1323844 九、發明說明： 【發明所屬之技術領域】 本發明係關於多主機系統或叢集系統之服務接管技術，特別 - 係關於一種高可用性叢集系統之服務接管系统及方法。 -【先前技術】 目鈾為了使運々亍重要業務之電腦糸統提供不間斷的服務，最 普通的做法是佈置高可用（HighAvailable)之叢集(Cluster)或多主 _ 機系統。南可用叢集通常由至少兩台主機構成，在對外提供服務 之過程中，一台主機提供正常的服務，其它主機則處於待命 狀態。並且各個主機之間透過〃心跳（Heartbeat)〃機制實現運作 狀態之互相監控。 例如「第1圖」顯示了-種典型高可用性之叢集結構示意圖， 在圖示實例中，整個系統即主機系統10由兩台主機構成，主機 12與域14，並分職有—私有網際協定㈤Vate Inters # Protocol ’ Private π>)位址 192 168 〇1 及 192 168 〇 2，但主機系統 10對外提供服務則透過一可訪問之公共網際協定位址，或稱網際 協定（P砸c Internet Protocol，publicIp)位址 1〇1〇11〇。客戶端 通過公共網際協定位址訪問主機系統1〇，從客戶端之角度來看， 整個系統就是-台提供—公共網協定位址:1() 1()]⑽之主機系 統，所以整個系統對客戶端隱藏了具體的結構。兩個主機Η盘 =之間通過“心跳”機制互相债測對方之狀態，當“待命，，主機 檢剩目則提供服務之主機發生故障而不能提供服 不穩定時’由“騎”錢鮮公共輯協定他及工作，對外 5 1323844 因此’在運作主機及時接管了網際協定’如果在接管前沒有 做好充分的軟硬體環境準備，尤其是需較長時間硬體準備之後才 能安全接管網路服務之情訂，例如網路小㈣财統介面服務 -必須保證在網際協定位址和網路服務本身接管前，硬碟和相應的 -廉價磁碟冗餘陣列（raid ’ Redundani〜h—e1323844 IX. Description of the Invention: [Technical Field] The present invention relates to a service takeover technology for a multi-master system or a cluster system, and more particularly to a service takeover system and method for a high availability cluster system. - [Prior Art] In order to provide uninterrupted service to the computer system of the important business, the most common method is to arrange a high-available cluster or multi-master system. The south available cluster usually consists of at least two hosts. In the process of providing external services, one host provides normal services, while other hosts are on standby. And each host communicates with each other through a heartbeat mechanism. For example, "Figure 1" shows a cluster structure diagram of typical high availability. In the illustrated example, the entire system, host system 10, consists of two hosts, host 12 and domain 14, and has a separate private network agreement. (5) Vate Inters # Protocol 'Private π>) addresses 192 168 〇 1 and 192 168 〇 2, but the host system 10 provides external services through an accessible public Internet Protocol address, or Internet Protocol (P砸c Internet Protocol). , publicIp) Address 1〇1〇11〇. The client accesses the host system through the public internet protocol address. From the client's point of view, the whole system is the host system provided by the station-public network protocol address: 1() 1()](10), so the whole system The specific structure is hidden from the client. The two host disks = the state of each other through the "heartbeat" mechanism to test each other's state, when "standby, the host checks the remaining host, the host that provides the service fails and can not provide service instability" The public series agreed with him and his work, external 5 1323844. Therefore, 'the Internet host was taken over in the operation host'. If the software and hardware environment preparation is not completed before the takeover, especially after a long time of hardware preparation, the network can be safely taken over. The service of the road service, such as the network small (four) financial interface service - must ensure that before the Internet Protocol address and the network service itself take over, the hard disk and the corresponding - cheap disk redundant array (raid 'Redundani ~ h- e
