HERMES-GIELIE/tests/test_docker_docs_and_readonly.py

315 lines
15 KiB
Python
Raw Normal View History

docs(docker): document agent-image upgrade flow + read-only WebUI source mount The hermes-agent-src named volume in the two- and three-container compose files is initialised from the agent image's /opt/hermes on first `up` and Docker reuses it verbatim on every subsequent `up` — even after a fresh `docker pull` of the agent image. This was the root cause of #1416 (the 'missing entrypoint' symptom was a stale cached volume hiding the new image's source tree). Changes: - Add an 'Upgrading the agent container' section to docs/docker.md with the canonical `down → docker volume rm → pull → up -d` recipe, plus the same pointer as a comment block in both multi-container compose files near the volume declarations. - Switch the WebUI's hermes-agent-src mount to `:ro` in both multi-container compose files. The WebUI only reads this volume to install the agent's Python deps at startup; mounting it read-only enforces that at the kernel layer and brings the actual mount mode in line with the existing docs/docker.md architecture diagram (which already labelled this edge as read-only). - Align the workspace bind default in both multi-container compose files with the single-container convention — `${HERMES_WORKSPACE:-${HOME}/workspace}` instead of `${HERMES_WORKSPACE:-~/workspace}` — so the default resolves the same way across Linux, macOS, WSL2, and Docker Desktop on Windows. - Add a 'What the multi-container setup isolates (and what it doesn't)' section to docs/docker.md to frame the two/three-container setups as process/network/resource isolation, not filesystem isolation, so users don't reach for multi-container expecting a trust boundary it doesn't provide. - Cross-link #1416 from the Related issues section. Adds 9 regression tests in tests/test_docker_docs_and_readonly.py covering: - :ro on the WebUI side of hermes-agent-src in both files - agent side stays read-write (still needs to populate /opt/hermes on first run) - ${HOME} (not ~) in workspace bind defaults in both files - single-container file already uses ${HOME} (pin to prevent drift) - docs/docker.md has the 'Upgrading the agent container' section + recipe - compose files reference docs/docker.md + show the upgrade step inline - docs/docker.md frames the isolation model honestly Test suite: 42 passed (33 existing Docker tests + 9 new). No behaviour change for users who set HERMES_WORKSPACE explicitly, and no migration is required for existing deployments — Docker rebinds the existing volume read-only on next `up`. Users upgrading the agent image should now follow the documented `docker volume rm hermes-agent-src` recipe. Closes #1416 (documented upgrade procedure) and addresses the read-only half of the multi-container coupling concern raised on #2453.
2026-05-17 17:18:39 +00:00
"""Regression tests for the Docker docs+readonly hygiene PR (post v0.51.83).
Pins three invariants:
1. The `hermes-agent-src` named volume is mounted READ-ONLY on the WebUI
service in both multi-container compose files. The WebUI only reads it to
install agent Python deps at startup; this is defence-in-depth against a
compromised WebUI writing into the agent's source tree (Concern raised by
RustyLopez on #2453 and #1416).
2. The workspace bind-mount default uses `${HOME}/workspace` (not `~/workspace`)
in both multi-container compose files, matching the single-container
convention so `~`/`${HOME}` doesn't disagree across Linux, macOS, WSL2, and
Docker Desktop on Windows.
3. `docs/docker.md` documents the agent-image upgrade procedure (`docker volume
rm hermes-agent-src`) — the root cause of #1416.
