feat(gitops): wire durable execution side effects

This commit is contained in:
chatgpt-final-release-writer
2026-08-08 15:15:40 +08:00
parent 3fbbc18fb2
commit baa3334324
8 changed files with 803 additions and 6 deletions
+149
View File
@@ -0,0 +1,149 @@
"""Fail-closed Authority Git side-effect boundary.
Only this module may issue the narrowly scoped ref CAS primitives used by the
durable executor. It never accepts force updates, arbitrary refs, tags, or
direct production mutations.
"""
from __future__ import annotations
from dataclasses import dataclass
import os
from pathlib import Path
import re
import subprocess
import tempfile
from typing import Mapping, Sequence
OID_RE = re.compile(r"^[0-9a-f]{40}$")
TASK_BRANCH_RE = re.compile(r"^refs/heads/agent/[A-Za-z0-9][A-Za-z0-9._/-]{2,179}$")
class AuthorityGitError(RuntimeError):
pass
def _oid(value: str, field: str) -> str:
if not isinstance(value, str) or not OID_RE.fullmatch(value):
raise AuthorityGitError(f"{field}_INVALID")
return value
@dataclass(frozen=True, slots=True)
class AuthorityGitExecutor:
git_dir: Path
worktree_root: Path = Path("/var/lib/aios-gitops/worktrees")
def __post_init__(self) -> None:
if not self.git_dir.is_absolute() or self.git_dir.is_symlink():
raise AuthorityGitError("AUTHORITY_GIT_DIR_INVALID")
if not self.worktree_root.is_absolute() or self.worktree_root.is_symlink():
raise AuthorityGitError("WORKTREE_ROOT_INVALID")
def _git(self, args: Sequence[str], *, check: bool = True) -> str:
result = subprocess.run(
["git", f"--git-dir={self.git_dir}", *args],
text=True,
capture_output=True,
check=False,
)
if check and result.returncode != 0:
raise AuthorityGitError("AUTHORITY_GIT_COMMAND_FAILED")
return result.stdout.strip()
def read_ref(self, ref: str) -> str | None:
if not ref.startswith("refs/") or ref.startswith(("refs/tags/", "refs/remotes/")):
raise AuthorityGitError("REF_READ_SCOPE_DENIED")
value = self._git(["rev-parse", "--verify", ref], check=False)
return value if value and OID_RE.fullmatch(value) else None
def ancestry_check(self, older_oid: str, newer_oid: str) -> bool:
_oid(older_oid, "older_oid")
_oid(newer_oid, "newer_oid")
result = subprocess.run(
["git", f"--git-dir={self.git_dir}", "merge-base", "--is-ancestor", older_oid, newer_oid],
capture_output=True,
check=False,
)
if result.returncode not in (0, 1):
raise AuthorityGitError("ANCESTRY_CHECK_FAILED")
return result.returncode == 0
def advance_task_branch_CAS(self, branch: str, expected_oid: str, new_oid: str) -> str:
if not TASK_BRANCH_RE.fullmatch(branch):
raise AuthorityGitError("TASK_BRANCH_SCOPE_DENIED")
_oid(expected_oid, "expected_oid")
_oid(new_oid, "new_oid")
current = self.read_ref(branch)
if current != expected_oid:
if current == new_oid:
return new_oid
raise AuthorityGitError("AUTHORITY_BRANCH_EXPECTATION_MISMATCH")
if not self.ancestry_check(expected_oid, new_oid):
raise AuthorityGitError("NEW_OID_NOT_DESCENDANT")
self._git(["update-ref", branch, new_oid, expected_oid])
readback = self.read_ref(branch)
if readback != new_oid:
raise AuthorityGitError("AUTHORITY_CAS_READBACK_MISMATCH")
return readback
def ff_CAS_main_after_verified_protected_merge(self, expected_oid: str, new_oid: str) -> str:
_oid(expected_oid, "expected_oid")
_oid(new_oid, "new_oid")
if not self.ancestry_check(expected_oid, new_oid):
raise AuthorityGitError("MAIN_FF_REQUIRED")
current = self.read_ref("refs/heads/main")
if current != expected_oid:
if current == new_oid:
return new_oid
raise AuthorityGitError("AUTHORITY_MAIN_EXPECTATION_MISMATCH")
self._git(["update-ref", "refs/heads/main", new_oid, expected_oid])
readback = self.read_ref("refs/heads/main")
if readback != new_oid:
raise AuthorityGitError("AUTHORITY_MAIN_CAS_READBACK_MISMATCH")
return readback
def commit_file_in_isolated_worktree(
self,
*,
base_oid: str,
files: Mapping[str, bytes],
message: str,
worktree_name: str,
) -> str:
_oid(base_oid, "base_oid")
if not worktree_name or "/" in worktree_name or "\\" in worktree_name or worktree_name in {".", ".."}:
raise AuthorityGitError("WORKTREE_NAME_INVALID")
root = self.worktree_root / worktree_name
if root.exists() or root.is_symlink():
raise AuthorityGitError("WORKTREE_ALREADY_EXISTS")
self.worktree_root.mkdir(parents=True, exist_ok=True)
