webapp: imaging UX overhaul + image management CRUD

Imaging dashboard
- services/imaging_log_tail.py: parses dnsmasq leases, Apache access log,
  Samba per-host log files, and dnsmasq syslog (DHCP/TFTP). Synthesizes
  inferred sessions keyed by MAC for bays that have only touched the boot
  chain but not yet pushed to /imaging/status. Active window 90 min.
- imaging_status.list_sessions() merges inferred sessions into the dashboard
  list. Real client-pushed sessions win for the same MAC.
- imaging_status: stage_history field tracks every stage transition (capped
  30); sidecar .log file per serial records every log_lines push uncapped
  (read_full_log() caps detail-page response to 1 MB).
- delete_session/delete_all_sessions clean up sidecar .log too.
- New SSE endpoint /imaging/stream emits a session-list hash every 5s.
  Client fetches /imaging/tiles (HTML partial) on hash change and swaps
  #imaging-tiles innerHTML. Polling fallback at 15s if SSE drops.
- Tile-swap preserves scroll, filter input, expanded state via localStorage,
  and any LAPS input the operator is mid-pasting (swap skipped when a
  laps-input is focused).
- imaging.html: removed 15s location.reload(). Added live-status dot in
  header (gray idle / green SSE connected / red SSE lost).
- _imaging_tiles.html: shared partial used by both /imaging full render and
  /imaging/tiles SSE refresh. Inferred bays render with yellow border +
  log-inferred badge + no progress bar (stage_index inference is coarse).
- imaging_detail.html (new): per-bay forensics page at /imaging/session/
  <serial>. Session metadata grid, stage timeline table, full sidecar log
  with truncation indicator, Copy-support-summary button. Linked from each
  client-pushed tile.
- qr-render.js exposes window.renderAllQRs() so the SSE swap can re-render
  Intune device-ID QRs in the swapped-in tiles.

Image management
- services/image_registry.py: JSON registry of image types at
  {SAMBA_SHARE}/image-registry.json. Bootstraps from baked-in
  config.IMAGE_TYPES on first run. create/clone/delete/rename_friendly
  mutate the file then call reload() which rewrites config.IMAGE_TYPES +
  config.FRIENDLY_NAMES in place. Sidebar reflects on next request.
- app.py routes: /images/new, /images/<t>/clone, /images/<t>/delete (with
  optional content-wipe checkbox), /images/<t>/rename.
- dashboard.html: + New image type button + Clone/Delete per row, all in
  Bootstrap modals with confirmation copy.
- Clone copies Deploy/ tree but preserves symlinks to shared dirs (Out-of-
  box Drivers, Operating Systems, Packages) so disk usage stays low.
- Delete with content checked unlinks symlinks (does not follow into shared
  dirs).

Driver / package upload + orphan adoption
- services/images.py: upload_driver, adopt_orphan, remove_orphans,
  upload_package. Filename sanitization blocks path traversal.
- app.py routes: /images/<t>/drivers/upload, /images/<t>/drivers/adopt,
  /images/<t>/drivers/orphans/delete, /images/<t>/packages/upload.
- image_config.html: Upload .zip button + modal on Drivers section. Orphan
  drivers card-footer rebuilt as interactive list with per-row Adopt inline
  form (family + destinationDir inputs) and bulk select+delete.
- Upload .zip on Packages section with optional destinationDir field that
  appends a packages.json entry.

Configuration
- config.py: new env vars DNSMASQ_LEASES, APACHE_ACCESS_LOG, SAMBA_LOG_DIR,
  DNSMASQ_SYSLOG for the log-tailer.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
cproudlock
2026-05-30 13:21:06 -04:00
parent c74148a222
commit 69a1682a7f
12 changed files with 2034 additions and 251 deletions

View File

@@ -0,0 +1,254 @@
"""Mutable registry for image types.
Previously IMAGE_TYPES + FRIENDLY_NAMES were hardcoded in config.py and
adding a new image type required a code edit + Ansible re-run. Now they
live in a JSON file (REGISTRY_PATH) that the webapp can read and write at
runtime.
On import (called from app.py startup), reload() rewrites
config.IMAGE_TYPES and config.FRIENDLY_NAMES in-place so the rest of the
code (and the base.html sidebar context) sees the live list without
threading the registry through every call.
Schema:
{
"image_types": [
{"key": "gea-standard", "friendly_name": "GE Aerospace Standard"},
...
]
}
The seed comes from whatever IMAGE_TYPES/FRIENDLY_NAMES are baked into
config.py at first run. Once the registry file exists, it wins.
