
request times out at 240 s (controller/node.py) and the Docker stop
query gets the value +30 s as its HTTP timeout, so anything above 210
would abort upstream first and surface an error while the stop keeps
running server-side. Cap at the derived ceiling and document the chain
in the clamp and the docstring.
Nine fixes from a review of the docker-shm-devices diff:
* GNS3_STOP_TIMEOUT >300 s aborted at the manager's default HTTP timeout
before Docker finished the stop — the stop query now gets a timeout
with a margin over the grace period.
* Overlapping bind targets (GNS3_MASK_UDEV + GNS3_MASK_SYSTEMD on the
same unit, a unit named twice, an extra_configs target equal to a
masked unit) made Docker reject the create with 'Duplicate mount
point' — Mounts are deduplicated by target.
* ExtraConfig.target now carries a pydantic validator (absolute file
path, no '..'), so bad targets 422 at template-save time instead of
failing at node-create time after a multi-GB image pull; directory
forms ('/', '/etc/') are also rejected by the runtime guard instead
of raising IsADirectoryError (raw 500).
* _check_host_readiness skipped every remaining check when one
/proc/sys key was unreadable (mid-loop return) — now continues.
* The base-class GNS3_* env parser strips trailing commas like the
vendor parser, so 'GNS3_MASK_UDEV=1,' composed from a list still
activates.
* Vendor env knobs are re-parsed on every create(), so a PUT to the
node's environment takes effect on the next (re)create.
* The graceful SIGTERM stop is now limited to the explicit user stop
route; delete/update/close/crash-cleanup keep the immediate kill
(those paths force-delete or recreate the container right after).
* An extra_configs target beneath a persisted volume is shadowed by the
volume bind — warn at create time.
The 60 s SIGTERM grace was hardcoded, unlike every other vendor knob
(GNS3_SHM_SIZE, GNS3_DEVICES, GNS3_MASK_UDEV, ...) which rides the
environment line. Parse GNS3_STOP_TIMEOUT=<seconds> (default 60,
clamped to 1-600, invalid values keep the default) and use it in
VendorDockerVM._terminate_container().
DockerVM.stop() terminated containers with an immediate SIGKILL — fine
for init.sh-based containers whose state is persisted beforehand, but a
systemd NOS (Cisco XRd, SR Linux) needs a graceful shutdown and treats
the abrupt kill as an unclean shutdown (exit 137 on every stop).
Extract the final termination into _terminate_container() and override
it in VendorDockerVM: POST /containers/{id}/stop?t=60 sends SIGTERM and
waits for systemd to stop services; Docker itself SIGKILLs the
container once the grace period expires, so no fallback is needed.
Docker's 304 (already stopped) is swallowed.
PortManager.get_free_udp_port had an unguarded find-then-add sequence.
A link allocates both ends concurrently (asyncio.gather in
UDPLink._prepare -> two POST /ports/udp) and FastAPI runs the sync
route handler in a threadpool, so both threads could probe the same
'free' port before either recorded it — handing lport == rport to both
ends. uBridge sets SO_REUSEADDR on UDP NIO sockets, so the double bind
succeeds silently and the kernel delivers everything to the last-bound
socket: one node starves, the other echoes to itself.
Make every TCP/UDP allocate/reserve/release path atomic with an RLock,
and rebuild _link_data in UDPLink._prepare so reset() commits the fresh
port pair instead of re-sending the stale, already-released one.
Regression tests: threaded barrier allocation never returns duplicates
(red on the old code, UDP and TCP); reset() leaves exactly one mirrored
NIO pair per side with lport != rport (red on the old code).
Add bugs/link-udp-self-loop.md: the intermittent one-way docker link
observed during XRd validation (one end's nio_udp rport pointing at its
own lport after a uBridge restart), with the evidence table, the uBridge
console capture diagnostics, the delete/recreate workaround, and the
narrowed suspects (batch port preallocation / link re-creation race).
Also note that a docker node's in-container ethN is a TAP device held by
uBridge (no veth host end exists).
Update the XRd feature doc: datapath validated end-to-end (XRd brings
its own interfaces up, ARP/ICMP bidirectional), add a troubleshooting
row pointing at the bug doc. Index the bug in docs/README.md.