self._git(["worktree", "add", "--detach", str(root), base_oid])
try:
for relative, content in files.items():
path = Path(relative)
if path.is_absolute() or ".." in path.parts or not path.parts or path.parts[0] == ".git":
raise AuthorityGitError("FILE_PATH_INVALID")
target = root / path
target.parent.mkdir(parents=True, exist_ok=True)
target.write_bytes(content)
result = subprocess.run(
["git", "-C", str(root), "add", "--", *files.keys()],
text=True,
capture_output=True,
check=False,
)
if result.returncode != 0:
raise AuthorityGitError("WORKTREE_ADD_FAILED")
result = subprocess.run(
["git", "-C", str(root), "-c", "user.name=aios-durable-job", "-c", "user.email=aios-durable-job@localhost", "commit", "--no-verify", "-m", message],
text=True,
capture_output=True,
check=False,
)
if result.returncode != 0:
raise AuthorityGitError("WORKTREE_COMMIT_FAILED")
return subprocess.check_output(["git", "-C", str(root), "rev-parse", "HEAD"], text=True).strip()
finally:
subprocess.run(["git", f"--git-dir={self.git_dir}", "worktree", "remove", "--force", str(root)], capture_output=True, check=False)
@@ -90,12 +90,15 @@ def load_recovered_authorization(path: str | Path) -> dict[str, Any]:
return {
"schema": receipt["schema"],
"state": receipt["recovery_state"],
"campaign_id": receipt["campaign_id"],
"original_event_id": receipt["original_event_id"],
"original_authorization_digest": receipt["original_authorization_digest"],
"signed_grant_id": receipt["signed_grant_id"],
"task_id": receipt["task_id"],
"repository": receipt["repository"],
"original_base_oid": receipt["original_base_oid"],
"branch": receipt["branch"],
"expected_branch_oid": receipt["expected_branch_oid"],
"digest": actual,
"new_owner_interaction": False,
}
@@ -0,0 +1,204 @@
"""Gitea HTTP/Git transport using only a projected systemd credential."""
from __future__ import annotations
from dataclasses import dataclass
import json
import os
from pathlib import Path
import subprocess
import tempfile
from typing import Any, Mapping
from urllib.error import HTTPError, URLError
from urllib.parse import quote, urlencode
from urllib.request import Request, urlopen
class GiteaTransportError(RuntimeError):
pass
def _safe_oid(value: str, field: str) -> str:
if not isinstance(value, str) or len(value) != 40 or any(ch not in "0123456789abcdef" for ch in value):
raise GiteaTransportError(f"{field}_INVALID")
return value
@dataclass(slots=True)
class GiteaMachineTransport:
api_base: str
repository_id: str
remote_url: str
authority_git_dir: Path
credential_name: str = "gitea-machine-token"
actor: str = "aios-autonomy-agent"
timeout: int = 20
def __post_init__(self) -> None:
if self.credential_name != "gitea-machine-token":
raise GiteaTransportError("CREDENTIAL_NAME_INVALID")
if "/" not in self.repository_id or self.repository_id.startswith("/"):
raise GiteaTransportError("REPOSITORY_INVALID")
if not self.api_base.startswith("https://") and not self.api_base.startswith("http://127.0.0.1"):
raise GiteaTransportError("GITEA_API_SCOPE_INVALID")
if not self.remote_url.startswith("https://"):
raise GiteaTransportError("GITEA_REMOTE_SCOPE_INVALID")
@property
def _credential_path(self) -> Path:
directory = os.environ.get("CREDENTIALS_DIRECTORY", "")
if not directory:
raise GiteaTransportError("CREDENTIAL_PROJECTION_MISSING")
path = Path(directory) / self.credential_name
if path.is_symlink() or not path.is_file():
raise GiteaTransportError("CREDENTIAL_PROJECTION_MISSING")
return path
def _token(self) -> str:
value = self._credential_path.read_text(encoding="utf-8").strip()
if not value:
raise GiteaTransportError("CREDENTIAL_EMPTY")
return value
def _request(self, method: str, path: str, body: Mapping[str, Any] | None = None) -> tuple[int, Any]:
data = None if body is None else json.dumps(dict(body), separators=(",", ":")).encode()
headers = {"Accept": "application/json", "Authorization": f"token {self._token()}"}
if data is not None:
headers["Content-Type"] = "application/json"
request = Request(self.api_base.rstrip("/") + path, data=data, headers=headers, method=method)
try:
with urlopen(request, timeout=self.timeout) as response:
raw = response.read()
return response.status, json.loads(raw.decode()) if raw else {}
except HTTPError as exc:
# Never include response bodies: Gitea may echo sensitive request context.
raise GiteaTransportError(f"GITEA_HTTP_{exc.code}") from exc
except (URLError, TimeoutError, OSError) as exc:
raise GiteaTransportError("GITEA_NETWORK_ERROR") from exc
def whoami(self) -> Mapping[str, Any]:
return self._request("GET", "/user")[1]
def repository_info(self) -> Mapping[str, Any]:
owner, name = self.repository_id.split("/", 1)
return self._request("GET", f"/repos/{quote(owner)}/{quote(name)}")[1]
def repository(self, owner: str, name: str) -> Mapping[str, Any]:
if f"{owner}/{name}" != self.repository_id:
raise GiteaTransportError("REPOSITORY_SCOPE_DENIED")
return self.repository_info()
def read_branch(self, branch: str) -> str | None:
owner, name = self.repository_id.split("/", 1)
try:
value = self._request("GET", f"/repos/{quote(owner)}/{quote(name)}/branches/{quote(branch, safe='/')}")[1]
except GiteaTransportError as exc:
if str(exc) == "GITEA_HTTP_404":
return None
raise
sha = value.get("commit", {}).get("id") if isinstance(value, dict) else None
return _safe_oid(sha, "remote_branch_oid")
def _askpass(self) -> tuple[Path, dict[str, str]]:
credential = self._credential_path
fd, name = tempfile.mkstemp(prefix="aios-gitea-askpass-", dir="/tmp")
path = Path(name)
os.fchmod(fd, 0o700)
script = "#!/bin/sh\ncase \"$1\" in *Username*) printf '%s\\n' \"$GITEA_MACHINE_ACTOR\" ;; *) cat \"$CREDENTIAL_FILE\" ;; esac\n"
with os.fdopen(fd, "w", encoding="utf-8") as stream:
stream.write(script)
stream.flush()
os.fsync(stream.fileno())
return path, {
"GIT_ASKPASS": str(path),
"GIT_TERMINAL_PROMPT": "0",
"GITEA_MACHINE_ACTOR": self.actor,
"CREDENTIAL_FILE": str(credential),
}
def _push_new_branch(self, branch: str, new_oid: str) -> None:
askpass, additions = self._askpass()
environment = os.environ.copy()
environment.update(additions)
try:
result = subprocess.run(
["git", f"--git-dir={self.authority_git_dir}", "push", self.remote_url, f"{new_oid}:refs/heads/{branch}"],
text=True,
capture_output=True,
check=False,
env=environment,
timeout=self.timeout,
)
if result.returncode != 0:
raise GiteaTransportError("GITEA_BRANCH_CREATE_FAILED")
finally:
askpass.unlink(missing_ok=True)
def compare_and_swap_branch(self, branch: str, expected_oid: str, new_oid: str) -> str:
_safe_oid(expected_oid, "expected_oid")
_safe_oid(new_oid, "new_oid")
current = self.read_branch(branch)
if current == new_oid:
return new_oid
if current is None:
self._push_new_branch(branch, new_oid)
readback = self.read_branch(branch)
if readback != new_oid:
raise GiteaTransportError("GITEA_BRANCH_CREATE_READBACK_MISMATCH")
return readback
if current != expected_oid:
raise GiteaTransportError("GITEA_BRANCH_EXPECTATION_MISMATCH")
owner, name = self.repository_id.split("/", 1)
status, value = self._request(
"PUT",
f"/repos/{quote(owner)}/{quote(name)}/branches/{quote(branch, safe='/')}",
{"new_commit_id": new_oid, "old_commit_id": expected_oid, "force": False},
)
if status not in (200, 201, 204):
raise GiteaTransportError("GITEA_BRANCH_CAS_FAILED")
readback = self.read_branch(branch)
if readback != new_oid:
raise GiteaTransportError("GITEA_BRANCH_CAS_READBACK_MISMATCH")
return readback
def _pulls(self) -> list[Mapping[str, Any]]:
owner, name = self.repository_id.split("/", 1)
query = urlencode({"state": "all", "base_branch": "main", "limit": 100})
value = self._request("GET", f"/repos/{quote(owner)}/{quote(name)}/pulls?{query}")[1]
return value if isinstance(value, list) else []
def create_or_update_pr(self, branch: str, target: str, title: str, body: str) -> Mapping[str, Any]:
if target != "main":
raise GiteaTransportError("PR_BASE_SCOPE_DENIED")
head = self.read_branch(branch)
if head is None:
raise GiteaTransportError("PR_HEAD_MISSING")
matches = [p for p in self._pulls() if (p.get("head") or {}).get("ref") == branch and (p.get("base") or {}).get("ref") == target]
open_matches = [p for p in matches if p.get("state") == "open"]
if len(open_matches) > 1:
raise GiteaTransportError("MULTIPLE_MATCHING_PRS")
if open_matches:
if (open_matches[0].get("head") or {}).get("sha") != head:
raise GiteaTransportError("PR_HEAD_CONFLICT")
return open_matches[0]
owner, name = self.repository_id.split("/", 1)
return self._request("POST", f"/repos/{quote(owner)}/{quote(name)}/pulls", {"head": branch, "base": target, "title": title, "body": body, "allow_maintainer_edit": False})[1]
def publish_status(self, branch: str, state: str, description: str) -> Mapping[str, Any]:
oid = self.read_branch(branch)
if oid is None:
raise GiteaTransportError("STATUS_BRANCH_MISSING")
owner, name = self.repository_id.split("/", 1)
return self._request("POST", f"/repos/{quote(owner)}/{quote(name)}/statuses/{oid}", {"state": state, "context": "aios/durable-release", "description": description})[1]
def protected_merge(self, pr_number: int) -> Mapping[str, Any]:
if not isinstance(pr_number, int) or pr_number < 1:
raise GiteaTransportError("PR_NUMBER_INVALID")
owner, name = self.repository_id.split("/", 1)
pr = self._request("GET", f"/repos/{quote(owner)}/{quote(name)}/pulls/{pr_number}")[1]
if pr.get("state") != "open" or (pr.get("base") or {}).get("ref") != "main":
raise GiteaTransportError("PR_MERGE_BINDING_INVALID")
if pr.get("mergeable") is False:
raise GiteaTransportError("PR_NOT_MERGEABLE")
body = {"do": "fast-forward-only", "force_merge": False, "merge_when_checks_succeed": False, "head_commit_id": (pr.get("head") or {}).get("sha")}
return self._request("POST", f"/repos/{quote(owner)}/{quote(name)}/pulls/{pr_number}/merge", body)[1]
@@ -275,7 +275,31 @@ class GitOpsJobController:
raise JobError("PAYLOAD_DIGEST_INVALID")
if request["branch"] in {"main", "master"} or request["branch"].startswith(("refs/", "tags/")):
raise JobError("PROTECTED_BRANCH_REJECTED")