"""
from __future__ import annotations
import json
import os
import re
import shutil
import tempfile
from datetime import datetime
from typing import Optional
import config
# Persist alongside the Samba share so an Ansible re-run that recreates
# SAMBA_SHARE keeps the registry. Override via env for tests.
REGISTRY_PATH = os.environ.get(
"IMAGE_REGISTRY_PATH",
os.path.join(config.SAMBA_SHARE, "image-registry.json"),
)
# image_type keys are used as directory names + URL path components. Match
# the existing baked-in convention (lowercase + hyphens + alnum).
_VALID_KEY = re.compile(r"^[a-z][a-z0-9-]{1,63}$")
class RegistryError(Exception):
pass
def _read_file() -> dict:
try:
with open(REGISTRY_PATH, "r", encoding="utf-8") as f:
return json.load(f)
except (OSError, json.JSONDecodeError):
return {}
def _write_file(data: dict):
os.makedirs(os.path.dirname(REGISTRY_PATH), exist_ok=True)
fd, tmp = tempfile.mkstemp(
dir=os.path.dirname(REGISTRY_PATH),
prefix=".tmp-registry-",
suffix=".json",
)
try:
with os.fdopen(fd, "w", encoding="utf-8") as f:
json.dump(data, f, indent=2, ensure_ascii=False)
f.write("\n")
os.replace(tmp, REGISTRY_PATH)
except Exception:
try:
os.unlink(tmp)
except OSError:
pass
raise
def _seed_from_config() -> dict:
"""First-run bootstrap: build a registry dict from whatever's baked into
config.py. Preserves order so the sidebar layout doesn't shuffle on the
first reload."""
items = []
for key in config.IMAGE_TYPES:
items.append({
"key": key,
"friendly_name": config.FRIENDLY_NAMES.get(key, key),
"created_at": None,
})
return {"image_types": items}
def load_registry() -> dict:
"""Return the registry dict. Bootstraps from config.py on first read."""
data = _read_file()
if not data.get("image_types"):
data = _seed_from_config()
try:
_write_file(data)
except OSError:
# Read-only filesystem / no perms; fall back to in-memory seed.
pass
return data
def reload():
"""Refresh config.IMAGE_TYPES + config.FRIENDLY_NAMES from the registry
file. Called at startup and after every CRUD op. Mutates in place so
existing references stay valid."""
data = load_registry()
keys = [item["key"] for item in data.get("image_types", []) if "key" in item]
friendly = {item["key"]: item.get("friendly_name", item["key"])
for item in data.get("image_types", [])}
# In-place replace so module-level references still point at the same
# list/dict objects.
config.IMAGE_TYPES[:] = keys
config.FRIENDLY_NAMES.clear()
config.FRIENDLY_NAMES.update(friendly)
def _validate_key(key: str):
if not key or not _VALID_KEY.match(key):
raise RegistryError(
f"Invalid image_type key '{key}'. Must be lowercase alphanumeric "
f"+ hyphens, 2-64 chars, starting with a letter."
)
def create(key: str, friendly_name: str) -> dict:
"""Add a new image type with no on-disk content. Caller is responsible
for populating Deploy/ via the import flow afterwards."""
_validate_key(key)
if not friendly_name:
friendly_name = key
data = load_registry()
existing_keys = {item["key"] for item in data["image_types"]}
if key in existing_keys:
raise RegistryError(f"image_type '{key}' already exists")
data["image_types"].append({
"key": key,
"friendly_name": friendly_name,
"created_at": datetime.now().astimezone().isoformat(timespec="seconds"),
})
_write_file(data)
reload()
return {"key": key, "friendly_name": friendly_name}
def clone(src_key: str, dst_key: str, friendly_name: Optional[str] = None) -> dict:
"""Duplicate a source image type's on-disk content + register the new
key. Copies Deploy/ tree (which contains Control/ JSON configs +
unattend XML), Tools/ if present, and the top-level Sources/ symlink if
present. Shared dirs (Out-of-box Drivers, Operating Systems, Packages)
are re-symlinked into the new image root, not duplicated."""
