feat(chv-guest): Gitea 同步变量经 seed 进 guest-env(执行器集成)

- 新增 cloud/chv/gitea_sync.py:6 键白名单,三源合并
  (宿主 CHV_GITEA_* < 任务 gitea_sync < extra_env),
  必填 4 键缺失即 fail-closed,空即不同步
- ssh_transport:GITEA_TOKEN 进密钥集(argv 拒绝),其余
  5 键进 guest-env 集(split_guest_env)
- hostkeys.write_seed:guest_env 与 secret_env 合并写同一
  条目,仍无 runcmd、0600
- config:_gitea_fragment 产 gitea_sync_env(函数内 import,
  叶子模块顶层只许标准库)
- executor.wake_vm:起 VMM 前合并+校验(gitea_sync_incomplete),
  guest-env 随 seed 投递;provision_host_key 透传 guest_env
- worker:_run_attempt 仅对 CHV 执行器把同步变量并进 gateway_env
- 测试:cloud/tests/test_chv_gitea_sync.py 20 项
- 边界:tasks 表无 JSON 列,DB 往返丢任务 gitea_sync;
  生产主来源是宿主 CHV_GITEA_* 环境变量
This commit is contained in:
host committed 2026-11-06 16:55:32 +00:00
1 parent 67ca0b86cd
commit e7f699c62b
7 files changed
+785 -37

No files matched your search

+55 -7
View File
@@ -59,6 +59,7 @@ from cloud.chv.ssh_transport import build_ssh_command as _build_ssh_cmd
from cloud.chv.ssh_transport import known_hosts_path as _known_hosts_path
from cloud.chv.ssh_transport import (
parse_keyscan_output,
split_guest_env,
split_secret_env,
write_known_hosts,
)
@@ -398,7 +399,8 @@ class CloudHypervisorExecutor(Executor):
"ssh_forward", "reuse_experiment",
"single_tenant_project", "snapshot_max_count",
"snapshot_max_total_gb", "snapshot_retention_days",
"snapshot_dir", "allow_remote_snapshot"):
"snapshot_dir", "allow_remote_snapshot",
"gitea_sync_env"):
setattr(self, _field, getattr(_cfg, _field))
# -- 名字 / 路径 ---------------------------------------------------
@@ -827,7 +829,8 @@ class CloudHypervisorExecutor(Executor):
[known_hosts_line(self.ssh_port, public_text,
host=ssh_host)])
def provision_host_key(self, task: dict, secret_env: dict | None = None) -> dict:
def provision_host_key(self, task: dict, secret_env: dict | None = None,
guest_env: dict | None = None) -> dict:
"""按 VM 预置 host key(补缺口轨强方案,见 hostkeys.py):
启动前为该 VM 生成一次性 ed25519 host key 对,公钥
立即写进该 VM 的 known_hosts(钉死),私钥只写进
@@ -837,7 +840,9 @@ class CloudHypervisorExecutor(Executor):
不同 VM 的 key 对不同(state_dir/<vm> 隔离),同一
VM 重启复用已有对(否则钉死立即失效)。
feat-chv-secfix(H2/M2):secret_env 给定时写进
seed 的客内密钥文件(0600);上次清理失败留下的
seed 的客内密钥文件(0600);guest_env 给定时与
secret_env 合并写同一客内文件(boot 时要用的非密钥
同步变量,见 gitea_sync);上次清理失败留下的
脏标记存在时直接拒绝(清理失败即拒起,不再复用
旧 key 对)。成功后审计 host_key_generated(只含
VM 名与密钥名,不含任何 key 值)。"""
@@ -862,7 +867,8 @@ class CloudHypervisorExecutor(Executor):
self._write_provisioned_known_hosts(paths, public_text)
seed_dir = write_seed(paths["seed_dir"], vm,
private_text, public_text,
secret_env=secret_env)
secret_env=secret_env,
guest_env=guest_env)
self._emit("host_key_generated", task,
"vm=%s secret_names=%s"
% (vm, ",".join(sorted(secret_env or {}))))
@@ -1305,6 +1311,37 @@ class CloudHypervisorExecutor(Executor):
"VMM api-socket 组/其它可写(%o),拒绝启动:%s"
% (mode, socket_path))
def _effective_extra_env(self, task: dict,
extra_env: dict | None) -> dict | None:
"""wake 生效的 extra_env:把 Gitea 同步变量并进来。
优先级(低→高):执行器宿主配置 CHV_GITEA_*(构造时
已映射成 gitea_sync_env)< 任务 dict 的 gitea_sync <
显式 extra_env 同名键(见 gitea_sync.extract_...)。
同步已请求(任一同步键出现)但必填键不全时,在起
VMM 之前就 ChvFailure(gitea_sync_incomplete,
fail-closed;错误只含缺失键名,不含值)——半份同步
配置绝不能让 VM 带着“以为同步了”的状态启动。无同步
变量时原样返回(本次启动不同步)。
必须在 _wake_prechecks 之前调:任务/配置来源的
GITEA_TOKEN 也要受 keyscan 模式 fail-closed 检查。"""
from cloud.chv.gitea_sync import (
extract_gitea_sync_env,
validate_gitea_sync_env,
)
sync = extract_gitea_sync_env(
task=task, extra_env=extra_env,
config_sync=getattr(self, "gitea_sync_env", None))
try:
validate_gitea_sync_env(sync)
except ValueError as exc:
raise ChvFailure("gitea_sync_incomplete", str(exc))
if not sync:
return extra_env
merged = dict(extra_env or {})
merged.update(sync)
return merged
def _wake_prechecks(self, task: dict, name: str, extra_env) -> tuple:
"""wake 前置:preflight/容量/配额/net/密钥模式,纯移动。
失败时发 vm_failed 审计并原样抛 ChvFailure。"""
@@ -1343,7 +1380,8 @@ class CloudHypervisorExecutor(Executor):
raise
return secrets, net_entries
def _provision_seed_disk(self, task: dict, name: str, secrets) -> str | None:
def _provision_seed_disk(self, task: dict, name: str, secrets,
guest_env: dict | None = None) -> str | None:
"""provisioned 模式的 seed 盘供给:生成→制盘→挂盘,
任一步失败即抛(调用方 stop_vm 清理),纯移动。"""
seed_disk = None
@@ -1366,7 +1404,8 @@ class CloudHypervisorExecutor(Executor):
seed_disk_path,
)
provisioned = self.provision_host_key(
task, secret_env=secrets or None)
task, secret_env=secrets or None,
guest_env=guest_env or None)
seed_disk = seed_disk_path(self.state_dir, name)
build_seed_disk(
provisioned["seed_dir"], seed_disk,
@@ -1446,10 +1485,19 @@ class CloudHypervisorExecutor(Executor):
if not isinstance(task, dict) or task.get("id") is None:
raise ValueError("任务缺少 id")
name = self.vm_name(task)
try:
extra_env = self._effective_extra_env(task, extra_env)
except ChvFailure as exc:
self._emit("vm_failed", task,
"vm=%s wake 失败 kind=%s" % (name, exc.kind),
exc.kind)
raise
secrets, net_entries = self._wake_prechecks(task, name, extra_env)
guest_env, _plain = split_guest_env(extra_env)
task_disk, socket_path = self._wake_prepare(task, name)
try:
seed_disk = self._provision_seed_disk(task, name, secrets)
seed_disk = self._provision_seed_disk(
task, name, secrets, guest_env=guest_env or None)
self._boot_vm(task, name, task_disk, seed_disk, net_entries)
except Exception:
try: