3tears-epoch 0.14.0__tar.gz

3tears_epoch-0.14.0/.gitignore

@@ -0,0 +1,216 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+#poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+#pdm.lock
+#pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+#pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# Abstra
+# Abstra is an AI-powered process automation framework.
+# Ignore directories containing user credentials, local state, and settings.
+# Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#  and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#  you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+# Claude Code local state
+.claude/
+
+# prawduct session evidence (local governance artifacts, never shipped)
+.prawduct/
+
+# macOS folder metadata
+.DS_Store

3tears_epoch-0.14.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Mark Pace
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

3tears_epoch-0.14.0/PKG-INFO

@@ -0,0 +1,74 @@
+Metadata-Version: 2.4
+Name: 3tears-epoch
+Version: 0.14.0
+Summary: Generation-stamped config epochs with NATS broadcast and per-message echo for cross-pod cache reload coherence
+Project-URL: Repository, https://github.com/pacepace/3tears
+Author: pace
+License-Expression: MIT
+License-File: LICENSE
+Classifier: Development Status :: 3 - Alpha
+Classifier: Framework :: AsyncIO
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Software Development :: Libraries
+Classifier: Typing :: Typed
+Requires-Python: >=3.14
+Requires-Dist: 3tears
+Requires-Dist: 3tears-nats
+Requires-Dist: 3tears-observe
+Requires-Dist: pydantic>=2
+Description-Content-Type: text/markdown
+
+# 3tears-epoch
+
+Generation-stamped configuration epochs with NATS broadcast and per-message echo for cross-pod cache-reload coherence.
+
+## Why
+
+Multiple in-memory configuration caches across the platform need to stay coherent across pods on admin writes:
+
+- a model capabilities registry (registered at startup from the `models` table)
+- a catalog cache (`gateway_models` + `gateway_providers` + `gateway_credit_rates`)
+- per-tool MCP RBAC grants
+
+Pure NATS broadcast (push) ships with a missed-message hole: a pod that didn't receive the broadcast (subscriber blip, pod just started during the window, JetStream redelivery edge) stays stale. Pure polling (pull) is correct but expensive on hot paths.
+
+This package combines both: a strictly-monotonic generation number (epoch) per *subject*, durable in Postgres, broadcast best-effort via NATS, and echoed in every relevant response message so consumers detect staleness on the next read and lazy-pull. Push for speed, pull for correctness.
+
+This is the standard pattern from etcd `mod_revision` + watch, K8s `resourceVersion` + informer, Envoy xDS `version_info` + ACK, DNS SOA serial + secondary refresh.
+
+## Identity
+
+The unit of identity is the **NATS subject path**. Each consumer:
+
+1. Defines or uses an existing `Subject` builder for the configuration domain it owns (e.g. `Subjects.capabilities_epoch()` -> `capabilities.epoch`).
+2. Calls `EpochClient.bump(subject, payload=...)` after committing the row mutation that motivates the reload.
+3. Subscribes via `EpochListener.subscribe(subject, on_bump=...)` from sibling pods.
+
+The `platform.config_epochs` row PK is the subject path string. Postgres is the source of truth; the NATS broadcast is best-effort. A subscriber that missed every broadcast still catches up on the next request whose response echoes the higher epoch (per-message echo is consumer-side wiring; the framework supplies the building blocks).
+
+## Wire envelope
+
+`EpochBumpMessage` is a frozen Pydantic v2 model:
+
+- `subject_path: str` is the namespaced subject the bump targets (matches the row PK)
+- `epoch: int` is the new strictly-monotonic value
+- `payload: dict[str, Any] | None` is an opaque hint for the consumer's reload callback (e.g. `{"model_id": "...", "action": "create"}`)
+
+The framework never inspects `payload`. Consumers parse if useful, ignore if not. The hint exists so a domain that only changes one row can avoid reloading the entire derived view.
+
+## Postgres
+
+```
+CREATE TABLE IF NOT EXISTS config_epochs (
+    subject_path TEXT PRIMARY KEY,
+    epoch BIGINT NOT NULL DEFAULT 0,
+    payload JSONB,
+    date_updated TIMESTAMPTZ NOT NULL DEFAULT now()
+);
+```
+
+`bump(subject, payload)` runs `INSERT ... ON CONFLICT (subject_path) DO UPDATE SET epoch = config_epochs.epoch + 1, payload = $2, date_updated = now() RETURNING epoch`. Atomic; concurrent bumps from different writers serialize on the row lock.
+
+Migration ships as a PLATFORM-scope `PackageMigrations` registration so consumers wire it via the canonical `MigrationRunner` alongside the rest of their platform tables.