_validate_key(dst_key)
data = load_registry()
keys = {item["key"] for item in data["image_types"]}
if src_key not in keys:
raise RegistryError(f"source image_type '{src_key}' not registered")
if dst_key in keys:
raise RegistryError(f"destination image_type '{dst_key}' already exists")
from services import fs # local import to avoid circular at module load
src_root = fs.image_root(src_key)
dst_root = fs.image_root(dst_key)
if os.path.exists(dst_root):
raise RegistryError(f"destination path already exists on disk: {dst_root}")
os.makedirs(dst_root, exist_ok=True)
if os.path.isdir(src_root):
for entry in os.listdir(src_root):
src_item = os.path.join(src_root, entry)
dst_item = os.path.join(dst_root, entry)
if os.path.islink(src_item):
# Preserve symlink (shared dir) instead of dereferencing.
target = os.readlink(src_item)
os.symlink(target, dst_item)
elif os.path.isdir(src_item):
shutil.copytree(src_item, dst_item, symlinks=True)
else:
shutil.copy2(src_item, dst_item)
if not friendly_name:
src_friendly = next(
(i["friendly_name"] for i in data["image_types"] if i["key"] == src_key),
src_key,
)
friendly_name = f"{src_friendly} (copy)"
data["image_types"].append({
"key": dst_key,
"friendly_name": friendly_name,
"created_at": datetime.now().astimezone().isoformat(timespec="seconds"),
"cloned_from": src_key,
})
_write_file(data)
reload()
return {"key": dst_key, "friendly_name": friendly_name, "cloned_from": src_key}
def delete(key: str, delete_content: bool = False) -> dict:
"""Remove key from the registry. When delete_content is True, also wipes
the on-disk Deploy/Tools/etc tree. Shared symlinked dirs are NOT
followed when deleting content (we unlink the symlink, not its target)."""
data = load_registry()
items = data.get("image_types", [])
found = [i for i in items if i["key"] == key]
if not found:
raise RegistryError(f"image_type '{key}' not registered")
data["image_types"] = [i for i in items if i["key"] != key]
_write_file(data)
reload()
removed_content = False
if delete_content:
from services import fs
root = fs.image_root(key)
if os.path.isdir(root):
# Walk top-level: unlink symlinks (don't follow into shared
# dirs), rmtree real directories, unlink files.
for entry in os.listdir(root):
p = os.path.join(root, entry)
try:
if os.path.islink(p):
os.unlink(p)
elif os.path.isdir(p):
shutil.rmtree(p)
else:
os.unlink(p)
except OSError:
pass
try:
os.rmdir(root)
except OSError:
pass
removed_content = True
return {"key": key, "removed_content": removed_content}
def rename_friendly(key: str, friendly_name: str) -> dict:
"""Update the human-readable name without touching the key or on-disk
content. Sidebar reflects on next request."""
if not friendly_name:
raise RegistryError("friendly_name required")
data = load_registry()
for item in data.get("image_types", []):
if item["key"] == key:
item["friendly_name"] = friendly_name
_write_file(data)
reload()
return {"key": key, "friendly_name": friendly_name}
raise RegistryError(f"image_type '{key}' not registered")

View File

@@ -1,11 +1,24 @@
"""Per-image-type state probes: status + config (drivers / OS / packages / models)."""
import os
import re
import shutil
import config
from services import fs
# File-name sanitizer for uploads. Strips anything outside [A-Za-z0-9._-]
# so a malicious filename can't break out of the destination directory.
_SAFE_NAME = re.compile(r"[^A-Za-z0-9._-]")
def _safe_filename(name: str) -> str:
base = os.path.basename(name or "")
cleaned = _SAFE_NAME.sub("_", base)
return cleaned or "upload.bin"
def image_status(image_type):
"""Return a dict describing the state of an image type."""
dp = fs.deploy_path(image_type)
@@ -121,3 +134,131 @@ def load_image_config(image_type):
"orphan_drivers": orphan_drivers,
"os_selection": os_selection,
}
# ---------------------------------------------------------------------------
# Driver / package upload + orphan adoption
# ---------------------------------------------------------------------------
DRIVERS_SUBDIR = "Out-of-box Drivers"
PACKAGES_SUBDIR = "Packages"
def _drivers_dir(image_type: str) -> str:
return os.path.join(fs.deploy_path(image_type), DRIVERS_SUBDIR)
def _packages_dir(image_type: str) -> str:
return os.path.join(fs.deploy_path(image_type), PACKAGES_SUBDIR)
def upload_driver(image_type: str, uploaded_file, family: str = "",
destination_dir: str = "", overwrite: bool = False) -> dict:
"""Save an uploaded driver .zip into the image's Out-of-box Drivers dir.
