Files
fips/testing/chaos/sim/control.py
T
Johnathan Corgan 77ac8c822e Add fipstop TUI monitoring tool with smoothed metrics and quality indices
fipstop: ratatui-based TUI for real-time monitoring of a running FIPS daemon.

Tabs and navigation:
- 8 navigable tabs: Node, Peers, Transports, Sessions, Tree, Filters,
  Performance, Routing
- Tab/BackTab navigation with group separators in tab bar
- Table views with selectable rows, detail drill-down panels, and scrollbars

Node tab:
- Runtime info: pid, exe path, uptime, control socket path, TUN adapter name
- Identity: npub, node_addr, ipv6 address
- State summary with peer/session/link/transport/connection counts
- TUN IPv6 traffic and forwarded transit traffic counters

Peers tab:
- Table with Name, Address, Conn, Depth, SRTT, Loss, LQI, Pkts Tx/Rx
- Detail panel: identity, connection info, transport cross-reference,
  tree/bloom state, link stats, MMP metrics with LQI

Sessions tab:
- Table with Name, Remote Addr, State, Role, SRTT, Loss, SQI, Path MTU,
  Last Activity
- Detail panel: identity, session info, traffic stats, MMP metrics with SQI

Transports tab:
- Hierarchical tree view: expandable transport parents with nested links
  (▼/▶ indicators, ├─/└─ tree chars, Space/Arrow to expand/collapse)
- Transport detail: type-specific stats (UDP/TCP/Ethernet)
- Link detail: peer cross-reference with MMP metrics and LQI

Performance tab:
- Link-layer MMP: SRTT, loss, ETX, LQI, goodput per peer
- Session-layer MMP: SRTT, loss, ETX, SQI, path MTU per session
- Trend indicators (rising/falling/stable) with context-aware coloring

Routing tab:
- Routing state: cache sizes, pending lookups, recent requests
- Coordinate cache: entries, fill ratio, TTL, expiry, avg age
- Statistics: forwarding, discovery request/response, error signal counters

Tree tab:
- Spanning tree position with 16 announce stats (inbound/outbound/cumulative)

Filters tab:
- Bloom filter announce stats, per-peer fill ratio and estimated node count

MMP metrics enhancements:
- Add etx_trend DualEwma for smoothed ETX tracking
- Add smoothed_loss() and smoothed_etx() accessors (long-term EWMA)
- LQI (Link Quality Index) = smoothed_etx * (1 + srtt_ms / 100)
- SQI (Session Quality Index) = same formula for session layer
- All loss/ETX displays prefer smoothed values with raw fallback

Control socket:
- Add smoothed_loss, smoothed_etx, lqi/sqi to show_peers, show_sessions,
  and show_mmp JSON responses
- Rename fips_address to ipv6_addr in show_status and show_peers
- Add tun_name and control_socket to show_status
- FHS-compliant 3-tier default path: $XDG_RUNTIME_DIR, /run/fips, /tmp

Node extensions:
- Add started_at/uptime() to Node
- Add tun_name() accessor

Docker sidecar updates:
- TCP transport support via FIPS_PEER_TRANSPORT env var
- Build scripts include fipstop binary
2026-03-01 16:33:33 +00:00

104 lines
3.3 KiB
Python

"""Control socket querying via docker exec.
Queries FIPS nodes' control sockets to observe runtime state (tree
structure, MMP metrics, peers) without requiring fipsctl in the
container. Uses a Python one-liner inside docker exec since python3
is available in the Docker image.
"""
from __future__ import annotations
import json
import logging
from .docker_exec import docker_exec, docker_exec_quiet
from .topology import SimTopology
log = logging.getLogger(__name__)
# Default control socket path inside containers (running as root,
# so /run/fips/ is created by the daemon).
CONTROL_SOCKET = "/run/fips/control.sock"
def query_node(container: str, command: str, timeout: int = 10) -> dict | None:
"""Send a command to a node's control socket, return the data dict.
Returns None if the query fails (node down, socket not ready, etc.).
"""
# Python one-liner that connects to the Unix socket, sends the JSON
# command, and prints the response. Runs inside the container.
script = (
"import socket,json,sys; "
"s=socket.socket(socket.AF_UNIX,socket.SOCK_STREAM); "
f"s.connect('{CONTROL_SOCKET}'); "
f"s.sendall(json.dumps({{'command':'{command}'}}).encode()+b'\\n'); "
"s.shutdown(socket.SHUT_WR); "
"chunks=[]; "
"[chunks.append(d) for d in iter(lambda:s.recv(65536),b'')]; "
"print(b''.join(chunks).decode())"
)
stdout = docker_exec_quiet(container, f"python3 -c \"{script}\"", timeout=timeout)
if stdout is None:
return None
try:
response = json.loads(stdout.strip())
except json.JSONDecodeError as e:
log.warning("Invalid JSON from %s: %s", container, e)
return None
if response.get("status") != "ok":
msg = response.get("message", "unknown error")
log.warning("Control query %s on %s failed: %s", command, container, msg)
return None
return response.get("data", {})
def query_tree(container: str) -> dict | None:
"""Query a node's spanning tree state."""
return query_node(container, "show_tree")
def query_mmp(container: str) -> dict | None:
"""Query a node's MMP metrics."""
return query_node(container, "show_mmp")
def query_peers(container: str) -> dict | None:
"""Query a node's peer list with MMP metrics."""
return query_node(container, "show_peers")
def snapshot_all_trees(topology: SimTopology) -> dict[str, dict]:
"""Query show_tree on all nodes, return {node_id: tree_data}.
Nodes that fail to respond are omitted from the result.
"""
result = {}
for node_id in sorted(topology.nodes):
container = topology.container_name(node_id)
data = query_tree(container)
if data is not None:
result[node_id] = data
else:
log.warning("No tree data from %s", node_id)
return result
def snapshot_all_mmp(topology: SimTopology) -> dict[str, dict]:
"""Query show_mmp on all nodes, return {node_id: mmp_data}.
Nodes that fail to respond are omitted from the result.
"""
result = {}
for node_id in sorted(topology.nodes):
container = topology.container_name(node_id)
data = query_mmp(container)
if data is not None:
result[node_id] = data
else:
log.warning("No MMP data from %s", node_id)
return result