return {field: str(request[field]) for field in fields}
binding = {field: str(request[field]) for field in fields}
# Operation-specific, non-secret binding inputs must survive the
# durable round trip and participate in idempotency. Arbitrary request
# fields are intentionally discarded rather than persisted.
optional = (
"task_id",
"new_branch_oid",
"authorization_recovery_digest",
"lineage_anchor_oid",
"generation",
"expected_main_oid",
"pr_number",
"target_branch",
)
for field in optional:
value = request.get(field)
if value is None:
continue
if isinstance(value, bool) or not isinstance(value, (str, int)):
raise JobError("JOB_BINDING_OPTION_INVALID")
text = str(value).strip()
if not text or len(text) > 512:
raise JobError("JOB_BINDING_OPTION_INVALID")
binding[field] = text
return binding
def submit(self, request: Mapping[str, Any], *, payload: bytes | None = None) -> JobReceipt:
binding = self._binding(request)
@@ -0,0 +1,238 @@
"""Durable Job side-effect executor for the AIOS campaign control plane."""
from __future__ import annotations
import argparse
import base64
from dataclasses import dataclass
import json
from pathlib import Path
from typing import Any, Mapping
from aios_authority_git_executor import AuthorityGitExecutor, AuthorityGitError
from aios_campaign_recovery import RecoveryEvidenceError, load_recovered_authorization
from aios_gitea_machine_broker import CampaignCapabilityReceipt, GiteaMachineBroker, MachineBrokerError
from aios_gitops_job_controller import GitOpsJobController, JobError, JobState, JobStore
TASK_ID = "AIOS_AUTONOMOUS_RELEASE_CAMPAIGN_V1"
CAMPAIGN_ID = "AIOS-AUTONOMOUS-RELEASE-CAMPAIGN-20260807"
REPOSITORY = "aios/aios"
BRANCH = "agent/autonomous-release-campaign-v1-20260807"
EXPECTED_BRANCH_OID = "bba675fab3b7e0c5fba7066720252b3c91b9f20d"
AUTHORIZATION_DIGEST = "6b0bf550eb702126e967eeff2d561fa808444b924d8a71f5399accad508ed3b2"
LINEAGE_ANCHOR = "69b45e81dcd2c3707c367003b08c55fd361faae8"
class JobExecutionError(RuntimeError):
pass
@dataclass(slots=True)
class DurableJobExecutor:
controller: GitOpsJobController
authority: AuthorityGitExecutor
broker: GiteaMachineBroker
recovery_path: Path
def _request(self, job_id: str) -> tuple[dict[str, Any], dict[str, Any]]:
row = self.controller.store.read(job_id)
try:
request = json.loads(row["request_json"])
except (KeyError, TypeError, json.JSONDecodeError) as exc:
raise JobExecutionError("JOB_REQUEST_INVALID") from exc
return row, dict(request.get("binding") or {})
def _validate_campaign(self, request: Mapping[str, Any]) -> str:
required = {
"task_id": TASK_ID,
"campaign_id": CAMPAIGN_ID,
"repository": REPOSITORY,
"branch": BRANCH,
"expected_branch_oid": EXPECTED_BRANCH_OID,
"authorization_recovery_digest": AUTHORIZATION_DIGEST,
}
for key, expected in required.items():
if request.get(key) != expected:
raise JobExecutionError(f"JOB_{key.upper()}_MISMATCH")
new_oid = request.get("new_branch_oid")
if not isinstance(new_oid, str) or len(new_oid) != 40 or any(ch not in "0123456789abcdef" for ch in new_oid):
raise JobExecutionError("NEW_BRANCH_OID_INVALID")
try:
recovery = load_recovered_authorization(self.recovery_path)
except RecoveryEvidenceError as exc:
raise JobExecutionError(str(exc)) from exc
if recovery["digest"] != AUTHORIZATION_DIGEST:
raise JobExecutionError("RECOVERY_DIGEST_MISMATCH")
for key, expected in required.items():
if key == "campaign_id":
actual = recovery.get(key)
elif key == "authorization_recovery_digest":
continue
else:
actual = recovery.get({"expected_branch_oid": "expected_branch_oid"}.get(key, key))
if actual is not None and actual != expected:
raise JobExecutionError("RECOVERY_BINDING_MISMATCH")
if recovery.get("expected_branch_oid") != EXPECTED_BRANCH_OID:
raise JobExecutionError("RECOVERY_BRANCH_OID_MISMATCH")
authority_head = self.authority.read_ref(f"refs/heads/{BRANCH}")
if authority_head not in {EXPECTED_BRANCH_OID, new_oid}:
raise JobExecutionError("AUTHORITY_BRANCH_EXPECTATION_MISMATCH")
if not self.authority.ancestry_check(EXPECTED_BRANCH_OID, new_oid):
raise JobExecutionError("NEW_OID_NOT_DESCENDANT")
identity = self.broker.verify_identity()
if identity.get("repository") != REPOSITORY or identity.get("login") == "aiman":
raise JobExecutionError("MACHINE_IDENTITY_SCOPE_INVALID")
return new_oid
def dry_run(self, job_id: str) -> dict[str, Any]:
_row, request = self._request(job_id)
new_oid = self._validate_campaign(request)
remote = self.broker.readback()
if remote is None:
action = "CREATE_REQUIRED"
elif remote == new_oid:
action = "ALREADY_IN_SYNC"
elif remote == EXPECTED_BRANCH_OID:
action = "FAST_FORWARD_REQUIRED"
else:
action = "MANUAL_OWNER_REVIEW_REQUIRED"
if action == "MANUAL_OWNER_REVIEW_REQUIRED":
raise JobExecutionError("GITEA_BRANCH_CONFLICT")
return {
"decision": "SAFE_TO_ADVANCE",
"authority_action": "CAS_REQUIRED",
"gitea_action": action,
"previous_branch_oid": EXPECTED_BRANCH_OID,
"final_branch_oid": new_oid,
"force_push": False,
}
def _recover_requested(self, job_id: str) -> dict[str, Any]:
row = self.controller.store.read(job_id)
if JobState(row["state"]) in {JobState.RECOVERABLE_FAILURE, JobState.RECONCILIATION_REQUIRED}:
self.controller.resume(job_id)
row = self.controller.store.read(job_id)
return row
def execute(self, job_id: str, *, dry_run: bool = False) -> dict[str, Any]:
row, request = self._request(job_id)
operation = request.get("operation")
if operation == "ADVANCE_EXISTING_CAMPAIGN_BRANCH":
if dry_run:
return self.dry_run(job_id)
return self._advance_campaign(job_id)
if operation == "COMMIT_FILE":
return self._commit_file(job_id, request)
if operation == "SYNC_TASK_BRANCH_TO_GITEA":
new_oid = self._validate_campaign(request)
final = self.broker.push_branch(new_oid)
return {"operation": operation, "remote_verified": final == new_oid, "final_network_oid": final}
if operation == "CREATE_PR":
return dict(self.broker.create_pr("AIOS autonomous release control plane", "Durable campaign continuation; exact head and recovery evidence are bound in the Job."))
if operation == "PUBLISH_CHECK":
return dict(self.broker.status("success", "Durable campaign control-plane checks passed"))
if operation == "PROTECTED_MERGE":
number = int(request.get("pr_number", "0"))
return dict(self.broker.merge(number))
if operation == "RECONCILE_AUTHORITY_MAIN":
old = request.get("expected_main_oid")
new = request.get("new_branch_oid")
if not isinstance(old, str) or not isinstance(new, str):
raise JobExecutionError("MAIN_RECONCILIATION_BINDING_INVALID")
return {"authority_main": self.authority.ff_CAS_main_after_verified_protected_merge(old, new)}
raise JobExecutionError("OPERATION_NOT_ALLOWED")
def _advance_campaign(self, job_id: str) -> dict[str, Any]:
self._recover_requested(job_id)
row, request = self._request(job_id)
new_oid = self._validate_campaign(request)
state = JobState(row["state"])
if state == JobState.REQUESTED:
receipt = self.controller.advance(job_id, JobState.AUTHORIZED, side_effect_state="AUTHORIZED", receipt={"authorization_digest": AUTHORIZATION_DIGEST, "task_id": TASK_ID})
row = self.controller.store.read(job_id)
if JobState(row["state"]) == JobState.AUTHORIZED:
self.controller.lease_job(job_id, lease_id=f"lease-{job_id}", lineage_anchor_oid=LINEAGE_ANCHOR, current_head_oid=EXPECTED_BRANCH_OID, generation=1, lease_expires_at="2099-01-01T00:00:00Z", expected_revision=int(row["revision"]))
row = self.controller.store.read(job_id)
if JobState(row["state"]) == JobState.LEASED:
self.controller.advance(job_id, JobState.WORKTREE_READY, side_effect_state="WORKTREE_READY", receipt={"branch": BRANCH, "expected_oid": EXPECTED_BRANCH_OID})
row = self.controller.store.read(job_id)
if JobState(row["state"]) == JobState.WORKTREE_READY:
self.controller.advance(job_id, JobState.EXECUTING, side_effect_state="IN_PROGRESS")
row = self.controller.store.read(job_id)
authority_before = self.authority.read_ref(f"refs/heads/{BRANCH}")
if authority_before == EXPECTED_BRANCH_OID:
authority_oid = self.authority.advance_task_branch_CAS(BRANCH and f"refs/heads/{BRANCH}", EXPECTED_BRANCH_OID, new_oid)
elif authority_before == new_oid:
authority_oid = new_oid
else:
raise JobExecutionError("AUTHORITY_BRANCH_CONFLICT")
row = self.controller.store.read(job_id)
if JobState(row["state"]) == JobState.EXECUTING:
self.controller.advance(job_id, JobState.LOCAL_VERIFIED, side_effect_state="LOCAL_VERIFIED", receipt={"authority": {"branch": BRANCH, "previous_oid": EXPECTED_BRANCH_OID, "new_oid": new_oid, "CAS_success": True, "readback_oid": authority_oid}})
row = self.controller.store.read(job_id)
remote_before = self.broker.readback()
if remote_before != new_oid:
if remote_before not in {None, EXPECTED_BRANCH_OID}:
raise JobExecutionError("GITEA_BRANCH_CONFLICT")
if JobState(row["state"]) == JobState.LOCAL_VERIFIED:
self.controller.advance(job_id, JobState.REMOTE_SYNCING, side_effect_state="IN_PROGRESS", receipt={"gitea": {"previous_oid": remote_before, "desired_oid": new_oid, "force_push": False}})
final_network = self.broker.push_branch(new_oid)