New feature doc covering why XRd takes the vendor docker_exec/SKIP_INIT
path (init.sh wrapper crashes its glibc loader), the four generic
mechanisms added for it (GNS3_SHM_SIZE/GNS3_DEVICES HostConfig injection,
extra_configs file injection, GNS3_MASK_UDEV + udevadm null-binding,
host-readiness check), the three host-disturbance root causes isolated by
plain docker-run A/B/C testing, the appliance recipe with XRd-specific
gotchas (Mg0/RP0/CPU0/0, /xr-storage-shadow persistence, first-boot
semantics), and troubleshooting. Indexed in docs/README.md alongside the
docker-exec-console base doc.
Document the SR Linux gns3a (35-adapter full chassis, matching
GNS3_INTERFACE_NAMES + custom_adapters), the three server-side schema fixes
needed for it to load (DockerConsoleType.docker_exec,
ApplianceV1_6.custom_adapters, extra_volumes docker-block passthrough), and
the symbol-theme behaviour that rewrites any :/symbols/-prefixed symbol to
the category default at load time (so router_cloud.svg cannot be used from
an appliance; use a custom symbol under symbols_path instead).
The reconnect-blank-screen bug: when sr_cli exited (quit / idle timeout /
crash) the while-true wrapper restarted it mid-session with no client
attached, so its startup CPR probe (\e[6n) went unanswered and the TUI
degraded/blocked. On reconnect lazy_started=True skipped recreation, so the
client saw a blank screen.
Fix: drop the while-true wrapper. Now when the CLI exits, the exec pty
closes (EOF), the broadcast task ends, and the next client connection
detects the dead upstream via _upstream_alive() and recreates the exec —
with a terminal attached, so CPR is answered. A live exec is reused
(just a Ctrl-L redraw).
_LazyExecTelnetServer is extracted from a closure to module level so the
reconnect/recreate logic is unit-testable. Add 9 tests covering
_upstream_alive states and the recreate-on-death / reuse-if-live /
close-half-dead-writer / no-while-true behaviors.
Full Docker suite (120) passes.
Override _mount_binds in VendorDockerVM: for GNS3_SKIP_INIT containers the
/etc/network volume (GNS3's own network config consumed by init.sh's ifup)
is dead weight — init.sh never runs and the NOS manages its own
interfaces. The override removes the bind, filters /etc/network out of
self._volumes (keeping GNS3_VOLUMES and the bridge/fix passes consistent)
and deletes the host-side skeleton directory created by the base class.
Without GNS3_SKIP_INIT the mount is kept, matching base behaviour.
_persistent_volumes() is removed — the mount override is now the single
filter point.
GNS3_SKIP_INIT containers never run init.sh, so /etc/network (GNS3's own
network config consumed by init.sh's ifup) has no consumer — the NOS
manages its own interfaces. VendorDockerVM._persistent_volumes() filters it
out for both _setup_skip_init_volumes and _fix_permissions, saving one
docker exec per pass. The shared _mount_binds is untouched, and without
GNS3_SKIP_INIT the full volume list is returned so behaviour matches the
base class.
Document why host-user ownership of volume files during runtime is
harmless for SR Linux (root processes, self-healing daemons like aaamgr
rewriting its managed files, ACL-based access) and the deviation from the
standard init.sh model, with the escape hatch of dropping the start-time
fix pass for strict-ownership NOS images.
Also document the boot-ordering caveat: the volume bridge comes up after
the NOS boots, so early boot reads overlay defaults — verify the
save/stop/start closed loop, and add troubleshooting entry #10 for config
present on the host but not applied after restart.
Document the aaamgr_local_user.json case: SR Linux's aaamgr daemon rewrites
the file during boot as the image's srlinux user (uid 1002) after the
start-time permission pass, leaving it unreadable until the stop-time pass.
Trace the warning to the file-browser API chain (Show in file manager ->
list_node_files -> magic.from_file) and note the impact is limited to the
file_type field.
The host-side pass could not work for unprivileged GNS3 processes: the
.gns3_perms marker is created root-owned by the container-side touch, and
chowning root-owned files from the host requires root.
Rewrite VendorDockerVM._fix_permissions to run the busybox
record/chmod/chown script inside the container (as root) on the
/gns3volumes bind-mount targets — they exist for the container's whole
lifetime and do not depend on the mount --bind bridge, so a container
restart can no longer make the fix hit the overlay copy. A
stopped/exited container is skipped (logged) instead of restarted; the
next start's pass fixes ownership.