"""
from __future__ import annotations
import re
docs(docker): document agent-image upgrade flow + read-only WebUI source mount The hermes-agent-src named volume in the two- and three-container compose files is initialised from the agent image's /opt/hermes on first `up` and Docker reuses it verbatim on every subsequent `up` — even after a fresh `docker pull` of the agent image. This was the root cause of #1416 (the 'missing entrypoint' symptom was a stale cached volume hiding the new image's source tree). Changes: - Add an 'Upgrading the agent container' section to docs/docker.md with the canonical `down → docker volume rm → pull → up -d` recipe, plus the same pointer as a comment block in both multi-container compose files near the volume declarations. - Switch the WebUI's hermes-agent-src mount to `:ro` in both multi-container compose files. The WebUI only reads this volume to install the agent's Python deps at startup; mounting it read-only enforces that at the kernel layer and brings the actual mount mode in line with the existing docs/docker.md architecture diagram (which already labelled this edge as read-only). - Align the workspace bind default in both multi-container compose files with the single-container convention — `${HERMES_WORKSPACE:-${HOME}/workspace}` instead of `${HERMES_WORKSPACE:-~/workspace}` — so the default resolves the same way across Linux, macOS, WSL2, and Docker Desktop on Windows. - Add a 'What the multi-container setup isolates (and what it doesn't)' section to docs/docker.md to frame the two/three-container setups as process/network/resource isolation, not filesystem isolation, so users don't reach for multi-container expecting a trust boundary it doesn't provide. - Cross-link #1416 from the Related issues section. Adds 9 regression tests in tests/test_docker_docs_and_readonly.py covering: - :ro on the WebUI side of hermes-agent-src in both files - agent side stays read-write (still needs to populate /opt/hermes on first run) - ${HOME} (not ~) in workspace bind defaults in both files - single-container file already uses ${HOME} (pin to prevent drift) - docs/docker.md has the 'Upgrading the agent container' section + recipe - compose files reference docs/docker.md + show the upgrade step inline - docs/docker.md frames the isolation model honestly Test suite: 42 passed (33 existing Docker tests + 9 new). No behaviour change for users who set HERMES_WORKSPACE explicitly, and no migration is required for existing deployments — Docker rebinds the existing volume read-only on next `up`. Users upgrading the agent image should now follow the documented `docker volume rm hermes-agent-src` recipe. Closes #1416 (documented upgrade procedure) and addresses the read-only half of the multi-container coupling concern raised on #2453.
2026-05-17 17:18:39 +00:00
from pathlib import Path
REPO = Path(__file__).resolve().parents[1]
# ── 1: hermes-agent-src must be read-only on the WebUI mount ────────────────
def test_two_container_webui_mounts_agent_src_readonly():
"""The WebUI only reads the agent source to install Python deps. Mounting
read-only enforces that at the kernel layer — a compromised WebUI process
cannot rewrite the agent source it then imports."""
src = (REPO / "docker-compose.two-container.yml").read_text(encoding="utf-8")
assert (
"hermes-agent-src:/home/hermeswebui/.hermes/hermes-agent:ro" in src
), (
"two-container: the WebUI must mount hermes-agent-src with :ro. "
"Without :ro, a compromised WebUI process can rewrite the agent's "
"Python source tree."
)
def test_three_container_webui_mounts_agent_src_readonly():
src = (REPO / "docker-compose.three-container.yml").read_text(encoding="utf-8")
assert (
"hermes-agent-src:/home/hermeswebui/.hermes/hermes-agent:ro" in src
), (
"three-container: the WebUI must mount hermes-agent-src with :ro."
)
def test_agent_service_keeps_writable_agent_src_mount():
"""The agent SERVICE writes the source tree to the volume on first up.
It must stay read-write — only the WebUI side is read-only."""
for fn in ("docker-compose.two-container.yml", "docker-compose.three-container.yml"):
src = (REPO / fn).read_text(encoding="utf-8")
# The agent's mount is `hermes-agent-src:/opt/hermes` (no :ro suffix).
# Look for the line that has /opt/hermes without :ro.
agent_lines = [
line for line in src.splitlines()
if "hermes-agent-src:/opt/hermes" in line
]
assert agent_lines, f"{fn}: agent must mount hermes-agent-src at /opt/hermes"
for line in agent_lines:
assert not line.rstrip().endswith(":ro"), (
f"{fn}: agent's hermes-agent-src mount must be writable "
f"(it populates /opt/hermes on first run): {line!r}"
)
# ── 2: ${HOME} (not ~) in workspace bind defaults ───────────────────────────
def test_two_container_workspace_uses_home_env_var():
"""Compose v2 expands `~` differently than `${HOME}` under sudo, on Docker
Desktop on Windows, and on some NAS appliances. Use `${HOME}` to match the
single-container `docker-compose.yml` and avoid platform drift."""
src = (REPO / "docker-compose.two-container.yml").read_text(encoding="utf-8")
assert "${HERMES_WORKSPACE:-${HOME}/workspace}:/workspace" in src, (
"two-container: workspace default must use ${HOME}/workspace, not ~/workspace, "
"to match docker-compose.yml's single-container convention."
)
assert "${HERMES_WORKSPACE:-~/workspace}" not in src, (
"two-container: tilde-form workspace default still present — change to ${HOME}/workspace."