else:
final_network = new_oid
final_api = self.broker.readback()
if final_api != new_oid:
raise JobExecutionError("GITEA_BRANCH_READBACK_MISMATCH")
row = self.controller.store.read(job_id)
if JobState(row["state"]) in {JobState.LOCAL_VERIFIED, JobState.REMOTE_SYNCING}:
self.controller.advance(job_id, JobState.REMOTE_VERIFIED, side_effect_state="REMOTE_VERIFIED", receipt={"gitea": {"machine_actor": "aios-autonomy-agent", "previous_oid": remote_before, "desired_oid": new_oid, "final_network_oid": final_network, "final_API_oid": final_api, "force_push": False}})
row = self.controller.store.read(job_id)
if JobState(row["state"]) == JobState.REMOTE_VERIFIED:
self.controller.advance(job_id, JobState.COMPLETED, side_effect_state="REMOTE_VERIFIED", receipt={"authority_oid": authority_oid, "network_oid": final_network, "gitea_api_oid": final_api, "force_push": False})
return self.controller.store._receipt(self.controller.store.read(job_id), duplicate=True).as_dict()
def _commit_file(self, job_id: str, request: Mapping[str, Any]) -> dict[str, Any]:
row = self.controller.store.read(job_id)
payload_ref = request.get("payload_ref")
if not payload_ref:
raise JobExecutionError("PAYLOAD_REQUIRED")
payload = self.controller.store.payloads.read_verified(payload_ref, row["payload_digest"], int(request["payload_size"]))
try:
spec = json.loads(payload.decode())
path = str(spec["path"])
content = base64.b64decode(spec["content_base64"], validate=True)
except Exception as exc:
raise JobExecutionError("COMMIT_FILE_PAYLOAD_INVALID") from exc
new_oid = self.authority.commit_file_in_isolated_worktree(base_oid=row["expected_branch_oid"], files={path: content}, message="chore(gitops): durable commit-file", worktree_name=job_id)
return {"operation": "COMMIT_FILE", "new_oid": new_oid}
def build_from_environment(job_root: Path) -> DurableJobExecutor:
from aios_gitea_machine_transport import GiteaMachineTransport
from aios_gitea_machine_broker import MachineCredentialProjection
account = __import__("os").environ.get("AIOS_GITEA_MACHINE_ACCOUNT", "aios-autonomy-agent")
api = __import__("os").environ.get("AIOS_GITEA_API", "https://g.cnnetmkt.top/git/api/v1")
remote = __import__("os").environ.get("AIOS_GITEA_REMOTE", "https://g.cnnetmkt.top/git/aios/aios.git")
authority_dir = Path(__import__("os").environ.get("AIOS_GITOPS_AUTHORITY_GIT_DIR", "/data/aios-gitops/repo.git"))
worktree_root = Path(__import__("os").environ.get("AIOS_GITOPS_WORKTREE_ROOT", "/var/lib/aios-gitops/worktrees"))
recovery = Path(__import__("os").environ.get("AIOS_RELEASE_RECOVERY_RECEIPT", "/var/lib/aios-gitops/release-campaigns/AIOS-AUTONOMOUS-RELEASE-CAMPAIGN-20260807/authorization-recovery.json"))
transport = GiteaMachineTransport(api, REPOSITORY, remote, authority_dir)
credential = MachineCredentialProjection.from_mapping({"scheme": "systemd-credential", "name": "gitea-machine-token", "account": account})
receipt = CampaignCapabilityReceipt(CAMPAIGN_ID, TASK_ID, REPOSITORY, BRANCH, EXPECTED_BRANCH_OID, frozenset({"exact_task_branch_push", "pr_create_update", "status_publish", "protected_merge", "remote_readback"}), frozenset({"direct_main_push", "force_push", "tag_update", "tag_delete", "baseline_write", "production_deploy"}))
return DurableJobExecutor(GitOpsJobController(JobStore(job_root)), AuthorityGitExecutor(authority_dir, worktree_root), GiteaMachineBroker(transport, credential, receipt), recovery)
def main(argv: list[str] | None = None) -> int:
parser = argparse.ArgumentParser()
parser.add_argument("--job-root", required=True, type=Path)
parser.add_argument("--job-id", required=True)
parser.add_argument("--dry-run", action="store_true")
args = parser.parse_args(argv)
result = build_from_environment(args.job_root).execute(args.job_id, dry_run=args.dry_run)
print(json.dumps(result, sort_keys=True))
return 0
if __name__ == "__main__":
raise SystemExit(main())
@@ -29,10 +29,16 @@ def runtime_policy(environment: Mapping[str, str] | None = None) -> dict[str, bo
return {"legacy_global_write": legacy, "durable_job_execution": durable, "repository_mutation_allowed": durable}
def handle_request(controller: GitOpsJobController, request: Mapping[str, Any]) -> dict[str, Any]:
def handle_request(controller: GitOpsJobController, request: Mapping[str, Any], executor: Any | None = None) -> dict[str, Any]:
job_id = str(request.get("job_id") or "")
if not job_id:
raise JobError("JOB_ID_REQUIRED")
if request.get("action") == "execute":
if executor is None:
raise JobError("JOB_EXECUTOR_UNAVAILABLE")
result = executor.execute(job_id, dry_run=bool(request.get("dry_run", False)))
receipt = controller.store._receipt(controller.store.read(job_id), duplicate=True)
return {"job": result, **receipt.as_dict()}
if request.get("action") == "resume":
receipt = controller.resume(job_id)
else:
@@ -41,7 +47,7 @@ def handle_request(controller: GitOpsJobController, request: Mapping[str, Any])
return receipt.as_dict()
def serve_socket(sock: socket.socket, controller: GitOpsJobController, stopping: threading.Event) -> None:
def serve_socket(sock: socket.socket, controller: GitOpsJobController, stopping: threading.Event, executor: Any | None = None) -> None:
sock.listen(32)
sock.settimeout(1.0)
while not stopping.is_set():
@@ -53,7 +59,7 @@ def serve_socket(sock: socket.socket, controller: GitOpsJobController, stopping:
try:
raw = connection.makefile("rb").readline(1_048_576)
request = json.loads(raw.decode("utf-8"))
response = {"status": "ok", **handle_request(controller, request)}
response = {"status": "ok", **handle_request(controller, request, executor)}
except ServiceUnavailable as exc:
response = {"status": "error", "error": str(exc), "retryable": True, "side_effect_state": "UNKNOWN"}
except Exception as exc:
@@ -77,12 +83,14 @@ def main(argv: list[str] | None = None) -> int:
if not policy["durable_job_execution"]:
raise RuntimeError("DURABLE_JOB_EXECUTION_GATE_CLOSED")
controller = GitOpsJobController(JobStore(args.job_root))
from aios_gitops_job_executor import build_from_environment
executor = build_from_environment(args.job_root)
stopping = threading.Event()
signal.signal(signal.SIGTERM, lambda *_: stopping.set())
signal.signal(signal.SIGINT, lambda *_: stopping.set())
sock = socket_from_systemd()
try:
serve_socket(sock, controller, stopping)
serve_socket(sock, controller, stopping, executor)
finally:
sock.close()
return 0
+7 -1
View File
@@ -12,6 +12,12 @@ Environment=AIOS_GITOPS_WRITES_ENABLED=false
Environment=AIOS_GITOPS_DURABLE_JOB_EXECUTION_ENABLED=true
Environment=AIOS_GITOPS_JOB_ROOT=/var/lib/aios-gitops/jobs
Environment=AIOS_GITEA_MACHINE_CREDENTIAL_REF=systemd-credential:gitea-machine-token
Environment=AIOS_GITOPS_AUTHORITY_GIT_DIR=/data/aios-gitops/repo.git
Environment=AIOS_GITOPS_WORKTREE_ROOT=/var/lib/aios-gitops/worktrees
Environment=AIOS_GITEA_API=https://g.cnnetmkt.top/git/api/v1
Environment=AIOS_GITEA_REMOTE=https://g.cnnetmkt.top/git/aios/aios.git
Environment=AIOS_GITEA_MACHINE_ACCOUNT=aios-autonomy-agent
Environment=AIOS_RELEASE_RECOVERY_RECEIPT=/var/lib/aios-gitops/release-campaigns/AIOS-AUTONOMOUS-RELEASE-CAMPAIGN-20260807/authorization-recovery.json
LoadCredentialEncrypted=gitea-machine-token:/etc/aios/credentials/gitea-machine.token.cred
Environment=PYTHONDONTWRITEBYTECODE=1
ExecStart=/usr/bin/python3 -B /opt/aios-gitops/active-runtime/app/aios_gitops_writer_worker.py --job-root ${AIOS_GITOPS_JOB_ROOT}
@@ -21,7 +27,7 @@ NoNewPrivileges=no
PrivateTmp=true
ProtectSystem=strict
ProtectHome=true
ReadWritePaths=/var/lib/aios-gitops/jobs /var/lib/aios-gitops/payloads
ReadWritePaths=/var/lib/aios-gitops/jobs /var/lib/aios-gitops/payloads /var/lib/aios-gitops/worktrees
[Install]
WantedBy=multi-user.target
+165
View File
@@ -0,0 +1,165 @@
from __future__ import annotations
import tempfile
from pathlib import Path
import pytest
import aios_gitops_job_executor as executor_module
from aios_authority_git_executor import AuthorityGitError, AuthorityGitExecutor
from aios_gitea_machine_transport import GiteaMachineTransport
from aios_gitea_machine_broker import CampaignCapabilityReceipt, GiteaMachineBroker, MachineCredentialProjection
from aios_gitops_job_controller import GitOpsJobController, JobError, JobStore, JobState
from aios_gitops_writer_worker import handle_request
OLD = "bba675fab3b7e0c5fba7066720252b3c91b9f20d"
NEW = "3fbbc18fb2131bbbd9e511004a59ee95cf3e86b7"
class FakeAuthority:
def __init__(self, descendant=True):
self.head = OLD
self.descendant = descendant
self.cas_calls = []
def read_ref(self, ref):
return self.head
def ancestry_check(self, older, newer):
return self.descendant and older == OLD and newer == NEW
def advance_task_branch_CAS(self, branch, expected, new):
if self.head != expected:
raise AuthorityGitError("AUTHORITY_BRANCH_EXPECTATION_MISMATCH")
self.cas_calls.append((branch, expected, new))
self.head = new
return new
class FakeBroker:
def __init__(self):
self.remote = None
self.push_calls = []
def verify_identity(self):
return {"login": "aios-autonomy-agent", "repository": "aios/aios"}