Replace the container-side _fix_permissions for vendor NOS containers with a
host-side pass that walks the node's project directories directly (they are
the Docker bind-mount sources): records mode:uid:gid into .gns3_perms and
chowns to the GNS3 user. No docker exec, no container restart — the base
implementation restarts an exited container just to chown, and after the
restart the mount --bind bridge is gone so it would fix the overlay copy
instead of the host files.
The pass runs at start (after _setup_skip_init_volumes seeds and bridges the
volumes) so the controller can read project files while the node runs, and
again at stop for files written during runtime.
Update docker-exec-console.md: VendorDockerVM architecture, hook points,
class-selection factory, volume-persistence lifecycle, and new
troubleshooting entries.
128 characters was far beyond any realistic marker label (icmp, arp, tcp-syn)
and would have collided with the pcap filename budget once a tag prefix is
added later. Cap the user-facing name at 32 in both MarkerCreate and
MarkerDefinitionCreate; the compute-side name guard now also rejects names
longer than 48, which covers the `global-{def_name}` inherited form (≤ 39).
A marker definition validated its BPF N times — once per link in the fan-out
(start_marker runs validate_bpf_syntax on every copy), spawning one tcpdump -d
subprocess per link for the same expression. Definitions did not validate BPF
at all; only direction was checked.
Move the single validation point to the definition layer (create/update), and
validate each definition's BPF on project load (dropping any that have gone
invalid, like private markers). The inherited fan-out (start_marker) and def
sync (update_marker) now skip validate_bpf_syntax for inherited copies, since
the BPF comes from an already-validated definition. Private per-link markers
still validate inline as before. uBridge still runs pcap_compile at install, so
an invalid expression can never slip through.
Creating a definition over N links now runs one tcpdump instead of N.
The 409 on a tx/rx definition now recommends encoding direction in the BPF
(e.g. icmp[icmptype]==8) as the primary fix, with per-link markers as the
single-link fallback. Doc updated: per-def rejects tx/rx (why + BPF), the
pause section no longer claims bpf changes reset+reapply (they rebuild one
filter), and a new Capture files section covers pcap cleanup + reset-preserves-mark.
The project-wide mute (POST /markers/pause|resume + _markers_paused) paused
every marker with one button. The actual need is per-rule control: pause one
definition and toggle only its inherited global-{name} copies across all links.
- Drop project-level: _markers_paused (init/asdict/load/start_all), the
pause_all/resume_all_markers methods, and the /markers/pause|resume routes.
- Add per-definition: a persisted `paused` flag on each definition;
pause/resume_marker_definition fan out update_marker(enabled) to every
global-{name} copy — instant, via the existing enable_packet_filter toggle
(no NIO rebuild, pcap/emitted preserved). New links inherit a paused
definition already off (inherit_marker passes enabled=not paused).
- start_marker takes an enabled kwarg; update_marker's enabled-only short-circuit
now also covers inherited copies so def pause/resume is instant.
- Routes: POST /marker-definitions/{name}/pause|resume.
- Docs + tests updated.
pause/resume was fire-and-forget: the controller sent marker pause/resume but
stored nothing, so the Web UI could only keep a local optimistic flag that was
lost on panel reopen. Treat the mute as a project-level config (like per-marker
enabled): record _markers_paused, persist it in the topology (asdict + load),
and echo it on the project object so the UI renders from server truth.
Because marker pause is a uBridge runtime flag that resets on node restart,
start_all re-applies the mute to freshly started uBridges after a project
reopen — a paused project stays paused across close/reopen.
A marker is single-sided — only the chosen capture node's uBridge installs the
mark filter — and dir=tx|rx is interpreted from that node's perspective. Until
now the observer was always auto-picked (_choose_marker_side), so dir=tx meant
"the auto-chosen endpoint is sending", which is unpredictable and makes the
direction filter hard to render meaningfully in the Web UI.
Add an optional create-only capture_node_id to MarkerCreate: when set, the
marker is pinned to that endpoint's uBridge (validated as a link endpoint and a
marker-capable type); when omitted, behavior is unchanged (auto-pick). The
chosen id is already echoed back as capture_node_id and in MARK signals, so the
UI can always render the observer regardless of who picked it.
capture_node_id is create-only (changing it would silently flip the meaning of
stored direction; recreate instead) and is not accepted on project-level
definitions — they are link-agnostic and have no endpoints, so inherited
markers keep auto-picking per link.