3tears_epoch-0.14.0/README.md

@@ -0,0 +1,52 @@
+# 3tears-epoch
+
+Generation-stamped configuration epochs with NATS broadcast and per-message echo for cross-pod cache-reload coherence.
+
+## Why
+
+Multiple in-memory configuration caches across the platform need to stay coherent across pods on admin writes:
+
+- a model capabilities registry (registered at startup from the `models` table)
+- a catalog cache (`gateway_models` + `gateway_providers` + `gateway_credit_rates`)
+- per-tool MCP RBAC grants
+
+Pure NATS broadcast (push) ships with a missed-message hole: a pod that didn't receive the broadcast (subscriber blip, pod just started during the window, JetStream redelivery edge) stays stale. Pure polling (pull) is correct but expensive on hot paths.
+
+This package combines both: a strictly-monotonic generation number (epoch) per *subject*, durable in Postgres, broadcast best-effort via NATS, and echoed in every relevant response message so consumers detect staleness on the next read and lazy-pull. Push for speed, pull for correctness.
+
+This is the standard pattern from etcd `mod_revision` + watch, K8s `resourceVersion` + informer, Envoy xDS `version_info` + ACK, DNS SOA serial + secondary refresh.
+
+## Identity
+
+The unit of identity is the **NATS subject path**. Each consumer:
+
+1. Defines or uses an existing `Subject` builder for the configuration domain it owns (e.g. `Subjects.capabilities_epoch()` -> `capabilities.epoch`).
+2. Calls `EpochClient.bump(subject, payload=...)` after committing the row mutation that motivates the reload.
+3. Subscribes via `EpochListener.subscribe(subject, on_bump=...)` from sibling pods.
+
+The `platform.config_epochs` row PK is the subject path string. Postgres is the source of truth; the NATS broadcast is best-effort. A subscriber that missed every broadcast still catches up on the next request whose response echoes the higher epoch (per-message echo is consumer-side wiring; the framework supplies the building blocks).
+
+## Wire envelope
+
+`EpochBumpMessage` is a frozen Pydantic v2 model:
+
+- `subject_path: str` is the namespaced subject the bump targets (matches the row PK)
+- `epoch: int` is the new strictly-monotonic value
+- `payload: dict[str, Any] | None` is an opaque hint for the consumer's reload callback (e.g. `{"model_id": "...", "action": "create"}`)
+
+The framework never inspects `payload`. Consumers parse if useful, ignore if not. The hint exists so a domain that only changes one row can avoid reloading the entire derived view.
+
+## Postgres
+
+```
+CREATE TABLE IF NOT EXISTS config_epochs (
+    subject_path TEXT PRIMARY KEY,
+    epoch BIGINT NOT NULL DEFAULT 0,
+    payload JSONB,
+    date_updated TIMESTAMPTZ NOT NULL DEFAULT now()
+);
+```
+
+`bump(subject, payload)` runs `INSERT ... ON CONFLICT (subject_path) DO UPDATE SET epoch = config_epochs.epoch + 1, payload = $2, date_updated = now() RETURNING epoch`. Atomic; concurrent bumps from different writers serialize on the row lock.
+
+Migration ships as a PLATFORM-scope `PackageMigrations` registration so consumers wire it via the canonical `MigrationRunner` alongside the rest of their platform tables.

3tears_epoch-0.14.0/pyproject.toml

@@ -0,0 +1,52 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "3tears-epoch"
+version = "0.14.0"
+description = "Generation-stamped config epochs with NATS broadcast and per-message echo for cross-pod cache reload coherence"
+readme = "README.md"
+requires-python = ">=3.14"
+authors = [{name = "pace"}]
+license = "MIT"
+license-files = ["LICENSE"]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Framework :: AsyncIO",
+    "Intended Audience :: Developers",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Software Development :: Libraries",
+    "Typing :: Typed",
+]
+dependencies = [
+    "3tears",
+    "3tears-nats",
+    "3tears-observe",
+    "pydantic>=2",
+]
+
+[project.urls]
+Repository = "https://github.com/pacepace/3tears"
+
+[tool.hatch.build.targets.wheel]
+packages = ["src/threetears"]
+
+[tool.uv.sources]
+3tears = { workspace = true }
+3tears-nats = { workspace = true }
+3tears-observe = { workspace = true }
+
+[tool.mypy]
+strict = true
+mypy_path = "src"
+packages = ["threetears.epoch"]
+explicit_package_bases = true
+ignore_missing_imports = true
+
+# no [tool.pytest.ini_options] block: pytest's rootdir detection
+# picks the closest pyproject with [tool.pytest.ini_options], so a
+# per-package block hides the workspace-level conftest.py and
+# unregisters the canonical ``pytest_plugins = ["threetears.core.
+# testing.fixtures"]`` line. inherit from the workspace instead.