def readback(self):
return self.remote
def push_branch(self, new):
self.push_calls.append(new)
self.remote = new
return new
def create_pr(self, *args):
return {"number": 1}
def status(self, *args):
return {"status": "success"}
def merge(self, *args):
return {"merged": True}
def _request(**extra):
value = {
"campaign_id": executor_module.CAMPAIGN_ID,
"repository": executor_module.REPOSITORY,
"operation": "ADVANCE_EXISTING_CAMPAIGN_BRANCH",
"branch": executor_module.BRANCH,
"expected_branch_oid": OLD,
"payload_digest": "0" * 64,
"task_id": executor_module.TASK_ID,
"new_branch_oid": NEW,
"authorization_recovery_digest": executor_module.AUTHORIZATION_DIGEST,
}
value.update(extra)
return value
def _executor(temp, authority=None, broker=None):
controller = GitOpsJobController(JobStore(temp))
job = controller.submit(_request())
executor = executor_module.DurableJobExecutor(controller, authority or FakeAuthority(), broker or FakeBroker(), Path(temp) / "recovery.json")
return controller, job, executor
def _recovery():
return {
"digest": executor_module.AUTHORIZATION_DIGEST,
"campaign_id": executor_module.CAMPAIGN_ID,
"task_id": executor_module.TASK_ID,
"repository": executor_module.REPOSITORY,
"branch": executor_module.BRANCH,
"expected_branch_oid": OLD,
}
def test_advance_existing_branch_success_and_duplicate_execute(monkeypatch):
monkeypatch.setattr(executor_module, "load_recovered_authorization", lambda _path: _recovery())
with tempfile.TemporaryDirectory() as temp:
controller, job, executor = _executor(temp)
first = executor.execute(job.job_id)
second = executor.execute(job.job_id)
assert first["state"] == JobState.COMPLETED.value
assert second["state"] == JobState.COMPLETED.value
assert executor.authority.cas_calls == [(f"refs/heads/{executor_module.BRANCH}", OLD, NEW)]
assert executor.broker.push_calls == [NEW]
def test_authorization_digest_required(monkeypatch):
monkeypatch.setattr(executor_module, "load_recovered_authorization", lambda _path: _recovery())
with tempfile.TemporaryDirectory() as temp:
controller = GitOpsJobController(JobStore(temp))
job = controller.submit(_request(authorization_recovery_digest="bad"))
executor = executor_module.DurableJobExecutor(controller, FakeAuthority(), FakeBroker(), Path(temp) / "recovery.json")
with pytest.raises(executor_module.JobExecutionError, match="AUTHORIZATION_RECOVERY_DIGEST_MISMATCH"):
executor.dry_run(job.job_id)
def test_wrong_branch_rejected(monkeypatch):
monkeypatch.setattr(executor_module, "load_recovered_authorization", lambda _path: _recovery())
with tempfile.TemporaryDirectory() as temp:
controller = GitOpsJobController(JobStore(temp))
job = controller.submit(_request(branch="agent/not-the-campaign"))
executor = executor_module.DurableJobExecutor(controller, FakeAuthority(), FakeBroker(), Path(temp) / "recovery.json")
with pytest.raises(executor_module.JobExecutionError, match="BRANCH_MISMATCH"):
executor.dry_run(job.job_id)
def test_non_descendant_new_oid_rejected(monkeypatch):
monkeypatch.setattr(executor_module, "load_recovered_authorization", lambda _path: _recovery())
with tempfile.TemporaryDirectory() as temp:
authority = FakeAuthority(descendant=False)
controller, job, executor = _executor(temp, authority=authority)
with pytest.raises(executor_module.JobExecutionError, match="NEW_OID_NOT_DESCENDANT"):
executor.dry_run(job.job_id)
def test_worker_execute_action_consumes_job():
with tempfile.TemporaryDirectory() as temp:
controller = GitOpsJobController(JobStore(temp))
job = controller.submit(_request())
class Stub:
def execute(self, job_id, *, dry_run=False):
assert job_id == job.job_id
assert dry_run is True
return {"decision": "SAFE_TO_ADVANCE"}
result = handle_request(controller, {"action": "execute", "job_id": job.job_id, "dry_run": True}, Stub())
assert result["job"]["decision"] == "SAFE_TO_ADVANCE"
def test_machine_transport_branch_cas_is_not_force_push(monkeypatch):
transport = GiteaMachineTransport("http://127.0.0.1:3000/api/v1", "aios/aios", "https://g.cnnetmkt.top/git/aios/aios.git", Path("/tmp/repo.git"))
calls = []
def read_branch(_self, _branch):
return NEW if calls else OLD
monkeypatch.setattr(GiteaMachineTransport, "read_branch", read_branch)
monkeypatch.setattr(GiteaMachineTransport, "_request", lambda _self, method, path, body=None: calls.append((method, path, body)) or (200, {}))
assert transport.compare_and_swap_branch(executor_module.BRANCH, OLD, NEW) == NEW
assert calls[0][0] == "PUT"
assert calls[0][2]["force"] is False