)
def test_three_container_workspace_uses_home_env_var():
src = (REPO / "docker-compose.three-container.yml").read_text(encoding="utf-8")
assert "${HERMES_WORKSPACE:-${HOME}/workspace}:/workspace" in src, (
"three-container: workspace default must use ${HOME}/workspace, not ~/workspace."
)
assert "${HERMES_WORKSPACE:-~/workspace}" not in src
def test_single_container_workspace_already_uses_home_env_var():
"""Sanity: the single-container file has used ${HOME} all along; pin it
so it doesn't drift back."""
src = (REPO / "docker-compose.yml").read_text(encoding="utf-8")
assert "${HERMES_WORKSPACE:-${HOME}/workspace}:/workspace" in src
# ── 3: docs/docker.md documents the agent-image upgrade procedure ──────────
def test_docker_md_documents_agent_image_upgrade():
"""The `hermes-agent-src` named volume caches the agent source on first
`up` and is reused verbatim on every subsequent `up`, even after a fresh
`docker pull` of the agent image. This is the root cause of #1416. The
docs must give users the explicit `docker volume rm` recipe so they don't
misdiagnose 'missing entrypoint' errors."""
docs = (REPO / "docs" / "docker.md").read_text(encoding="utf-8")
assert "Upgrading the agent container" in docs, (
"docs/docker.md must have an 'Upgrading the agent container' section."
)
assert "docker volume rm" in docs, (
"docs/docker.md must show the `docker volume rm` step in the upgrade recipe."
)
assert "hermes-agent-src" in docs
# Cross-reference to the original issue so users searching for the
# symptom land in the right place
assert "#1416" in docs
def test_compose_files_point_to_docker_md_for_upgrades():
"""Both multi-container compose files should reference docs/docker.md
near the named-volumes block so anyone reading the compose file directly
finds the upgrade procedure."""
for fn in ("docker-compose.two-container.yml", "docker-compose.three-container.yml"):
src = (REPO / fn).read_text(encoding="utf-8")
assert "docs/docker.md" in src, (
f"{fn}: must reference docs/docker.md so users reading the compose "
f"file see the agent upgrade pointer."
)
assert "docker volume rm" in src, (
f"{fn}: must show the `docker volume rm` upgrade step inline."
)
# ── 4: docs/docker.md frames the isolation model honestly ──────────────────
def test_docker_md_documents_isolation_model():
"""The multi-container setups give process + network + resource isolation
but NOT filesystem isolation. Document that explicitly so users don't
reach for multi-container expecting a trust boundary it doesn't provide
(RustyLopez's concern on #2453)."""
docs = (REPO / "docs" / "docker.md").read_text(encoding="utf-8")
assert "What the multi-container setup isolates" in docs, (
"docs/docker.md must have a section calibrating multi-container "
"isolation expectations — process/network/resource isolation, NOT "
"filesystem isolation."
)
fix(docker): stage agent source to writable build dir before pip install The Docker smoke gate added in this same PR caught a real production regression on its very first CI run. v0.51.84 (PR #2470) mounted hermes-agent-src read-only on the WebUI side and widened the chown prune to keep the read-only walk happy, but missed that the WebUI's startup also runs: uv pip install "$_agent_src[all]" against the same now-read-only mount. setuptools' egg_info step writes hermes_agent.egg-info/ inside the source tree even under PEP 517 build isolation (this is by design -- PEP 517 isolates the BUILD environment, not the source tree's metadata directory). On a :ro mount this returns EROFS, the install fails, error_exit fires, and every multi-container deploy dies at startup. The smoke gate flagged it on both the two-container and three-container variants. The fix ------- Stage the agent source into a writable build dir under /tmp BEFORE invoking pip install, then point pip at the staged copy. _stage_src="/tmp/hermes-agent-build" rm -rf "$_stage_src" && mkdir -p "$_stage_src" rsync -a --exclude='*.egg-info' --exclude='build' --exclude='dist' \ --exclude='__pycache__' --exclude='.git' \ "$_agent_src"/ "$_stage_src"/ uv pip install "$_stage_src[all]" ... rm -rf "$_stage_src" The exclusion list matters: when setuptools sees a pre-baked *.egg-info, build, or dist directory, it takes a timestamp-update code path that also reads/writes inside that directory -- which itself fails on a :ro source. Excluding them keeps the build on the fresh-build path unconditionally. rsync is in the production image (Dockerfile line 41-44). For users running custom WebUI images without rsync, the script falls back to cp -a + post-copy rm -rf of the same artifacts. Tests ----- Two new source-level invariants in tests/test_docker_docs_and_readonly.py: test_docker_init_stages_agent_source_for_writable_install -- asserts _stage_src=... is declared -- asserts every `uv pip install ...[all]` line uses _stage_src, NOT raw $_agent_src test_docker_init_excludes_egg_info_during_staging -- asserts the staging path excludes *.egg-info (rsync exclude form or cp-fallback's explicit rm -rf both pass) These would have caught the v0.51.84 regression at the source level (once written; they're new). The Docker runtime smoke gate is the durable defence for the broader class of :ro x init-script interactions, since source-level invariants only catch what they're written to catch. Verification ------------ - pytest tests/test_docker_docs_and_readonly.py: 11 passed (9 existing + 2 new) - pytest tests/ -q --timeout=60: 5891 passed, 6 skipped (was 5889; delta is exactly the 2 new tests) - bash -n docker_init.bash: clean Once this lands, the Docker smoke gate's two/three-container variants should go green, completing the self-validating loop.