Ms)、邏輯卷（LV)設置已經準備好，因為涉及到硬體準備通 常需要較多的時間’通常至少需要3〇秒鐘左右，在沒有準備好前 籲如果通過此崎協定位址訪問這個服務，就會造成“服務拒 ，、、邑，進而發生拒絕訪問錯誤。系統會以此給出服務報告錯誤， 因此習知技術無法實現服務的無間斷及透明接管。 【發明内容】 為了解決上述習知技術中的問題與缺陷，本發明之 提供-種多主機系統之服務接管系統及方法，藉以在多主機系統 之提供服務之主機發生故障時，其它運作主機能夠安全、無間斷 籲且透明地接官其網際協定位址及服務，以保證服務的正常運行及 功能的正常使用。 為達上述目的，本發明所揭露之—種服務接管系統，應用於 包含-服務主機及至少-個待命主機之多主機系統中，服務主機 -透過-對外網際協定位址對外提供服務，待命主機處於待命狀 -態，且祕域與絲少-鱗命域之間透如_制互相於 控運作狀態，其中此服務接管系統包含有：一網際協定位址接^ 模組、-服務接管模組及-請求處理池；於此，此網際協定: 址接管模組係藉由心跳機制判斷服務主機之運作狀態，並當服務 7 料包 =發生故料發送—資轉放請麵知服務域職其佔有的 2網際協定健及藝，以鮮服務主機之料網際協定位址 中-待命主機，服聽管模組係用以準備接管服務主機之服 =至此待命主機所需之服務環境，以及接管服務。請求處理模组 糸細i服務接管池之服顧鮮備絲，並且在服務環境未準 借完全之前’去棄透過對外網際協定位址對此服務訪問之請求資 此外’本發日月所揭露之一種服務接管方法，應用於包含一服 機及至^個待命主機之多主機系統中，服務主機盘至少一 個待命主機之間魏㈣_㈤目監控運触n，此方法包含以 :步驟：跡__情服務主機之運作狀態，並當服務主機 發生故障時槪—資轉放請求通知服務域，轉放其佔有的 渺卜網際齡健及服務；鮮服務主機之料提供服務之一對 二際協疋仙:至其巾—待命主機；準備接管服務主機之服務至 t主機所需之服務環境；_服務環境之準備狀態，在服務環 t奉備完全之前丢棄透過對外崎協定位址對此服務之訪問請 ;、貝;斗1 α及在服務％鮮備完全之後接管服務，並接收對此 服務之關請求資料包，輯外提供服務。 ^發日脉多主機系統及_環境下透翻_定位址及服務 执官提供面可用服務時，對於需要時間準備之服務接管，為了保 Λ 1^之域服務在崎協粒址被接管後安全無誤地提供 廿^務—日此在服務接管之鱗備接管服務所f之服務環境， 、；備凡成之則丢棄對服務的請求資料包。進而不斷地偵測服 務環境的準備狀態， 提供服務。 直至準備完权後喊行服務的接管與對外 的二具有以下優點：不僅保證了具有快速接管特性 食’絲料驗賴供服務之料下客戶端 務實酬 【實施方式】 明如7本發日狀倾與實作，脑合_作最佳實_詳細說 請參考「第2圖」’係顯示了本發明之多主機服務接管系統， 多主機系統包含-服務主機及至少-個待命主機，例如「第！圖」 所示之實施例中，多主機系統1〇包含主機12及主機Μ，這裡假 設主機U為服務主機，主機14為一待命主機，且服務主機η與 待命主機14之間透過心跳機制互相監控運作狀態。因此，針對習 知技術之上關題’本剌之多主機服務接管系統包含了一網際 協定位址纽2〇、—服務·额22及—料處_組％: 下文中將對上述各個模組作詳細說明。 本發明之網際協定位址接管模組2〇係用以在當前提供服務 之服務主機12發生轉讀，麟其錄命狀紅針—主機將 此服務主機12之對外網際協定位址1〇 1〇 i 1〇快速接管過去。合 存在多個待命主機時，由哪個待命主機接管服務可以隨機選 擇。任何-鐵多懈命主機可能會細㈣故駐機之故障，因 此這些待命主機都會試圖接管此故障主機的網際協定位址和服 丄 官準備後才能透過所接管之網際協定位址及時、安全地提供服 務因此’服務接管模組22則需要在服務接管之前，準備接管此 故障主機12之服務至待命主機14所需之軟硬體環境。 服務接官桓、组U之接管環境準備隨服務類型的不同而不 ^有^接官服務需要提前準備好接管所需的軟硬體環境，因而 可能非她時’而有些服_不需要準備鮮環境，因此直接快 速地接官。因此，服務接管模組22 ^要對接管服務是否需要環境 準備觸。如果不需要準備，則立即接管此服務；否則，服 務接官彳碰22執行接管服務所f之服務環鮮備。關於接管是否 需娜竟準備，服務接管模組22可透過接管服務類型來判斷，如 果所提供之服務與齡㈣或縣内容相關，例如網路小型電腦 系統介面（ISCSI)、槽案傳送協定（FTp)、超文件傳送協定 r = )、/網路檔案系統（^ )、伺服器訊息區塊/公用網際網路 =糸統（SMB/OPS)料與存間或文件_目_服務， 備服務環境；相反，對於純網路服務，例如純網路 «之_主機組態财⑽cp)、領域名稱服務（卿）等， 服務接讀組22财f要時_備服務财。 、 非=輸咖概她機瓣 醉解f耗時（如料補被其他設備釋放，等 ανΠΐ開頭位置’創建廉價磁碟冗餘陣列（RAID)、邏輯卷 ，、w等），而有麵鮮制f要特—购時時門。 运有一 i準備的服務可能只需要更改配置魅或者更改路 11 對於服務接管概較科，只需要麵啓誠轉本域上之服 務程式即可。 4多主機系統之對外服務不但包含網路小型電_統介面等區 塊叹備訪問魏，也提供了職傳送協定（剛、伺服器訊息區 ,υ、’周際網路擋案系統（smb/cifs )、網路槽案系統（鹏） 等槽案訪問功能。另外，還提供了遠端登人協定（SSH)、終端機 模擬協定（Telnet)、用戶介面（WebuJ)等管理功能，同時提供 2動態主機_協定（DHCP)、領域名稱服務⑽s)等網路功 月匕k些服務大致可以分為兩類：第一類服務如網路小型電腦系 統介面、齡傳送協K司服器訊息區塊/公用網際網路槽案系 統、網路财线等服務’麵服務需要與硬體資源較，例如 網路小型電财、統介面必須是對某個確定的磁盤的操作，而槽案 傳，協^、伺服nm塊/公㈣際網路檔齡統、網路槽案系 統等共用必須基於某個確定的磁盤上的某個目錄。第二類服務如 遠端登人協定、終端_擬協定等管理魏和動態线組態協 定領域名稱服務等網路功能。這類服務基本與硬體資源無關。 只要计异機JL常作’網際協定錄正冑提佩可以提供對外的 服矛力’所以在以上雙糊||轉接管過財，這_服務需要需 要分別處理。 對於第-類服務，除保證故障後的連接性外，還需要保證訪 問的二間與發生故障前訪_空間是相同的。否顧戶訪問空間 會發生交化’導致不能JL常提供服務。所料__類服務在故障切 換前需要先將硬鮮備好’然後才能提供真正的服務。 12 _對於第—類服務，需要條故障發生後的快速連通性，需要 保證在故障發生後不能有明顯的延遲。因為這些服務尤盆是遠端 f入協定、終端機模擬協定、用戶介面之管理服務都與用戶體驗 孩切相關’明顯的延遲會降低用戶的體驗品質。其中對於第一類 T務之故轉料境，由服務·雜22之資源準賴組完 成。資源準備模組24係提供接管該服務之網路連接，以及提供與 =服務主機發纽障之_同之關㈣。射當接管服務為槽 =服務時’料醉備模組储供與該服務域發生故障之前相 1位置之播案存放空間；當接管服務為區塊設備訪問服務時，該 Z職綱她樓瓣之糊—訪問服務區 塊S又備之區塊設備。 7如對於磁碟陣列（DiskArray)之儲存空間準備，資源準 :組24可執行步驟如下：發送指令要求故障主機釋放所佔用的 m如果故障主機還可以活動，則釋放這些硬碟設備，否 潍’”、已、Mt掉’也就不需要釋放硬碟設傷；重新起始化這些硬 t公共=碟郎，__廉價刺冗物彳、卷之組裝 磁碟冗餘陣列之組裝資料將這些硬碟分組組裝成 翔’此時廉價磁碟冗餘陣列還原；根據邏輯卷之 將廉價磁碟冗餘陣_分或啟動為不_邏輯卷，此時 ==對於網路小型_系統介面服務，需要將設備輸出 1/ (Initlat〇r) 5 侧：路槽案系統、網路播案系統服務，將這些設備掛載 (_nt)到指定的目錄，根據組裝資料分不同的廉價磁碟冗餘陣 13 ，此時所 列、邏輯卷全部絲完成，直到所有設備都準備好為止 有的硬體資源才準備完畢。 際協定位址接管模組20及服務接管模組22 ^早主機之位址接如及服務接管前之環鮮備進行了說明。名 協====22係依照_機透過_ 絲確定财要接管的服務，並依·昭 L服務雜的;行相顧轉 =瓣之準備過程中，對應服務之端口被‘= 峨咖咖σ，_絲務被 ^ _客戶端會放棄此服務訪問 及透明==:猶環境準備時，實現服務的無間斷 多主機服務接管系統包含—請求處理模組 接管服務之準=命令調用或函數調用來實現服務環境或 操：::=’：，-個梅，-表 某個磁碟上魏麵些命令結束時於 在。