Harbor v2.0 鏡像回收那些事

Harbor 鏡像回收流程

Harbor 鏡像回收分兩部分：

• 一、刪除鏡像 TAG
• 二、執行垃圾清理

早期 Harbor 1.0 版本時，刪除鏡像TAG能夠經過UI界面手動一個個選取刪除，也能夠調用官方提供的SDK刪除（很早之前官方不維護了）。後來官方在 1.9 版本提供TAG保留策略，解決之前鏡像Tag回收麻煩的問題。

TAG保留策略有哪些策略，以下圖：

Harbor Python SDK

做者寫了一個 Python SDK 方法，錦上添花吧！

代碼項目地址：https://github.com/yangpeng14/harbor_sdk_v2.0

注意：Harbor v2.0+ 和 Harbor v1 版本 API調用方式發生很大的改變。安全方面：修復跨域攻擊問題，API登錄Harbor方法也發現變化。

#!/usr/bin/env python3
# -*- coding=utf8 -*-

import logging
import requests

logging.basicConfig(level=logging.INFO)


class HarborClient(object):
    def __init__(self, host, user, password, protocol="http"):
        self.host = host
        self.user = user
        self.password = password
        self.protocol = protocol

        # 第一次get請求，獲取 cookie 信息
        self.cookies, self.headers = self.get_cookie()

        # 獲取登錄成功 session
        self.session_id = self.login()

        # 把登錄成功的 sid值 替換 get_cookie 方法中 cookie sid值，用於 delete 操做
        self.cookies_new = self.cookies
        self.cookies_new.update({'sid': self.session_id})

    # def __del__(self):
    #     self.logout()

    def get_cookie(self):
        response = requests.get("{0}://{1}/c/login".format(self.protocol, self.host))
        csrf_cookie = response.cookies.get_dict()
        headers = {'X-Harbor-CSRF-Token': csrf_cookie['__csrf']}
        return csrf_cookie, headers

    def login(self):
        login_data = requests.post('%s://%s/c/login' %
                                   (self.protocol, self.host),
                                   data={'principal': self.user,
                                         'password': self.password}, cookies=self.cookies, headers=self.headers)

        if login_data.status_code == 200:
            session_id = login_data.cookies.get('sid')

            logging.debug("Successfully login, session id: {}".format(
                session_id))
            return session_id
        else:
            logging.error("Fail to login, please try again")
            return None

    def logout(self):
        requests.get('%s://%s/c/logout' % (self.protocol, self.host),
                     cookies={'sid': self.session_id})
        logging.debug("Successfully logout")

    # GET /projects
    def get_projects(self, project_name=None, is_public=None):
        # TODO: support parameter
        result = []
        page = 1
        page_size = 15

        while True:
            path = '%s://%s/api/v2.0/projects?page=%s&page_size=%s' % (self.protocol, self.host, page, page_size)
            response = requests.get(path,
                                    cookies={'sid': self.session_id})
            if response.status_code == 200:
                logging.debug("Successfully get projects result: {}".format(
                    result))
                if isinstance(response.json(), list):
                    result.extend(response.json())
                    page += 1
                else:
                    break
            else:
                logging.error("Fail to get projects result")
                result = None
                break
        return result

    # GET /projects/{project_name}/repositories
    def get_repositories(self, project_name, query_string=None):
        # TODO: support parameter
        result = []
        page = 1
        page_size = 15

        while True:
            path = '%s://%s/api/v2.0/projects/%s/repositories?page=%s&page_size=%s' % (
                self.protocol, self.host, project_name, page, page_size)
            response = requests.get(path,
                                    cookies={'sid': self.session_id})
            if response.status_code == 200:
                logging.debug(
                    "Successfully get repositories with name: {}, result: {}".format(
                        project_name, result))
                if len(response.json()):
                    result.extend(response.json())
                    page += 1
                else:
                    break
            else:
                logging.error("Fail to get repositories result with name: {}".format(
                    project_name))
                result = None
                break
        return result

    # GET /projects/{project_name}/repositories/{repository_name}/artifacts
    # GET /projects/{project_name}/repositories/{repository_name}/artifacts?with_tag=true&with_scan_overview=true&with_label=true&page_size=15&page=1
    def get_repository_artifacts(self, project_name, repository_name):
        result = []
        page = 1
        page_size = 15

        while True:
            path = '%s://%s/api/v2.0/projects/%s/repositories/%s/artifacts?with_tag=true&with_scan_overview=true&with_label=true&page_size=%s&page=%s' % (
                self.protocol, self.host, project_name, repository_name, page_size, page)
            response = requests.get(path,
                                    cookies={'sid': self.session_id}, timeout=60)
            if response.status_code == 200:
                logging.debug(
                    "Successfully get repositories artifacts with name: {}, {}, result: {}".format(
                        project_name, repository_name, result))
                if len(response.json()):
                    result.extend(response.json())
                    page += 1
                else:
                    break
            else:
                logging.error("Fail to get repositories artifacts result with name: {}, {}".format(
                    project_name, repository_name))
                result = None
                break
        return result