2026-05-18 00:21:31 +00:00
# ── 5: docker_init.bash stages agent source to a writable build dir ─────────
#
# The :ro mount fixed in PR #2470 broke a second, less obvious surface:
# `uv pip install "$_agent_src[all]"` invokes setuptools' egg_info build step,
# which touches `hermes_agent.egg-info/` *inside the source tree* even under
# PEP 517 build isolation. On a `:ro` mount this returns `EROFS` and (under
# `set -e`) kills container startup. The fix: copy the source tree into a
# writable tmpfs build dir, run the install against THAT, then clean up.
#
# This was caught the first time the Docker smoke gate ran on its own PR — a
# real regression that 5800+ source-level pytests had no way to surface
# because none of them invoked `docker_init.bash` against a real :ro mount.
def test_docker_init_stages_agent_source_for_writable_install():
"""docker_init.bash must NOT pass the raw _agent_src path to `uv pip
install` — that hits the :ro mount and fails. It must stage the source
into a writable build dir first (the staged path is used in the install
invocation)."""
src = (REPO / "docker_init.bash").read_text(encoding="utf-8")
# The fix uses a /tmp staging path that's clearly distinct from the
# mounted source path. Pin the staging marker.
assert "_stage_src=" in src, (
"docker_init.bash must declare a _stage_src writable build dir "
"before invoking `uv pip install` against the (potentially :ro) "
"hermes-agent source."
)
# The install line must reference the staged path, NOT the raw _agent_src
# path. The pre-fix code was:
# uv pip install "$_agent_src[all]" ...
# The fixed code is:
# uv pip install "$_stage_src[all]" ...
install_lines = [
line for line in src.splitlines()
if "uv pip install" in line and "[all]" in line
]
assert install_lines, "expected an `uv pip install ...[all]` line in docker_init.bash"
for line in install_lines:
assert '"$_agent_src[all]"' not in line, (
"docker_init.bash invokes `uv pip install $_agent_src[all]` "
"directly — this fails with EROFS when the hermes-agent volume "
"is mounted :ro (the production multi-container default). "
"Use the writable $_stage_src path instead. "
f"Offending line: {line!r}"
)
assert "_stage_src" in line, (
"the `uv pip install ...[all]` line must use the staged writable "
f"path. Offending line: {line!r}"
)
def test_docker_init_excludes_egg_info_during_staging():
"""The staging copy must exclude pre-baked *.egg-info / build / dist
directories. setuptools takes a different (timestamp-update) code path
when one is already present in the source tree, which itself hits the
:ro mount through stat/utime calls. Excluding them keeps the build
harden(docker-smoke): catch !!ERROR/!!Exiting + tighten egg_info test Two non-blocking observations from the review, both addressed: 1. The bad-pattern grep listed `error_exit` as a literal token, but the `error_exit()` function at docker_init.bash:5-10 only echoes the strings `"!! ERROR: "` and `"!! Exiting script (ID: $$)"` — the function name itself never appears in container logs. So `grep -E -i "error_exit"` would only fire on stray debug prints of the name, not on actual failures. The other patterns (`Failed to set (UID|GID|...)`, `groupmod: cannot`, etc.) DO catch real error_exit output, so this wasn't a coverage gap — just a dead token. Add `!! ERROR` and `!! Exiting script` to the bad-pattern set so the grep actually matches the function's output. Keep the literal `error_exit` token as belt-and-suspenders for any debug/echo of the name. 2. `test_docker_init_excludes_egg_info_during_staging` was a single `assert "egg-info" in src` check. That passes if any occurrence appears — including the explanatory comment block above the staging logic. A maintainer removing the `--exclude='*.egg-info'` from rsync but keeping the comment would slip past the test. Tighten to: - scope to the staging block (between `_stage_src=` and the `uv pip install` line) so comments outside that window can't satisfy the assertion; - require the literal `--exclude='*.egg-info'` rsync flag; - require `*.egg-info` in the block so the cp-fallback cleanup is also pinned; - additionally require `--exclude='build'`, `--exclude='dist'`, `--exclude='__pycache__'` so all four setuptools-touchable artifact dirs stay excluded. Verified: - tests/test_docker_docs_and_readonly.py — 11/11 pass. - YAML parses cleanly via `yaml.safe_load`. - Full suite: 5770 passed, 0 failed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-18 00:34:46 +00:00
happily on the fresh-build code path.
Tight assertions on both the rsync and cp-fallback paths — a loose
`"egg-info" in src` check would pass on a stray comment mention, so
we require the actual exclusion mechanics to be present.
"""
fix(docker): stage agent source to writable build dir before pip install The Docker smoke gate added in this same PR caught a real production regression on its very first CI run. v0.51.84 (PR #2470) mounted hermes-agent-src read-only on the WebUI side and widened the chown prune to keep the read-only walk happy, but missed that the WebUI's startup also runs: uv pip install "$_agent_src[all]" against the same now-read-only mount. setuptools' egg_info step writes hermes_agent.egg-info/ inside the source tree even under PEP 517 build isolation (this is by design -- PEP 517 isolates the BUILD environment, not the source tree's metadata directory). On a :ro mount this returns EROFS, the install fails, error_exit fires, and every multi-container deploy dies at startup. The smoke gate flagged it on both the two-container and three-container variants. The fix ------- Stage the agent source into a writable build dir under /tmp BEFORE invoking pip install, then point pip at the staged copy. _stage_src="/tmp/hermes-agent-build" rm -rf "$_stage_src" && mkdir -p "$_stage_src" rsync -a --exclude='*.egg-info' --exclude='build' --exclude='dist' \ --exclude='__pycache__' --exclude='.git' \ "$_agent_src"/ "$_stage_src"/ uv pip install "$_stage_src[all]" ... rm -rf "$_stage_src" The exclusion list matters: when setuptools sees a pre-baked *.egg-info, build, or dist directory, it takes a timestamp-update code path that also reads/writes inside that directory -- which itself fails on a :ro source. Excluding them keeps the build on the fresh-build path unconditionally. rsync is in the production image (Dockerfile line 41-44). For users running custom WebUI images without rsync, the script falls back to cp -a + post-copy rm -rf of the same artifacts. Tests ----- Two new source-level invariants in tests/test_docker_docs_and_readonly.py: test_docker_init_stages_agent_source_for_writable_install -- asserts _stage_src=... is declared -- asserts every `uv pip install ...[all]` line uses _stage_src, NOT raw $_agent_src test_docker_init_excludes_egg_info_during_staging -- asserts the staging path excludes *.egg-info (rsync exclude form or cp-fallback's explicit rm -rf both pass) These would have caught the v0.51.84 regression at the source level (once written; they're new). The Docker runtime smoke gate is the durable defence for the broader class of :ro x init-script interactions, since source-level invariants only catch what they're written to catch. Verification ------------ - pytest tests/test_docker_docs_and_readonly.py: 11 passed (9 existing + 2 new) - pytest tests/ -q --timeout=60: 5891 passed, 6 skipped (was 5889; delta is exactly the 2 new tests) - bash -n docker_init.bash: clean Once this lands, the Docker smoke gate's two/three-container variants should go green, completing the self-validating loop.