Plumbed through REST create_marker, the MCP link_marker tool, and base
Link.start_marker. update_marker does not forward it.
Direction field in marker.match events
=======================================
Read ubridge dir=<tx|rx> from MARK signal datagrams and forward it
as a "dir" key in the marker.match notification event. The field is
additive -- older ubridge builds omit it and the parser leaves it null,
so consumers fall back to undirected rendering with no version coupling.
Semantics are relative to the capture node (the signal's node=<id>):
tx = capture node is sending (ingressed device-side NIO), rx = it is
receiving (ingressed link-side NIO).
Per-marker direction filter (opt-in, server-side pipeline)
==========================================================
Add a direction field to MarkerCreate and MarkerDefinitionCreate
schemas ("tx" | "rx" | null). Plumb it through the full pipeline:
Schema -> controller (start_marker/update_marker, marker_entry,
_markers_for_node, update_marker_definition sync)
-> REST/MCP handlers -> compute _ubridge_add_marker_filter +
IOU _ubridge_apply_markers -> bridge add_packet_filter dir <tx|rx>
When set, ubridge only fires the mark handler (signal + pcap) for
packets matching the chosen direction. null (default/legacy) = both
directions -- zero behavioural change for existing markers.
Docs and tests
==============
- docs/features/marker-traffic-insight.md: signal format updated,
new Direction section with NIO mapping, arrow mapping, and additive
compatibility note.
- tests/compute/marker/test_marker_manager.py: 3 new parser tests
(dir tx/rx/absent) plus existing test extended to assert dir=None.
13 files, +123/-22, 72 tests pass (zero breakage)
For Docker nodes, reload bottoms out as a raw POST /containers/{id}/restart
to the Docker daemon, bypassing GNS3's start/stop lifecycle (uBridge
re-attach, console servers, NIC setup). The container restarts at the
Docker level but GNS3's plumbing goes out of sync, and the daemon call can
block up to the controller's 240s timeout — manifesting as "reload hangs /
no response". stop+start runs the full lifecycle and is reliable.
Remove the node_reload and node_reload_all MCP tools (tool functions,
handlers, NODE_TOOLS entry, tests, docs). The underlying REST endpoints
(POST /nodes/{id}/reload, POST /nodes/reload) are kept for native API
users. MCP callers should use node_stop + node_start (partial) or
close/open project (full restart) instead.
Add a Per-link attribution section (link field semantics, pcap path keyed on link) with an IOU subsection explaining the shared IOL-BRIDGE and per-port filter lists. Note that the marker.match link_id comes from the signal link= field, and call out shared capture-side node behavior in Notes.
Document the marker feature following the gns3-documentation standard:
overview, Mermaid architecture and business-process diagrams, API endpoint
tables, request/response and field reference, error responses, and notes
covering the key design points — immutable name (rename = delete + create),
global-prefix reservation, read-only inherited markers, render hints
(color / highlight_duration), supported node types, and persistence.
Indexed under Features in docs/README.md.
- Move fastmcp from core requirements.txt to mcp-requirements.txt
- Add MCP_AVAILABLE feature flag in agent/__init__.py (graceful degradation)
- Guard MCP imports/registration in server.py and tasks.py
- Replace ai-copilot/mcp/ai-support extras with single ai-features extra
- Add stub MCP routes returning 501 when MCP is not installed
- Add gns3server-uninstall-ai-features CLI command
- Remove old gns3server-uninstall-ai-copilot command
- Update all error messages and docs to reference ai-features
Closes#2794
The name device_command_run was ambiguous and easily confused with
device_config_send. device_show_run makes the read-only intent explicit,
forming a clear pair: config_send (write) vs show_run (read).
Add documentation explaining how _server_url() resolves the host
when Server.host is 0.0.0.0 or :: — using the default route
interface IP instead of hardcoding 127.0.0.1.
Added Transport Security section to MCP service documentation covering:
- Default behaviour (disabled, allow all hosts)
- How to enable protection via gns3_server.conf
- Protection mechanism (Host header validation)
- DNS rebinding attack prevention explanation
- Behaviour summary table