When family + destination_dir are provided, also append a HardwareDriver
.json entry so the driver is recognized at deploy time. Returns a dict
describing what landed on disk + whether an entry was registered."""
fname = _safe_filename(uploaded_file.filename)
if not fname.lower().endswith(".zip"):
raise ValueError("Driver upload must be a .zip file")
dst_dir = _drivers_dir(image_type)
os.makedirs(dst_dir, exist_ok=True)
dst_path = os.path.join(dst_dir, fname)
if os.path.exists(dst_path) and not overwrite:
raise FileExistsError(f"{fname} already exists in {DRIVERS_SUBDIR}/")
uploaded_file.save(dst_path)
registered = False
if family and destination_dir:
adopt_orphan(image_type, fname, family, destination_dir)
registered = True
return {
"filename": fname,
"path": dst_path,
"registered": registered,
}
def adopt_orphan(image_type: str, filename: str, family: str,
destination_dir: str) -> dict:
"""Append a HardwareDriver.json entry for an existing .zip in the image's
Out-of-box Drivers dir, so it stops showing up as an orphan + gets
deployed for the named hardware family. Idempotent: a second adopt with
the same filename is a no-op."""
safe_name = _safe_filename(filename)
drivers_path = _drivers_dir(image_type)
if not os.path.isfile(os.path.join(drivers_path, safe_name)):
raise FileNotFoundError(f"{safe_name} not found in {DRIVERS_SUBDIR}/")
if not family:
raise ValueError("family is required (matches HardwareModelSelection.Id)")
if not destination_dir:
raise ValueError("destination_dir is required")
ctrl = fs.control_path(image_type)
hw_file = os.path.join(ctrl, "HardwareDriver.json")
entries = fs.load_json(hw_file)
for e in entries:
if (e.get("FileName") or e.get("fileName") or "").lower() == safe_name.lower():
return {"filename": safe_name, "already_registered": True}
entries.append({
"FileName": safe_name,
"DestinationDir": destination_dir,
"family": family,
})
fs.save_json(hw_file, entries)
return {"filename": safe_name, "already_registered": False}
def remove_orphans(image_type: str, filenames: list[str]) -> dict:
"""Delete the named files from Out-of-box Drivers/. Caller is
responsible for confirming this is what the user wants (orphan files
have no JSON entry, so deleting them is safe in the sense that nothing
references them; but they may be stash for future adoption)."""
drivers_path = _drivers_dir(image_type)
removed, missing = [], []
for fn in filenames:
safe = _safe_filename(fn)
path = os.path.join(drivers_path, safe)
if not os.path.isfile(path):
missing.append(safe)
continue
try:
os.unlink(path)
removed.append(safe)
except OSError:
missing.append(safe)
return {"removed": removed, "missing": missing}
def upload_package(image_type: str, uploaded_file, destination_dir: str = "",
overwrite: bool = False) -> dict:
"""Save an uploaded package (any extension) into the image's Packages
dir, and append an entry to packages.json when destination_dir is set
so it deploys at imaging time."""
fname = _safe_filename(uploaded_file.filename)
dst_dir = _packages_dir(image_type)
os.makedirs(dst_dir, exist_ok=True)
dst_path = os.path.join(dst_dir, fname)
if os.path.exists(dst_path) and not overwrite:
raise FileExistsError(f"{fname} already exists in {PACKAGES_SUBDIR}/")
uploaded_file.save(dst_path)
registered = False
if destination_dir:
ctrl = fs.control_path(image_type)
pkg_file = os.path.join(ctrl, "packages.json")
entries = fs.load_json(pkg_file)
already = any(
(e.get("fileName") or e.get("FileName") or "").lower() == fname.lower()
for e in entries
)
if not already:
entries.append({
"fileName": fname,
"destinationDir": destination_dir,
})
fs.save_json(pkg_file, entries)
registered = True
return {
"filename": fname,
"path": dst_path,
"registered": registered,
}

View File

@@ -0,0 +1,376 @@
"""Infer in-progress imaging sessions from server-side logs.
The /imaging/status endpoint only sees clients that have reached the WinPE
startnet.cmd push (stage_index 2). A bay that gets stuck earlier (no DHCP,
TFTP fail, boot.wim download stall) is invisible to the operator.
This module tails dnsmasq leases + Apache access log + per-host Samba logs
and synthesizes a "session" record per active MAC, with a coarse inferred
stage based on which boot assets the client has actually fetched. Output
shape matches services/imaging_status so list_sessions() can merge both.
Inferred sessions carry source="inferred" and never overwrite a real
client-pushed session for the same serial (correlated by MAC when present).
