feat: events journal write primitives (log_event, backfill, get_events_journal)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
2026-07-10 14:47:42 -04:00
co-authored by Claude Fable 5
parent 8d268adf70
commit fb5570155b
2 changed files with 164 additions and 4 deletions
+54 -4
View File
@@ -11,7 +11,7 @@ import os
import asyncio
from datetime import datetime
from pathlib import Path
from typing import Dict, Any, Optional
from typing import Dict, Any, Optional, Callable, Iterable, List, Tuple, Union
logger = logging.getLogger(__name__)
@@ -149,10 +149,35 @@ class MessageJournal:
self._current_size += len(json_bytes)
logger.debug(f"Logged message from {addr[0]}: {msg.get('ID', 'UNKNOWN')}")
except Exception as e:
logger.error(f"Error writing to journal: {e}")
async def log_event(self, event: Dict[str, Any]):
"""Write a caller-provided dict verbatim as one JSONL line (with rotation)."""
if not self.enabled or not self._initialized:
return
async with self._lock:
try:
line = json.dumps(event, separators=(',', ':')) + '\n'
nbytes = len(line.encode('utf-8'))
if self._current_size + nbytes > self.max_size:
await self._rotate()
if self._file_handle:
self._file_handle.write(line)
self._file_handle.flush()
self._current_size += nbytes
except Exception as e:
logger.error(f"Error writing event to journal: {e}")
async def backfill(self, events: Iterable[Dict[str, Any]]):
"""One-time seed: write *events* only if the journal file is currently empty."""
if not self.enabled or not self._initialized or self._current_size > 0:
return
for ev in events:
if isinstance(ev, dict):
await self.log_event(ev)
async def _rotate(self):
"""
Rotate the journal file.
@@ -332,7 +357,7 @@ def get_journal(config: Optional[Dict[str, Any]] = None) -> MessageJournal:
async def log_message(msg: Dict[str, Any], addr: tuple, timestamp: Optional[float] = None):
"""
Convenience function to log a message using the global journal.
Args:
msg: Parsed message dictionary
addr: Source address (ip, port) tuple
@@ -340,3 +365,28 @@ async def log_message(msg: Dict[str, Any], addr: tuple, timestamp: Optional[floa
"""
journal = get_journal()
await journal.log_message(msg, addr, timestamp)
# Global events journal instance (human-readable event log, written by notify.eventlog)
_events_journal_instance: Optional[MessageJournal] = None
def get_events_journal(config: Optional[Dict[str, Any]] = None) -> MessageJournal:
"""Get or create the global events journal instance.
Uses the events_journal_* config keys; shares journal_dir and
journal_enabled with the raw-datagram journal.
"""
global _events_journal_instance
if _events_journal_instance is None:
cfg = config or {}
_events_journal_instance = MessageJournal(
{
'journal_dir': cfg.get('journal_dir', '/var/log/heartbeat'),
'journal_file': cfg.get('events_journal_file', 'events.journal'),
'journal_max_size': cfg.get('events_journal_max_size', 10 * 1024 * 1024),
'journal_max_backups': cfg.get('events_journal_max_backups', 10),
'journal_enabled': cfg.get('journal_enabled', True),
}
)
return _events_journal_instance
+110
View File
@@ -0,0 +1,110 @@
"""Tests for the dedicated events journal (write path and read path)."""
import asyncio
import json
from hbd.server import journal
def _make_journal(tmp_path, **overrides):
cfg = {"journal_dir": str(tmp_path), "journal_file": "events.journal"}
cfg.update(overrides)
j = journal.MessageJournal(cfg)
assert asyncio.run(j.initialize())
return j
def _read_lines(tmp_path, name="events.journal"):
return (tmp_path / name).read_text(encoding="utf-8").splitlines()
EV1 = {"ts": 1000.0, "host": "h1", "level": "INFO", "service": None, "message": "host up"}
EV2 = {"ts": 2000.0, "host": "h2", "level": "CRITICAL", "service": "cpu", "message": "cpu high"}
def test_log_event_writes_one_json_line(tmp_path):
j = _make_journal(tmp_path)
asyncio.run(j.log_event(EV1))
asyncio.run(j.close())
lines = _read_lines(tmp_path)
assert len(lines) == 1
assert json.loads(lines[0]) == EV1
def test_log_event_appends_in_order(tmp_path):
j = _make_journal(tmp_path)
asyncio.run(j.log_event(EV1))
asyncio.run(j.log_event(EV2))
asyncio.run(j.close())
lines = _read_lines(tmp_path)
assert [json.loads(ln)["ts"] for ln in lines] == [1000.0, 2000.0]
def test_log_event_rotates_at_max_size(tmp_path):
# max_size fits one serialized event line (~77 bytes) but not two, so the
# second write triggers exactly one rotation
j = _make_journal(tmp_path, journal_max_size=120)
asyncio.run(j.log_event(EV1))
asyncio.run(j.log_event(EV2))
asyncio.run(j.close())
backups = list(tmp_path.glob("events.journal.*"))
assert len(backups) == 1
assert json.loads(backups[0].read_text().splitlines()[0]) == EV1
assert json.loads(_read_lines(tmp_path)[0]) == EV2
def test_log_event_noop_when_disabled(tmp_path):
j = journal.MessageJournal(
{"journal_dir": str(tmp_path), "journal_file": "events.journal", "journal_enabled": False}
)
asyncio.run(j.initialize())
asyncio.run(j.log_event(EV1))
assert not (tmp_path / "events.journal").exists()
def test_backfill_seeds_empty_journal(tmp_path):
j = _make_journal(tmp_path)
asyncio.run(j.backfill([EV1, EV2]))
asyncio.run(j.close())
lines = _read_lines(tmp_path)
assert len(lines) == 2
assert json.loads(lines[0]) == EV1
def test_backfill_skipped_when_journal_nonempty(tmp_path):
(tmp_path / "events.journal").write_text(json.dumps(EV1) + "\n")
j = _make_journal(tmp_path) # initialize() picks up the existing size
asyncio.run(j.backfill([EV2]))
asyncio.run(j.close())
assert len(_read_lines(tmp_path)) == 1
def test_get_events_journal_uses_events_config_keys(tmp_path):
journal._events_journal_instance = None
try:
ej = journal.get_events_journal(
{
"journal_dir": str(tmp_path),
"events_journal_file": "ev.jsonl",
"events_journal_max_size": 12345,
"events_journal_max_backups": 3,
}
)
assert ej.journal_file == "ev.jsonl"
assert ej.max_size == 12345
assert ej.max_backups == 3
assert ej.journal_dir == tmp_path
# singleton: second call returns the same instance
assert journal.get_events_journal() is ej
finally:
journal._events_journal_instance = None
def test_get_events_journal_defaults(tmp_path):
journal._events_journal_instance = None
try:
ej = journal.get_events_journal({"journal_dir": str(tmp_path)})
assert ej.journal_file == "events.journal"
assert ej.max_size == 10 * 1024 * 1024
assert ej.max_backups == 10
finally:
journal._events_journal_instance = None