2026-05-18 00:21:31 +00:00
src = (REPO / "docker_init.bash").read_text(encoding="utf-8")
harden(docker-smoke): catch !!ERROR/!!Exiting + tighten egg_info test Two non-blocking observations from the review, both addressed: 1. The bad-pattern grep listed `error_exit` as a literal token, but the `error_exit()` function at docker_init.bash:5-10 only echoes the strings `"!! ERROR: "` and `"!! Exiting script (ID: $$)"` — the function name itself never appears in container logs. So `grep -E -i "error_exit"` would only fire on stray debug prints of the name, not on actual failures. The other patterns (`Failed to set (UID|GID|...)`, `groupmod: cannot`, etc.) DO catch real error_exit output, so this wasn't a coverage gap — just a dead token. Add `!! ERROR` and `!! Exiting script` to the bad-pattern set so the grep actually matches the function's output. Keep the literal `error_exit` token as belt-and-suspenders for any debug/echo of the name. 2. `test_docker_init_excludes_egg_info_during_staging` was a single `assert "egg-info" in src` check. That passes if any occurrence appears — including the explanatory comment block above the staging logic. A maintainer removing the `--exclude='*.egg-info'` from rsync but keeping the comment would slip past the test. Tighten to: - scope to the staging block (between `_stage_src=` and the `uv pip install` line) so comments outside that window can't satisfy the assertion; - require the literal `--exclude='*.egg-info'` rsync flag; - require `*.egg-info` in the block so the cp-fallback cleanup is also pinned; - additionally require `--exclude='build'`, `--exclude='dist'`, `--exclude='__pycache__'` so all four setuptools-touchable artifact dirs stay excluded. Verified: - tests/test_docker_docs_and_readonly.py — 11/11 pass. - YAML parses cleanly via `yaml.safe_load`. - Full suite: 5770 passed, 0 failed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-18 00:34:46 +00:00
# Find the staging block: rsync invocation OR cp-fallback. Both must
# actually exclude *.egg-info — a comment mention is not enough.
stage_idx = src.index("_stage_src=")
install_idx = src.index("uv pip install", stage_idx)
stage_block = src[stage_idx:install_idx]
# Rsync path must carry --exclude='*.egg-info'.
assert "--exclude='*.egg-info'" in stage_block, (
"docker_init.bash rsync invocation must include "
"--exclude='*.egg-info' so setuptools' timestamp-update code path "
"doesn't fire (which itself hits the :ro mount through stat/utime)."
)
# cp-fallback path must explicitly rm the egg-info dir after copy
# (cp -a has no --exclude flag, so the cleanup happens post-copy).
assert "*.egg-info" in stage_block, (
"docker_init.bash cp-fallback must remove $_stage_src/*.egg-info "
"after copy so the install runs on the fresh-build code path."
)
# Both build and dist must also be excluded — setuptools touches them
# under different conditions but the failure mode is identical.
assert "--exclude='build'" in stage_block, (
"rsync staging must --exclude='build' (setuptools build artifacts)."
)
assert "--exclude='dist'" in stage_block, (
"rsync staging must --exclude='dist' (setuptools build artifacts)."
)
assert "--exclude='__pycache__'" in stage_block, (
"rsync staging must --exclude='__pycache__' to keep the copy minimal."
fix(docker): stage agent source to writable build dir before pip install The Docker smoke gate added in this same PR caught a real production regression on its very first CI run. v0.51.84 (PR #2470) mounted hermes-agent-src read-only on the WebUI side and widened the chown prune to keep the read-only walk happy, but missed that the WebUI's startup also runs: uv pip install "$_agent_src[all]" against the same now-read-only mount. setuptools' egg_info step writes hermes_agent.egg-info/ inside the source tree even under PEP 517 build isolation (this is by design -- PEP 517 isolates the BUILD environment, not the source tree's metadata directory). On a :ro mount this returns EROFS, the install fails, error_exit fires, and every multi-container deploy dies at startup. The smoke gate flagged it on both the two-container and three-container variants. The fix ------- Stage the agent source into a writable build dir under /tmp BEFORE invoking pip install, then point pip at the staged copy. _stage_src="/tmp/hermes-agent-build" rm -rf "$_stage_src" && mkdir -p "$_stage_src" rsync -a --exclude='*.egg-info' --exclude='build' --exclude='dist' \ --exclude='__pycache__' --exclude='.git' \ "$_agent_src"/ "$_stage_src"/ uv pip install "$_stage_src[all]" ... rm -rf "$_stage_src" The exclusion list matters: when setuptools sees a pre-baked *.egg-info, build, or dist directory, it takes a timestamp-update code path that also reads/writes inside that directory -- which itself fails on a :ro source. Excluding them keeps the build on the fresh-build path unconditionally. rsync is in the production image (Dockerfile line 41-44). For users running custom WebUI images without rsync, the script falls back to cp -a + post-copy rm -rf of the same artifacts. Tests ----- Two new source-level invariants in tests/test_docker_docs_and_readonly.py: test_docker_init_stages_agent_source_for_writable_install -- asserts _stage_src=... is declared -- asserts every `uv pip install ...[all]` line uses _stage_src, NOT raw $_agent_src test_docker_init_excludes_egg_info_during_staging -- asserts the staging path excludes *.egg-info (rsync exclude form or cp-fallback's explicit rm -rf both pass) These would have caught the v0.51.84 regression at the source level (once written; they're new). The Docker runtime smoke gate is the durable defence for the broader class of :ro x init-script interactions, since source-level invariants only catch what they're written to catch. Verification ------------ - pytest tests/test_docker_docs_and_readonly.py: 11 passed (9 existing + 2 new) - pytest tests/ -q --timeout=60: 5891 passed, 6 skipped (was 5889; delta is exactly the 2 new tests) - bash -n docker_init.bash: clean Once this lands, the Docker smoke gate's two/three-container variants should go green, completing the self-validating loop.
2026-05-18 00:21:31 +00:00
)
assert "--exclude='.playwright'" in stage_block, (
"rsync staging must --exclude='.playwright' so unreadable Playwright "
"browser dependency files (e.g. deb.deps, rpm.deps) don't cause "
"rsync error code 23 and kill container startup."
)
assert ".playwright" in stage_block, (
"Both rsync (--exclude) and cp-fallback (rm -rf) must handle "
".playwright so the build doesn't choke on unreadable browser files."
)
def test_docker_init_makes_staged_dir_writable_after_ro_mount_copy():
"""Regression test for the docker-init "could not create hermes_agent.egg-info:
Permission denied" failure on :ro multi-container mounts.
`rsync -a` / `cp -a` preserve the source tree's mode bits, so a hermes-agent
source mounted mode 555 leaves the staged copy also mode 555 even though the
staging dir itself was created writable by hermeswebui. setuptools then can't
create `<pkg>.egg-info/` next to the package and dies with "Permission denied"
during `uv pip install`. The fix is a `chmod -R u+w` on the staged tree AFTER
both copy paths and BEFORE the install. This test pins that ordering and the
`u+w` form (so the staged tree isn't accidentally made world-writable).
"""
src = (REPO / "docker_init.bash").read_text(encoding="utf-8")
stage_idx = src.index("_stage_src=")
install_idx = src.index("uv pip install", stage_idx)
stage_block = src[stage_idx:install_idx]
# The fix MUST add owner-write to the staged tree after the rsync/cp copy.
# A naked `chmod` call that strips 0555 (or equivalent) would also work,
# but `u+w` is the minimal-permission form and matches the rest of the
# docker-init.bash style (never widen perms beyond what the operator needs).
assert re.search(r"chmod\s+-R\s+u\+w\s+\"?\\?\$?_stage_src\"?", stage_block), (
"docker_init.bash staging block must `chmod -R u+w \"$_stage_src\"` "
"after the rsync/cp copy. Without it, a :ro hermes-agent mount leaves "
"the staged tree mode 555 and `uv pip install` fails with "
"'could not create hermes_agent.egg-info: Permission denied' under "
"PEP 517 build isolation."
)
# The chmod must live in the SHARED tail (after `fi`), not inside either
# branch — otherwise dropping back to cp -a (when rsync is missing) would
# silently re-introduce the bug.
assert "chmod -R u+w \"$_stage_src\"" in stage_block, (
"the post-staging chmod must use the exact `chmod -R u+w \"$_stage_src\"` "
"form so the substring-check pattern below matches both code paths."
)
# Both copy branches (rsync and cp -a) close with `fi`; the chmod must
# come AFTER the closing `fi` so it covers whichever path was taken.
fi_idx = stage_block.rindex("\n fi\n")
chmod_idx = stage_block.index("chmod -R u+w")
assert chmod_idx > fi_idx, (
"chmod must live AFTER the rsync/cp if/else closes — putting it "
"inside one branch means the other copy path skips it and the "
":ro mount perm leak returns."
)