"""
from __future__ import annotations
import os
import re
import time
from datetime import datetime, timezone
from typing import Optional
import config
# Active window: an inferred session is shown if any evidence is newer than
# this many seconds. Past that we assume the bay is idle / done / off.
INFERRED_ACTIVE_WINDOW_S = 90 * 60 # 90 min
# Tail size caps so a giant log doesn't pull the whole file into memory on
# each dashboard refresh.
APACHE_TAIL_BYTES = 512 * 1024 # 512 KB
SAMBA_TAIL_BYTES = 64 * 1024 # 64 KB per file
SYSLOG_TAIL_BYTES = 256 * 1024 # 256 KB
# Path prefixes Apache serves, mapped to coarse imaging stage signals.
# Order matters: we take the highest-numbered match (latest in the boot chain).
_APACHE_STAGE_HITS = [
("/menu.ipxe", ("stage_0_menu", 0)),
("/win11/boot/", ("stage_0_boot_pre", 0)),
("/win11/efi/", ("stage_0_boot_pre", 0)),
("/win11/sources/boot.wim", ("stage_1_wim_get", 1)),
("/win11/sources/", ("stage_1_wim_get", 1)),
]
# TFTP bootloader fetches arrive via dnsmasq, not Apache. Mapped same way.
_TFTP_STAGE_HITS = [
("undionly.kpxe", ("stage_0_tftp_bios", 0)),
("ipxe.efi", ("stage_0_tftp_uefi", 0)),
]
# Apache combined log: <ip> - - [DD/Mon/YYYY:HH:MM:SS +ZZZZ] "GET /path HTTP/x.y" status bytes "ref" "ua"
_APACHE_RE = re.compile(
r'^(?P<ip>\S+)\s+\S+\s+\S+\s+'
r'\[(?P<ts>[^\]]+)\]\s+'
r'"(?P<method>\S+)\s+(?P<path>\S+)\s+\S+"\s+'
r'(?P<status>\d+)\s+(?P<bytes>\S+)'
)
_APACHE_TS_FMT = "%d/%b/%Y:%H:%M:%S %z"
# dnsmasq syslog: "<MMM DD HH:MM:SS> host dnsmasq-tftp[pid]: sent /tftp/path to ip"
# or "dnsmasq-dhcp[pid]: DHCPACK(...) ip mac hostname"
_SYSLOG_DNSMASQ_RE = re.compile(
r'^(?P<mon>\w{3})\s+(?P<day>\d+)\s+(?P<time>\d{2}:\d{2}:\d{2})\s+'
r'\S+\s+dnsmasq-(?P<kind>\w+)\[\d+\]:\s+(?P<msg>.+)$'
)
def _tail_bytes(path: str, max_bytes: int) -> str:
"""Read last max_bytes of a text log, return decoded string. Returns
empty string if file is missing / unreadable. Skips partial first line."""
try:
size = os.path.getsize(path)
except OSError:
return ""
start = max(0, size - max_bytes)
try:
with open(path, "rb") as f:
f.seek(start)
data = f.read()
except OSError:
return ""
text = data.decode("utf-8", errors="replace")
if start > 0:
# Drop partial first line.
nl = text.find("\n")
if nl >= 0:
text = text[nl + 1:]
return text
def _now_iso() -> str:
return datetime.now().astimezone().isoformat(timespec="seconds")
def _epoch_to_iso(epoch: float) -> str:
return datetime.fromtimestamp(epoch).astimezone().isoformat(timespec="seconds")
def parse_leases() -> list[dict]:
"""dnsmasq.leases format: <expires_epoch> <mac> <ip> <hostname> <client_id>."""
out = []
try:
with open(config.DNSMASQ_LEASES, "r") as f:
for line in f:
parts = line.split()
if len(parts) < 4:
continue
try:
expires = int(parts[0])
except ValueError:
continue
out.append({
"expires": expires,
"mac": parts[1].lower(),
"ip": parts[2],
"hostname": parts[3] if parts[3] != "*" else "",
})
except OSError:
pass
return out
def parse_apache_hits() -> dict:
"""Return {ip: {"last_ts": epoch, "max_stage": int, "stage_label": str,
"paths": [recent_paths]}}."""
out: dict = {}
text = _tail_bytes(config.APACHE_ACCESS_LOG, APACHE_TAIL_BYTES)
for line in text.splitlines():
m = _APACHE_RE.match(line)
if not m:
continue
ip = m.group("ip")
path = m.group("path").lower()
try:
ts = datetime.strptime(m.group("ts"), _APACHE_TS_FMT).timestamp()
except ValueError:
continue
# Match against known stage prefixes; keep the highest-numbered.
for prefix, (label, stage) in _APACHE_STAGE_HITS:
if path.startswith(prefix):
rec = out.setdefault(ip, {
"last_ts": 0.0, "max_stage": -1,
"stage_label": "", "paths": [],
})
if ts > rec["last_ts"]:
rec["last_ts"] = ts
if stage > rec["max_stage"]:
rec["max_stage"] = stage
rec["stage_label"] = label
rec["paths"].append(path)
# Cap path list so a busy client doesn't explode the dict.
rec["paths"] = rec["paths"][-10:]
break
return out
def parse_syslog_tftp() -> dict:
"""Return {mac_or_ip: {"last_ts": epoch, "max_stage": int,
"stage_label": str, "evidence": [str]}} from dnsmasq syslog lines."""