    # DELETE /projects/{project_name}/repositories/{repository_name}
    def delete_repository(self, project_name, repository_name, tag=None):
        # TODO: support to check tag
        # TODO: return 200 but the repo is not deleted, need more test
        result = False
        path = '%s://%s/api/v2.0/projects/%s/repositories/%s' % (
            self.protocol, self.host, project_name, repository_name)
        response = requests.delete(path,
                                   cookies=self.cookies_new, headers=self.headers)
        if response.status_code == 200:
            result = True
            print("Delete {} successful!".format(repository_name))
            logging.debug("Successfully delete repository: {}".format(
                repository_name))
        else:
            logging.error("Fail to delete repository: {}".format(repository_name))
        return result

    # Get /projects/{project_name}/repositories/{repository_name}/artifacts/{reference}/tags
    def get_repository_tags(self, project_name, repository_name, reference_hash):
        result = None
        path = '%s://%s/api/v2.0/projects/%s/repositories/%s/artifacts/%s/tags' % (
            self.protocol, self.host, project_name, repository_name, reference_hash)
        response = requests.get(path,
                                cookies={'sid': self.session_id}, timeout=60)
        if response.status_code == 200:
            result = response.json()
            logging.debug(
                "Successfully get tag with repository name: {}, result: {}".format(
                    repository_name, result))
        else:
            logging.error("Fail to get tags with repository name: {}".format(
                repository_name))
        return result

    # Del /projects/{project_name}/repositories/{repository_name}/artifacts/{reference}/tags/{tag_name}
    def del_repository_tag(self, project_name, repository_name, reference_hash, tag):
        result = False
        path = '%s://%s/api/v2.0/projects/%s/repositories/%s/artifacts/%s/tags/%s' % (
            self.protocol, self.host, project_name, repository_name, reference_hash, tag)
        response = requests.delete(path, cookies=self.cookies_new, headers=self.headers)
        if response.status_code == 200:
            result = True
            print("Delete {} {} {} {} successful!".format(project_name, repository_name, reference_hash, tag))
            logging.debug(
                "Successfully delete repository project_name: {}, repository_name: {}, reference_hash: {}, tag: {}".format(
                    project_name, repository_name, reference_hash, tag))
        else:
            logging.error("Fail to delete repository project_name: {}, repository_name: {}, reference_hash: {}, tag: {}".format(
                project_name, repository_name, reference_hash, tag))
        return result

    # Del /projects/{project_name}/repositories/{repository_name}/artifacts/{reference}
    def del_artifacts_hash(self, project_name, repository_name, reference_hash):
        result = False
        path = '%s://%s/api/v2.0/projects/%s/repositories/%s/artifacts/%s' % (
            self.protocol, self.host, project_name, repository_name, reference_hash)
        response = requests.delete(path, cookies=self.cookies_new, headers=self.headers)
        if response.status_code == 200:
            result = True
            print("Delete artifacts hash {} {} {} successful!".format(project_name, repository_name, reference_hash))
            logging.debug(
                "Successfully delete repository project_name: {}, repository_name: {}, artifacts hash: {}".format(
                    project_name, repository_name, reference_hash))
        else:
            logging.error("Fail to delete repository project_name: {}, repository_name: {}, artifacts hash: {}".format(
                project_name, repository_name, reference_hash))
        return result

使用舉例

列出 Harbor Projects

import harborclient_modify_v2_0

class GetHarborApi(object):
    def __init__(self, host, user, password, protocol="http"):
        self.host = host
        self.user = user
        self.password = password
        self.protocol = protocol
        self.client = harborclient_modify_v2_0.HarborClient(self.host, self.user, self.password, self.protocol)

    def main(self):
        print(self.client.get_projects())

if __name__ == '__main__':
    host = "harbor.example.com"
    user = "admin"
    password = "******"
    protocol = "https"
    cline_get = GetHarborApi(host, user, password, protocol)
    cline_get.main()

熱門文章推薦

• IT運維面試問題總結-簡述Etcd、Kubernetes、Lvs、HAProxy等
• IT運維面試問題總結-數據庫、監控、網絡管理
• IT運維面試問題總結-運維工具、開源應用(Ceph、Docker、Apache、Nginx等)
• IT運維面試問題總結-基礎服務、磁盤管理、虛擬平臺和系統管理
• IT運維面試問題總結-Linux基礎
• Harbor v1.7.0自動鏡像回收
  

最後

• 歡迎您加我微信【 ypxiaozhan01】，拉您進技術羣，一塊兒交流學習...
• 歡迎您關注【 YP小站】，學習互聯網最流行的技術，作個專業的技術人...

---

  【文章讓您有收穫，👇 贊 或者 在看 支持我吧】

本文分享自微信公衆號 - YP小站（ypxiaozhan）。
若有侵權，請聯繫 support@oschina.cn 刪除。
本文參與「OSC源創計劃」，歡迎正在閱讀的你也加入，一塊兒分享。