3tears_epoch-0.14.0/src/threetears/epoch/__init__.py

@@ -0,0 +1,44 @@
+"""3tears-epoch: cross-pod config-epoch coherence.
+
+generation-stamped configuration epochs with NATS broadcast (push) and
+per-message epoch echo (pull-on-stale) to keep in-memory caches
+coherent across pods on admin writes.
+
+provides three modules:
+
+- :mod:`threetears.epoch.wire` -- :class:`EpochBumpMessage` typed wire
+  envelope
+- :mod:`threetears.epoch.client` -- :class:`EpochClient` publish-side
+  bump + current-read against ``config_epochs``
+- :mod:`threetears.epoch.listener` -- :class:`EpochListener`
+  subscribe-side dispatcher with monotonic dedupe + echo helper
+"""
+
+from __future__ import annotations
+
+# Version derived from pyproject.toml so the metadata is the single
+# source of truth -- a future release that bumps pyproject without
+# updating ``__init__.py`` can't drift the runtime ``__version__``.
+# The except guard handles the rare case where the package isn't
+# installed via importlib.metadata (e.g. running directly from a
+# checked-out source tree without ``uv sync``); the fallback keeps
+# imports working but reports ``unknown`` rather than crashing.
+from importlib.metadata import PackageNotFoundError as _PackageNotFoundError
+from importlib.metadata import version as _version
+
+try:
+    __version__ = _version("3tears-epoch")
+except _PackageNotFoundError:  # pragma: no cover - dev fallback
+    __version__ = "unknown"
+
+from threetears.epoch.client import EpochClient, PoolLike
+from threetears.epoch.listener import BumpCallback, EpochListener
+from threetears.epoch.wire import EpochBumpMessage
+
+__all__ = [
+    "BumpCallback",
+    "EpochBumpMessage",
+    "EpochClient",
+    "EpochListener",
+    "PoolLike",
+]