out: dict = {}
text = _tail_bytes(config.DNSMASQ_SYSLOG, SYSLOG_TAIL_BYTES)
year = datetime.now().year
for line in text.splitlines():
m = _SYSLOG_DNSMASQ_RE.match(line)
if not m:
continue
kind = m.group("kind")
msg = m.group("msg")
# Best-effort ts parse. Syslog format has no year; assume current.
try:
ts = datetime.strptime(
f"{year} {m.group('mon')} {m.group('day')} {m.group('time')}",
"%Y %b %d %H:%M:%S"
).timestamp()
except ValueError:
continue
if kind == "tftp":
# "sent /srv/tftp/ipxe.efi to 172.16.9.42"
tm = re.search(r'(?:sent|read|file)\s+(?P<file>\S+)\s+to\s+(?P<ip>\S+)', msg)
if not tm:
continue
key = tm.group("ip")
fname = os.path.basename(tm.group("file")).lower()
for needle, (label, stage) in _TFTP_STAGE_HITS:
if needle in fname:
rec = out.setdefault(key, {
"last_ts": 0.0, "max_stage": -1,
"stage_label": "", "evidence": [],
})
if ts > rec["last_ts"]:
rec["last_ts"] = ts
if stage > rec["max_stage"]:
rec["max_stage"] = stage
rec["stage_label"] = label
rec["evidence"].append(f"tftp:{fname}")
rec["evidence"] = rec["evidence"][-5:]
break
elif kind == "dhcp":
# DHCPACK lines confirm a lease handshake completed; useful as a
# very-early "this MAC is alive" signal even before TFTP fires.
dm = re.search(
r'DHCP(?:ACK|OFFER|REQUEST)\([^)]+\)\s+'
r'(?P<ip>\d+\.\d+\.\d+\.\d+)\s+'
r'(?P<mac>[0-9a-f:]{17})',
msg
)
if not dm:
continue
key = dm.group("ip")
rec = out.setdefault(key, {
"last_ts": 0.0, "max_stage": -1,
"stage_label": "", "evidence": [],
})
if ts > rec["last_ts"]:
rec["last_ts"] = ts
if rec["max_stage"] < 0:
rec["max_stage"] = 0
rec["stage_label"] = "stage_0_dhcp"
rec["evidence"].append(f"dhcp:{dm.group('mac')}")
rec["evidence"] = rec["evidence"][-5:]
return out
def parse_samba_hits() -> dict:
"""Return {ip: {"last_ts": epoch, "shares": set[str]}} for clients that
have opened a Samba session in the recent tail. Samba writes per-host log
files: /var/log/samba/log.<ip>. Mere existence + recent mtime = client
has connected. Share names from inside the file are best-effort."""
out: dict = {}
log_dir = config.SAMBA_LOG_DIR
if not os.path.isdir(log_dir):
return out
try:
entries = os.listdir(log_dir)
except OSError:
return out
for name in entries:
if not name.startswith("log."):
continue
ip = name[len("log."):]
# Skip the daemon's own log files (log.smbd, log.nmbd, log.winbindd).
if ip in ("smbd", "nmbd", "winbindd") or "." not in ip:
continue
path = os.path.join(log_dir, name)
try:
st = os.stat(path)
except OSError:
continue
rec = {"last_ts": st.st_mtime, "shares": set()}
text = _tail_bytes(path, SAMBA_TAIL_BYTES)
for share in ("winpeapps", "clonezilla", "enrollment", "image-upload", "blancco-reports"):
if f"[{share}]" in text or f"connect to service {share}" in text:
rec["shares"].add(share)
out[ip] = rec
return out
def infer_sessions(
active_window_s: int = INFERRED_ACTIVE_WINDOW_S,
) -> list[dict]:
"""Build inferred-session records from all log sources. Keyed by MAC when
available, else IP. Only returns sessions with evidence in the active
window."""
leases = parse_leases()
apache = parse_apache_hits()
samba = parse_samba_hits()
tftp = parse_syslog_tftp()
# IP -> lease (MAC, hostname).
by_ip = {lease["ip"]: lease for lease in leases}
cutoff = time.time() - active_window_s
sessions: dict[str, dict] = {}
def _ensure(ip: str) -> dict:
lease = by_ip.get(ip, {})
key = lease.get("mac") or ip
rec = sessions.get(key)
if rec is None:
rec = {
"source": "inferred",
"serial": None,
"mac": lease.get("mac", ""),
"ip": ip,
"hostname_target": lease.get("hostname", ""),
"started_at": None,
"last_updated": None,
"stage_index": 0,
"stage_total": 8,
"current_stage": "",
"status": "in_progress",
"evidence": [],
"log_tail": [],
}
sessions[key] = rec
return rec
def _bump_stage(rec: dict, stage: int, label: str):
if stage > rec["stage_index"]:
rec["stage_index"] = stage
rec["current_stage"] = label
def _touch(rec: dict, ts: float):
iso = _epoch_to_iso(ts)
if rec["started_at"] is None or iso < rec["started_at"]:
rec["started_at"] = iso
if rec["last_updated"] is None or iso > rec["last_updated"]:
rec["last_updated"] = iso
for ip, hit in tftp.items():
if hit["last_ts"] < cutoff:
continue
rec = _ensure(ip)
_touch(rec, hit["last_ts"])
_bump_stage(rec, hit["max_stage"], hit["stage_label"])
rec["evidence"].extend(hit["evidence"])
for ip, hit in apache.items():
if hit["last_ts"] < cutoff:
continue
rec = _ensure(ip)
_touch(rec, hit["last_ts"])
_bump_stage(rec, hit["max_stage"], hit["stage_label"])
for p in hit["paths"]:
rec["evidence"].append(f"http:{p}")
for ip, hit in samba.items():
if hit["last_ts"] < cutoff:
continue
rec = _ensure(ip)
_touch(rec, hit["last_ts"])
# WinPE-stage SMB activity = boot.wim applied + apps stage starting.
_bump_stage(rec, 2, "stage_2_smb_winpe")
for share in sorted(hit["shares"]):
rec["evidence"].append(f"smb:{share}")
# Trim evidence + bake log_tail for display.
for rec in sessions.values():
rec["evidence"] = rec["evidence"][-15:]
rec["log_tail"] = list(rec["evidence"])
return list(sessions.values())
def merge_with_client_sessions(
client_sessions: list[dict],
inferred: Optional[list[dict]] = None,
) -> list[dict]:
"""Return a single list. Real (client-pushed) sessions win over inferred
ones for the same MAC. Inferred sessions are dropped when a real session
with the same MAC already exists, since the real one has strictly more
information."""
if inferred is None:
inferred = infer_sessions()
real_macs = {(s.get("mac") or "").lower() for s in client_sessions if s.get("mac")}
real_ips = {s.get("ip") for s in client_sessions if s.get("ip")}
# Tag real sessions so the template can style them differently from
# inferred ones (and not break older sessions that lack the field).
for s in client_sessions:
s.setdefault("source", "client")
merged = list(client_sessions)
for inf in inferred:
mac = (inf.get("mac") or "").lower()
ip = inf.get("ip")
if mac and mac in real_macs:
continue
if ip and ip in real_ips:
continue
merged.append(inf)
merged.sort(key=lambda s: s.get("last_updated") or "", reverse=True)
return merged

View File

@@ -20,8 +20,16 @@ from typing import Optional
import config
# How many recent log lines to keep per session.
# How many recent log lines to keep per session in the JSON (dashboard tile
# quick view). The full unbounded log is appended to a sidecar .log file
# next to the .json so the detail page can show everything.
LOG_TAIL_MAX = 50
# Cap how many stage transitions we record per session (bounds JSON size on
# pathological loops; 30 covers more than any real run uses).
STAGE_HISTORY_MAX = 30
# Detail page caps how many bytes of the sidecar .log it sends to the
# browser, to avoid blowing up the response for a runaway log.
DETAIL_LOG_MAX_BYTES = 1024 * 1024 # 1 MB
# Sessions older than this are considered stale and dropped from the dashboard
# "active" list. Still readable individually.
ACTIVE_WINDOW_HOURS = 6
@@ -40,6 +48,30 @@ def _path_for(serial: str) -> str:
return os.path.join(config.IMAGING_DIR, f"{safe}.json")
def _log_path_for(serial: str) -> str:
safe = _SAFE_SERIAL.sub("_", serial.strip()) or "unknown"
return os.path.join(config.IMAGING_DIR, f"{safe}.log")
def _append_full_log(serial: str, lines):
"""Best-effort append to the per-serial sidecar log file. Each line is
timestamped. Failures are swallowed (status-tracking is not the
authoritative log source; the .log is a convenience for the detail
page)."""
if not lines:
return
if isinstance(lines, str):
lines = [lines]
try:
_ensure_dir()
with open(_log_path_for(serial), "a") as f:
ts = _now_iso()
for line in lines:
f.write(f"{ts} {line}\n")
except OSError:
pass
def _now_iso() -> str:
return datetime.now().astimezone().isoformat(timespec="seconds")
@@ -100,7 +132,9 @@ def update_session(payload: dict) -> dict:
# Fresh state after a rewind - mint a new started_at.
state["started_at"] = _now_iso()
# Append any new log lines (preserve old; cap to LOG_TAIL_MAX).
# Append any new log lines: capped tail in the JSON for the dashboard
# quick view, and unbounded append to the sidecar .log for the detail
# page.
new_lines = payload.pop("log_lines", None)
if new_lines:
if isinstance(new_lines, str):
@@ -108,6 +142,33 @@ def update_session(payload: dict) -> dict:
tail = list(state.get("log_tail", []))
tail.extend(new_lines)
state["log_tail"] = tail[-LOG_TAIL_MAX:]
_append_full_log(serial, new_lines)
# Stage history: record a transition row whenever stage_index increases
# or status changes. Bounded to STAGE_HISTORY_MAX so a bouncing client
# can't blow up the JSON. The dashboard tile only needs current state;
# the detail page renders the timeline from this list.
history = list(state.get("stage_history", []))
try:
new_idx = int(payload.get("stage_index") or 0)
except (TypeError, ValueError):
new_idx = 0
try:
old_idx = int(state.get("stage_index") or 0)
except (TypeError, ValueError):
old_idx = 0
new_status = payload.get("status") or state.get("status") or "in_progress"
old_status = state.get("status") or ""
stage_changed = new_idx > old_idx
status_changed = new_status != old_status and new_status in ("succeeded", "failed")
if stage_changed or status_changed or not history:
history.append({
"ts": _now_iso(),
"stage_index": new_idx or old_idx,
"current_stage": payload.get("current_stage") or state.get("current_stage", ""),
"status": new_status,
})
state["stage_history"] = history[-STAGE_HISTORY_MAX:]
for key, value in payload.items():
if value is None or value == "":
@@ -132,8 +193,14 @@ def update_session(payload: dict) -> dict:
return state
def list_sessions() -> list[dict]:
"""Return all sessions sorted by last_updated desc."""
def list_sessions(include_inferred: bool = True) -> list[dict]:
"""Return all sessions sorted by last_updated desc.
When include_inferred is True (default for the dashboard), also pull
server-side log-tail evidence and append synthesized sessions for any
bay that has touched DHCP/TFTP/boot.wim but not yet pushed status.
Real client-pushed sessions always win for the same MAC.
"""
_ensure_dir()
out: list[dict] = []
for name in os.listdir(config.IMAGING_DIR):
@@ -145,7 +212,11 @@ def list_sessions() -> list[dict]:
out.append(json.load(f))
except (json.JSONDecodeError, OSError):
continue
out.sort(key=lambda s: s.get("last_updated", ""), reverse=True)
if include_inferred:
from services import imaging_log_tail
out = imaging_log_tail.merge_with_client_sessions(out)
else:
out.sort(key=lambda s: s.get("last_updated", ""), reverse=True)
return out
@@ -160,27 +231,58 @@ def get_session(serial: str) -> Optional[dict]:
return None
def read_full_log(serial: str, max_bytes: int = DETAIL_LOG_MAX_BYTES) -> tuple[str, bool]:
"""Return (text, truncated). Reads the trailing max_bytes of the sidecar
.log. `truncated` is True when the file was larger than max_bytes and
the leading slice was dropped."""
path = _log_path_for(serial)
try:
size = os.path.getsize(path)
except OSError:
return ("", False)
truncated = size > max_bytes
start = max(0, size - max_bytes)
try:
with open(path, "rb") as f:
f.seek(start)
data = f.read()
except OSError:
return ("", False)
text = data.decode("utf-8", errors="replace")
if truncated:
nl = text.find("\n")
if nl >= 0:
text = text[nl + 1:]
return (text, truncated)
def delete_session(serial: str) -> bool:
path = _path_for(serial)
if not os.path.isfile(path):
return False
try:
os.unlink(path)
return True
except OSError:
return False
# Best-effort sidecar cleanup.
try:
os.unlink(_log_path_for(serial))
except OSError:
pass
return True
def delete_all_sessions() -> int:
"""Wipe every per-bay JSON in IMAGING_DIR. Returns count removed."""
"""Wipe every per-bay JSON + sidecar .log in IMAGING_DIR. Returns count
of JSON files removed."""
_ensure_dir()
removed = 0
for fn in os.listdir(config.IMAGING_DIR):
if not fn.endswith(".json"):
continue
try:
os.unlink(os.path.join(config.IMAGING_DIR, fn))
removed += 1
except OSError:
pass
if fn.endswith(".json") or fn.endswith(".log"):
try:
os.unlink(os.path.join(config.IMAGING_DIR, fn))
except OSError:
continue
if fn.endswith(".json"):
removed += 1
return removed