zioinfo-mail/workspace/guardia-itsm/routers/auto_remediation.py

"""
자동 교정 실행 — 승인 기반 (PAM 패턴 재사용)

드리프트 교정 명령을 관리자 승인 후 SSH 경유로 실행.
롤백 명령 포함.

엔드포인트:
  GET  /api/remediation/jobs           — 교정 작업 목록
  GET  /api/remediation/jobs/{id}      — 교정 작업 상세
  POST /api/remediation/approve/{id}   — 승인 후 실행
  POST /api/remediation/reject/{id}    — 거부
  POST /api/remediation/rollback/{id}  — 롤백 실행
"""
from __future__ import annotations

import logging
from datetime import datetime

import paramiko
from fastapi import APIRouter, Depends, HTTPException
from sqlalchemy import select, desc
from sqlalchemy.ext.asyncio import AsyncSession

from core.auth import get_current_user, require_admin_role
from core.ssh_exec import _decrypt_password as decrypt_password
from database import get_db
from models import User, Server, AutoRemediationJob, AuditLog

logger = logging.getLogger(__name__)
router = APIRouter(prefix="/api/remediation", tags=["자동 교정"])


async def _run_fix(server: Server, cmd: str) -> tuple[bool, str]:
    """SSH 경유 교정 명령 실행 — 에이전트리스."""
    try:
        pw = decrypt_password(server.os_pw_enc)
        ssh = paramiko.SSHClient()
        ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
        ssh.connect(server.ip_addr, username=server.ssh_user, password=pw, timeout=10)
        _, stdout, stderr = ssh.exec_command(cmd, timeout=30)
        out = stdout.read().decode('utf-8', 'replace').strip()
        err = stderr.read().decode('utf-8', 'replace').strip()
        exit_code = stdout.channel.recv_exit_status()
        ssh.close()
        if exit_code == 0:
            return True, out
        return False, f"exit={exit_code}: {err[:200]}"
    except Exception as e:
        return False, str(e)[:200]


@router.get("/jobs")
async def list_jobs(
    status: str = None,
    limit: int = 50,
    db: AsyncSession = Depends(get_db),
    user: User = Depends(get_current_user),
):
    q = select(AutoRemediationJob).order_by(desc(AutoRemediationJob.created_at)).limit(limit)
    if status:
        q = q.where(AutoRemediationJob.status == status)
    rows = await db.execute(q)
    jobs = rows.scalars().all()
    return [
        {
            "id": j.id, "server_id": j.server_id, "item_key": j.item_key,
            "fix_cmd": j.fix_cmd[:80] + "..." if len(j.fix_cmd or "") > 80 else j.fix_cmd,
            "status": j.status, "result": j.result_message,
            "created_at": j.created_at, "executed_at": j.executed_at,
        }
        for j in jobs
    ]


@router.get("/jobs/{job_id}")
async def get_job(
    job_id: int,
    db: AsyncSession = Depends(get_db),
    user: User = Depends(get_current_user),
):
    row = await db.execute(select(AutoRemediationJob).where(AutoRemediationJob.id == job_id))
    job = row.scalar_one_or_none()
    if not job:
        raise HTTPException(404)
    return {
        "id": job.id, "server_id": job.server_id,
        "item_key": job.item_key, "fix_cmd": job.fix_cmd,
        "status": job.status, "result": job.result_message,
        "approved_by": job.approved_by,
        "created_at": job.created_at, "executed_at": job.executed_at,
    }


@router.post("/approve/{job_id}")
async def approve_and_execute(
    job_id: int,
    db: AsyncSession = Depends(get_db),
    user: User = Depends(require_admin_role),
):
    """승인 후 즉시 교정 실행."""
    row = await db.execute(select(AutoRemediationJob).where(AutoRemediationJob.id == job_id))
    job = row.scalar_one_or_none()
    if not job:
        raise HTTPException(404)
    if job.status != "PENDING_APPROVAL":
        raise HTTPException(400, f"현재 상태: {job.status}")

    srv_row = await db.execute(select(Server).where(Server.id == job.server_id))
    server = srv_row.scalar_one_or_none()
    if not server:
        raise HTTPException(404, "서버 없음")

    job.status = "EXECUTING"
    job.approved_by = user.id
    await db.commit()

    success, result = await _run_fix(server, job.fix_cmd)
    job.status = "SUCCESS" if success else "FAILED"
    job.result_message = result[:500]
    job.executed_at = datetime.utcnow()

    # 감사 로그
    log = AuditLog(
        user_id=user.id,
        action="AUTO_REMEDIATION",
        detail=f"서버 {server.hostname}: {job.item_key} 교정 {'성공' if success else '실패'}",
        created_at=datetime.utcnow(),
    )
    db.add(log)
    await db.commit()

    return {"ok": success, "job_id": job_id, "status": job.status, "result": result[:200]}


@router.post("/reject/{job_id}")
async def reject_job(
    job_id: int,
    reason: str = "관리자 거부",
    db: AsyncSession = Depends(get_db),
    user: User = Depends(require_admin_role),
):
    row = await db.execute(select(AutoRemediationJob).where(AutoRemediationJob.id == job_id))
    job = row.scalar_one_or_none()
    if not job:
        raise HTTPException(404)
    job.status = "REJECTED"
    job.result_message = reason
    await db.commit()
    return {"ok": True}


@router.post("/rollback/{job_id}")
async def rollback_job(
    job_id: int,
    db: AsyncSession = Depends(get_db),
    user: User = Depends(require_admin_role),
):
    """교정 롤백 실행."""
    row = await db.execute(select(AutoRemediationJob).where(AutoRemediationJob.id == job_id))
    job = row.scalar_one_or_none()
    if not job or not job.rollback_cmd:
        raise HTTPException(400, "롤백 명령 없음")
    if job.status != "SUCCESS":
        raise HTTPException(400, "실행 완료된 작업만 롤백 가능")

    srv_row = await db.execute(select(Server).where(Server.id == job.server_id))
    server = srv_row.scalar_one_or_none()
    if not server:
        raise HTTPException(404)

    success, result = await _run_fix(server, job.rollback_cmd)
    job.status = "ROLLED_BACK" if success else "ROLLBACK_FAILED"
    job.result_message = f"ROLLBACK: {result[:400]}"
    await db.commit()

    return {"ok": success, "status": job.status, "result": result[:200]}