3tears_epoch-0.14.0/src/threetears/epoch/client.py

@@ -0,0 +1,214 @@
+"""epoch client -- atomic Postgres bump plus best-effort NATS broadcast.
+
+:class:`EpochClient` is the publish-side companion to
+:class:`~threetears.epoch.listener.EpochListener`. it owns one pair of
+operations against the ``config_epochs`` table:
+
+- :meth:`current` -- read the latest epoch for a subject (used by
+  listeners on cold start and by periodic catch-up ticks)
+- :meth:`bump` -- atomically increment the epoch for a subject, then
+  publish an :class:`~threetears.epoch.wire.EpochBumpMessage` on the
+  same subject so sibling pods notice immediately
+
+postgres is the source of truth. nats is the fast notify. a missed
+broadcast is recovered by either the next periodic tick (calls
+:meth:`current`) or by the next response that echoes a higher epoch
+(per-message echo, consumer-side).
+
+the row PK in ``config_epochs`` is the subject path string. the
+publisher always knows its own current epoch (just returned by the
+``RETURNING epoch`` clause); subscribers learn it from broadcasts and
+from echoes. this is the etcd ``mod_revision`` shape minus the
+multi-key transaction support: every domain is independent.
+"""
+
+from __future__ import annotations
+
+import json
+from typing import Any, Protocol
+
+from threetears.nats import NatsClient
+from threetears.nats.errors import PublishError
+from threetears.nats.subjects import Subject
+from threetears.observe import get_logger, traced
+
+from threetears.epoch.wire import EpochBumpMessage
+
+__all__ = [
+    "EpochClient",
+    "PoolLike",
+]
+
+log = get_logger(__name__)
+
+
+class PoolLike(Protocol):
+    """minimal pool surface :class:`EpochClient` needs.
+
+    matches the top-level ``fetchrow``/``fetchval`` methods that
+    :class:`asyncpg.Pool` exposes (the pool acquires + releases a
+    connection internally). callers pass an :class:`asyncpg.Pool`
+    directly in production; tests pass a duck-typed mock.
+
+    typed as a :class:`Protocol` so the package does not take a
+    runtime dependency on asyncpg.
+    """
+
+    async def fetchrow(self, query: str, *args: object) -> Any:
+        """run query and return the first row (or None if empty).
+
+        :param query: parameterized SQL with ``$N`` placeholders
+        :ptype query: str
+        :param args: positional parameter values
+        :ptype args: object
+        :return: row record or None
+        :rtype: Any
+        """
+        ...
+
+    async def fetchval(self, query: str, *args: object) -> Any:
+        """run query and return the first column of the first row.
+
+        :param query: parameterized SQL with ``$N`` placeholders
+        :ptype query: str
+        :param args: positional parameter values
+        :ptype args: object
+        :return: scalar value or None
+        :rtype: Any
+        """
+        ...
+
+
+_BUMP_SQL = (
+    "INSERT INTO config_epochs (subject_path, epoch, payload) "
+    "VALUES ($1, 1, $2::jsonb) "
+    "ON CONFLICT (subject_path) DO UPDATE SET "
+    "epoch = config_epochs.epoch + 1, "
+    "payload = EXCLUDED.payload, "
+    "date_updated = now() "
+    "RETURNING epoch"
+)
+
+_CURRENT_SQL = "SELECT epoch FROM config_epochs WHERE subject_path = $1"
+
+
+class EpochClient:
+    """publish-side client for cross-pod config-epoch coherence.
+
+    one instance per process; safe to call from multiple admin
+    handlers concurrently (the bump statement serializes on the row
+    lock). the client never caches the last-seen epoch -- it always
+    round-trips Postgres on :meth:`bump` because the
+    ``RETURNING`` value is the only guaranteed-monotonic answer
+    available to a single writer in a multi-writer system.
+
+    :param pool: asyncpg-compatible pool exposing ``fetchrow`` /
+        ``fetchval``; production passes :class:`asyncpg.Pool`
+    :ptype pool: PoolLike
+    :param nats_client: connected typed NATS wrapper for broadcast
+    :ptype nats_client: NatsClient
+    """
+
+    def __init__(self, pool: PoolLike, nats_client: NatsClient) -> None:
+        """capture pool + nats client; no I/O.
+
+        :param pool: postgres pool implementing :class:`PoolLike`
+        :ptype pool: PoolLike
+        :param nats_client: connected NatsClient
+        :ptype nats_client: NatsClient
+        :return: nothing
+        :rtype: None
+        """
+        self._pool = pool
+        self._nats = nats_client
+
+    @traced
+    async def current(self, subject: Subject) -> int:
+        """read the latest epoch recorded for a subject.
+
+        used by :class:`~threetears.epoch.listener.EpochListener` on
+        cold start to prime its last-seen, and by periodic catch-up
+        ticks. returns ``0`` when no row exists yet -- the bump-side
+        ``ON CONFLICT`` clause guarantees the first successful
+        :meth:`bump` returns ``1``, so a returned ``0`` here means
+        "nobody has bumped this domain in this database."
+
+        :param subject: target subject; the subject's ``path`` is
+            the row PK
+        :ptype subject: Subject
+        :return: latest epoch, or ``0`` if no row exists
+        :rtype: int
+        """
+        value = await self._pool.fetchval(_CURRENT_SQL, subject.path)
+        if value is None:
+            result = 0
+        else:
+            result = int(value)
+        return result
+
+    @traced
+    async def bump(
+        self,
+        subject: Subject,
+        payload: dict[str, Any] | None = None,
+    ) -> int:
+        """atomically increment the epoch for a subject, then broadcast.
+
+        the upsert is serialized on the row lock; concurrent bumps
+        from different writers wait briefly and each receive a
+        unique strictly-increasing epoch. broadcast is best-effort:
+        :class:`~threetears.nats.errors.PublishError` is logged and
+        swallowed because the row commit is the source of truth and
+        any subscriber that missed the broadcast catches up via
+        :meth:`current` on the next periodic tick or via a per-
+        message epoch echo.
+
+        callers MUST invoke after the row mutation that motivates
+        the bump has committed. bumping inside an open transaction
+        broadcasts a phantom epoch if the transaction rolls back.
+
+        :param subject: target subject; the subject's ``path`` is
+            the row PK and the broadcast subject
+        :ptype subject: Subject
+        :param payload: opaque hint forwarded to subscribers in the
+            broadcast envelope; framework never inspects
+        :ptype payload: dict[str, Any] | None
+        :return: the new epoch returned by ``RETURNING``
+        :rtype: int
+        :raises RuntimeError: if the upsert returns no row (should
+            never happen on a healthy database -- the ``RETURNING``
+            clause is unconditional)
+        """
+        # asyncpg does not auto-encode dict to jsonb without a per-pool
+        # type codec; serialize at the call site so callers do not have
+        # to register codecs to use this client. the ``$2::jsonb`` cast
+        # in the SQL parses the resulting text back to jsonb.
+        payload_json = json.dumps(payload) if payload is not None else None
+        row = await self._pool.fetchrow(_BUMP_SQL, subject.path, payload_json)
+        if row is None:
+            raise RuntimeError(
+                f"config_epochs upsert returned no row for subject={subject.path!r}",
+            )
+        new_epoch = int(row["epoch"])
+
+        message = EpochBumpMessage(
+            subject_path=subject.path,
+            epoch=new_epoch,
+            payload=payload,
+        )
+        try:
+            await self._nats.publish(subject=subject, message=message)
+        except PublishError as exc:
+            log.warning(
+                "epoch bump broadcast failed; row commit is durable, "
+                "subscribers will catch up via current() or per-message echo",
+                extra={
+                    "extra_data": {
+                        "subject": subject.path,
+                        "epoch": new_epoch,
+                        "error": str(exc),
+                    },
+                },
+            )
+
+        return new_epoch