即如果標諸或檔案存在：、龜務經j 則代表環境尚林備好, m麵好’否 局限於上述方法m完成之躺本發明並不 請求處理模㊆26在^_=個目的之方法均可以使用。 口之 過接官之多主機系統之對外網際協定位址訪問“端亚對透 14 求，行處理。在服務環鮮備轉備完全之前，請求處理模組26 :對此服務之關請求資料包。由於關請求在送達至相應服 1端口之㈣|棄’系統不會產生服務被拒絕之回應至客戶端， 各戶端則ϋ為沒有收到回應而繼續重新發送請求重試。 並且在服務接官模組22服務接管環境準備完全之後，接管服 '打開對應服務端口。請求處理模組26同時停止對此服務端 I之訪問糾龍包的丢棄，開始魏發送此端σ之訪問請求資 厂進而只現了服務的對外正常提供。對於其它需要準備時間 ，待接管服務之關，均透過上述方案實職務的正常運行 旎的正當#用。 _.因此’在關服務的客戶端看來，服務實現了錢斷、透明 雖齡問服務㈣間會暫時延缓，但最終服務並沒有中 所，貝料沒有丢失而保證了安全性及可靠性。 法竹㈣ 「第3目」對本㈣之多主、統之服務接管戈 程圖21’此圖為本發明多主齡統之服務接管方法之步驟洁 线㈣Γ鋪胁包含—服務域及至少—餘命主機之多 錢刪過心跳機 其它待命域透過務域發生故障時， 中-待人主旛_, 相測到此故障主機之狀態，因此其 接管。响她 ========_ 服務環境，在準備服務接管環境需花費_定的二，因二= 系2之所有服務在服顧管/_触巾沒有或沒有完全進入到 正常工作的狀態。 r 5裡可以：義系統之所有服務均完全進入到正常工作狀態 沈、者狀怨’即所有類型之服務已接收至上述接管故障主機 網際協疋位址即服務之主機中，並能夠對外提供完整 、正常的 服務，則稱作整個多主機系統進人r就緒1態。反之’如果 t先，於保遵狀恶，則表示整個系統在網路協定位址接管/服 ^接S，中或其它故障切換過程中沒有或沒有完全進入到，/就 、狀態。此外’定義服務〃保護〃狀態是對於某些需要準備接 體環境才能接管之服務，所採取的保護狀態，即在服務環 ^備未完狀前，服務不能接管科能正常提供料服務之狀 ^由於服務不能正f斜提供，因此在客戶端關此服務之請 =達對應服務端π之前，將此訪問請求資料包丢棄^直至服務 :就緒狀態，即在完成服務接管環境準備之後接管服務，並 常料提絲務時的狀態。料停均減服務端口之訪 :山桃的丟棄’開始接收發送此端^之關請求資料包， 只現服務的對外正常提供。 現在請參考「_」，物_命_統之狀態為保護 _)，記錄龄，同時設置待命_統之所有服務 她_ (步驟104)。所有服務為"保護夕狀‘態則可以透過簡 早的默認去棄所有服務請求之方式達到訪問請求處理之处果。上 =態=為本發明之重要部份，下面結合「第4圖」詳細說γ 保獲狀態之系統及服務之請求處理步驟。 16 丄J厶：>δ44 冬—第二圖」為保5蒦狀態服務之訪問請求處理流程示意圖， 各戶端訪問此待命主機系統時，處理客戶端服務訪問請求之流 程^圖所示。接收客戶端發送對某—服務之訪問請求至,/保護„ 狀態系統（步驟202)’判斷此服務是否處於々就緒狀態（步驟 204)。如果服務沒有處於，/就，狀態，即當前為m態 則丢棄對此服務之訪問請求資料包（步驟第）；否則，發送至相 應服務進行處理（步驟208 )。Ms), logical volume (LV) settings are already in place, because it usually takes more time to get involved in hardware preparation. 'It usually takes at least 3 seconds or so. If it is not ready, please call this address through this Sakis agreement. The service will cause "service rejection, 邑, 邑, and then the access denied error. The system will give a service report error, so the conventional technology can not achieve the uninterrupted and transparent takeover of the service. [Summary of the invention] Problems and disadvantages in the prior art, the present invention provides a service takeover system and method for a multi-host system, whereby other host computers can be safe, uninterrupted and transparent when a host serving a multi-host system fails The address is an Internet Protocol address and service to ensure the normal operation of the service and the normal use of the function. To achieve the above purpose, the service takeover system disclosed in the present invention is applied to the include-service host and at least one standby In the host system of multiple hosts, the service host provides external services through the external Internet Protocol address, and the standby host is waiting. The life-state, and the secret domain and the silk-small-sense domain are in a state of mutual control, wherein the service takeover system includes: an internet protocol address interface module, and a service takeover module. And - request processing pool; here, the Internet Protocol: the address takeover module judges the operation status of the service host by the heartbeat mechanism, and when the service 7 material package = the occurrence of the accident is sent - the transfer to the service area It occupies 2 Internet Protocols and Arts, with the service host's Internet Protocol address-standby host, the listening module is used to prepare to take over the service host's service = the service environment required by the standby host, and Take over the service. The request processing module 糸 i 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务 服务The method for service takeover disclosed is applied to a multi-host system including a server and a standby host, and the service host disk has at least one standby host between Wei (four) and (five) monitoring and monitoring n, and the method includes the following steps: Trace __ the service status of the service host, and when the service host fails, the service transfer request notification service domain, transfer its possession of the Internet age and service; one of the services provided by the fresh service host The second international cooperation agreement: to its towel - standby host; ready to take over the service host service to the t host required service environment; _ service environment preparation state, before the service ring t is fully prepared to discard through the external agreement Address access to this service;, Bei; Dou 1 α and take over the service after the service is fully prepared, and receive the request package for this service, provide services outside the series. ^ 发日脉多主机系统和_ In the environment, when the address is provided and the service provider provides the available services, the service is taken over for the time-prepared service. In order to protect the domain service, the service is safely and unconditionally provided after the takeover of the company. In the service environment of the service takeover, the service environment of the service is taken over, and the request packet of the service is discarded. In turn, it constantly detects the readiness of the service environment and provides services. Until the preparation for the right, the takeover of the service and the external two have the following advantages: not only guarantees the customer with the quick takeover characteristics of the food 'wire inspection service for the customer's remuneration [implementation] Deformation and implementation, brain integration _ for best practice _ please refer to "Figure 2" for details. The system shows the multi-host service takeover system of the present invention. The multi-host system includes - the service host and at least one standby host. For example, in the embodiment shown in the "FIG. Map", the multi-host system 1 includes the host 12 and the host port. Here, it is assumed that the host U is a service host, the host 14 is a standby host, and the service host η is connected to the standby host 14 Monitor the operational status of each other through a heartbeat mechanism. Therefore, for the above-mentioned topic of the above-mentioned technology, the multi-host service takeover system of Benedict contains an Internet Protocol address, 2, service, amount 22, and material information group %: The group is detailed. The internet protocol address takeover module 2 of the present invention is used for the transfer of the service host 12 currently providing the service, and the host has recorded the red needle of the service. The host has the external internet protocol address of the service host 12 of 1〇1. 〇i 1〇 quickly took over. When there are multiple standby hosts, which standby host takes over the service can be randomly selected. Anything that is too much for the host may be fine (four), so the host will try to take over the Internet address of the failed host and the service provider can prepare it in time and safely through the Internet Protocol address that is taken over. Providing the service Therefore, the service takeover module 22 needs to prepare to take over the service of the failed host 12 to the soft hardware environment required by the standby host 14 before the service takes over. Service takeovers, group U takeover environment preparations vary with the type of service, and there is no need to prepare for the takeover of the required software and hardware environment, so it may not be her and some services are not required Fresh environment, so directly and quickly pick up the official. Therefore, the service takeover module 22 has to be ready to touch the takeover service. If you do not need to prepare, you will take over the service immediately; otherwise, the service pick-up officer will perform the service ring of the takeover service. Regarding whether the takeover needs to be prepared, the service takeover module 22 can be judged by the type of takeover service, if the service provided is related to the age (4) or county content, such as the network small computer system interface (ISCSI), the slot transfer protocol ( FTp), super file transfer protocol r = ), / network file system (^), server message block / public internet = SMB / OPS material and storage or file _ directory _ service, preparation Service environment; on the contrary, for pure network services, such as pure network « _ host configuration (10) cp), domain name service (Qing), etc., service access group 22 financial time. , non = loss of coffee, her machine is drunk, f is time consuming (such as material supplement is released by other devices, etc. at the beginning of ανΠΐ location to create a redundant array of cheap disks (RAID), logical volume, w, etc.) Fresh f must be special - when buying time. If you have a service that you have prepared, you may only need to change the configuration or change the way. 11 For the service takeover profile, you only need to turn the service program on this domain. The external services of the 4 multi-host systems not only include the network small _ _ interface and other block sighing access Wei, but also provide the job transfer agreement (just, server message area, υ, 'internet network file system (smb /cifs), network slot system (Peng) and other slot access functions. In addition, remote access protocol (SSH), terminal emulation protocol (Telnet), user interface (WebuJ) and other management functions are also provided. Provide 2 dynamic host _ agreement (DHCP), domain name service (10) s) and other network power 匕 k Some services can be roughly divided into two categories: the first type of services such as network small computer system interface, age transmission association K server The service block/public internet slot system, network finance line and other services need to be compared with hardware resources, such as network small money, the interface must be the operation of a certain disk, and the slot The case, the cooperation, the servo nm block / the public (four) network age system, the network slot system and other sharing must be based on a certain directory on a certain disk. The second type of services are network functions such as remote access protocol, terminal_protocol agreement, etc., which manage Wei and dynamic line configuration agreement domain name services. This type of service is basically independent of hardware resources. As long as the JL machine is often used as an 'internet agreement, you can provide external spears.' So in the above double paste||transfer tube, this service needs to be handled separately. For the first-class service, in addition to ensuring the connectivity after the failure, it is also necessary to ensure that the two accesses are the same as the pre-access _space. No visitor space will occur. This will result in the inability of JL to provide services. It is expected that the __ class service needs to be prepared before the failover, and then the real service can be provided. 12 _ For the first type of service, fast connectivity after the occurrence of a fault is required, and it is necessary to ensure that there is no significant delay after the failure occurs. Because these services are remote access protocols, terminal emulation protocols, and user interface management services are all related to the user experience. The obvious delay will reduce the user experience quality. Among them, for the first type of T-transfer, the resource is completed by the resources of the service and miscellaneous. The resource preparation module 24 provides a network connection to take over the service and provides the same as the service host (4). When the takeover service is slot=service, the material storage space of the location is 1 before the failure of the service domain; when the takeover service is a block device access service, the Z grade is her building. Paste paste - access to the service block S and the block device. 7 For the storage space preparation of the disk array (DiskArray), the resource quasi-group 24 executable steps are as follows: sending the command requires the faulty host to release the occupied m. If the faulty host can also be active, the hard disk device is released, otherwise '', has, Mt off' does not need to release the hard disk set damage; reinitialize these hard t public = disc Lang, __ cheap stabs, the assembly of the redundant array of assembled disks will These hard disk groupings are assembled into a 'redundant disk redundant array restore at this time; according to the logical volume, the cheap disk redundant array is divided or started as a non-logical volume, at this time == for the network small_system interface Service, you need to output the device 1 / (Initlat〇r) 5 side: road slot system, network broadcast system service, mount these devices (_nt) to the specified directory, according to the assembly data, different cheap disk Redundant array 13, at this time, the logical volumes are all completed, until all the devices are ready, and some hardware resources are ready. The agreement address takeover module 20 and the service takeover module 22 Address and service takeover The former ring is freshly prepared. The name association ====22 is based on the _ machine through _ silk to determine the services to be taken over by the financial, and according to the Zhao L service miscellaneous; line care = the preparation of the valve, corresponding The port of the service is '= 峨 咖 咖 、, 丝 丝 ^ ^ _ client will give up this service access and transparent ==: When the environment is ready, the service of the non-stop multi-host service takeover system contains - request processing module Take over the service = command call or function call to implement the service environment or operation:::=':, - a plum, - table on a disk on the surface of some of the commands at the end of the command. That is, if the standard or file exists :, turtle service j, on behalf of the environment is still ready, m surface is good 'no limited to the above method m to complete the lying. The invention does not request to deal with the model VII 26 in ^_= a purpose of the method can be used. The external Internet protocol address of the host system is accessed by the end of the network. Before the service ring is fully prepared, the request processing module 26: requests the data package for this service. Since the off request does not generate a response to the client that the service is rejected, the client will continue to resend the request and retry. And after the service receiving module 22 service takeover environment is ready, the takeover service 'opens the corresponding service port. The request processing module 26 simultaneously stops the discarding of the access token of the server I, and starts to send the access request resource of the terminal σ to further provide the normal external provision of the service. For other requirements that require preparation time and to be taken over by the above-mentioned program, the normal operation of the position is used. _. Therefore, in the view of the client of the service, the service has achieved money and transparency. Although the service is temporarily delayed, the service is not temporarily delayed, but the final service is not lost. The safety and reliability are guaranteed without loss of material. . Fa Zhu (4) "3rd Head" to take over the service of the main and unified services of this (4) Gou Tu 21' This figure is the step of the service takeover method of the multi-owner system of the present invention. (4) Γ Γ Included - service domain and at least - When the rest of the host computer has deleted the other standby domain of the heartbeat machine, when the service domain fails, the middle-waiting owner _, the status of the faulty host is measured, so it takes over. Ringing her ========_ service environment, in the preparation of service to take over the environment, it takes _ fixed two, because the second = system 2 all services in the service / _ touch towel did not or did not fully enter the normal work status. r 5 can: all the services of the system are completely in the normal working state, and the complaints are that all types of services have been received into the host that takes over the faulty host network address, that is, the service, and can provide external A complete, normal service is called the entire multi-master system. On the other hand, if t is first, if it is in compliance with the law, it means that the whole system has not or does not fully enter, / state, in the network protocol address takeover/service S, S or other failover process. In addition, the 'definition service 〃 protection 〃 status is the protection status that can be taken for some services that need to be prepared to take over the environment. That is, the service cannot take over the service of the service normally before the service ring is completed. Since the service cannot be provided in the positive direction, the access request packet is discarded until the service: ready state, that is, after the completion of the service takeover environment, the service is taken over before the client closes the service to the server π. And often the state of the wire. Material stop and decrease service port visit: Shantao's discarding 'starts to receive the request packet for sending this terminal ^, only the service is normally provided externally. Now please refer to "_", the status of the object_命_ is the protection _), record the age, and set all the services of the standby__ (step 104). All services are "protective eves' can achieve access request processing by means of a simple default to abandon all service requests. The upper = state = an important part of the invention. The following is a detailed description of the system and service request processing steps of the gamma-receiving state in conjunction with "fourth figure". 16 丄J厶:>δ44 Winter—Second Diagram” is a schematic diagram of the access request processing flow for the security service, and the process of processing the client service access request when each client accesses the standby host system is shown in the figure. The receiving client sends an access request to the service to /, the protection status system (step 202) determines whether the service is in the ready state (step 204). If the service is not in, the status is currently m The state discards the access request packet for this service (step); otherwise, it sends it to the corresponding service for processing (step 208).
對於m態之服務訪問請求包丢棄具有多種實現方式， 在Unix/Limix平台最簡單的方法是使用响㈣/此職沈實現，例 如下面的齡可㈣棄所有發往“網路小型電腦系統介面，，服務 的請求。 #iptables -A INPUT -p tcp -dport 3260 -j DROP，其中 3260 是 iSCSI的服務埠。 對於非保護〃狀態服務，即服務處於〃就緒〃狀態時，則 φ 取‘消對此服務之訪問請求丟棄操作，即取消對服務的保護，而是 - 要求服務處理此訪問請求。例如取消訪問請求包丟棄之指令為： #iptables -D INPUT -p tcp -dport 3260 -j DROP #iptables -A INPUT -p tcp -dport 3260 -j ACCEPT 上面兩個過程將服務之，，保護〃狀態去除，使系統可以接收 並處理上個步驟丟棄的發往〃網路微型電腦系統介面〃之服務請 求。 17 丄JZJO畔斗 :要心出的κ ’這裡只是給出實現上述操作之—概括實 並非用以蚊本發私保魏圍，㈣纽 知 均可以應祕本發日种。 代U技術 際協德址（步驟廳）。關於網際協定位址之 t 術’例如可參見開源項目_，u酿Virtuals叫 =際協定接管之實現代喝。然後，則可以接管各個服務。李統 &供了多個對外服務，對於不需要 … =贿务，系統立刻可以提供。因此，判斷需接管:服務是： utrr管魏料（步_)，若不f要，_卩接管此 〜11G)。例如提供管理魏及提供網路魏之服務，這 ;=本細__ ’麵_健增供後即能夠 務_=步驟1G8巾疋否需要準備服務接f環境，可依據接管服 _⑻心。觸’如果所提供之服務與儲存空間紐案内容相關， =網路小型電腦系統介面（ISCSI)、齡傳送齡（阳）、超 /文讀送協定（HTTP)、網频㈣、統⑽）、恤器訊息區塊 :用網際剩t㈣統（SMB/CIFS)料與存儲如或文件内 二相關的服務’則需要時間準備服務環境；相反，對於純網路服 、^例如純網路服務之動態主機組_定⑽⑻、領域名稱服 力DNS )等則不需要時間準備服務環境。 剞· ^文所达’對於某些與硬體資源約定之服務，例如網路小 mi面；}#案傳达協定等，由於需要服務接管環境準備 18 因而不能立即提供， 環境準備步驟（步驟 詳細說明。 因此前進至辣112，執行服務接管之資源 U2)。關於環境準備的具體步驟下面將給出 不同服務接管之環境準備時，接管環境準備隨服務類型的 關而右^非常耗時的軟硬體準備主要是與硬魏等待時間有 可==備_等待一個超糊 务要更魏置_或者更改路彳轉。騎料接管則比 乂合易’只需要鱗鶴或機本主機上之服務程式即可。 對於需要與硬體資源約定之服務，除保證輯後之接管服 摘路連馳外，還需要賴關之郎與發生故障前訪問触 P早主機空間是_的。否_戶訪問㈣會發生變化，導致不能 正常提供服務。所以此舰務在輯城”要先將硬體準2 好，然後才能提供真正的服務。當與硬體資源約定之接管服務為 檔案服務時，需提供與服務域發纽障之前_位置之樓^存 放空間；當接管服務為區塊設備訪問服務時，則準備與服務主機 發生故障之前為同一訪問服務區塊設備之區塊設備。 在接管服務之資源環境準備完成之後接管這個服務（步驟 114) ’然後此服務才進入“就緒，，狀態（步驟n6)，並正常提供 對外服務（步驟120)。這些接管耗時的服務在進行資源準備時的 時間雖然很長，但因為一直處於系統及服務的雙重“保護，，狀態 下，系統對透過網際協定位址訪問這個服務之所有請求，即網際 協定資料包都會被丟棄掉，所以不會產生服務拒絕的消息，而客 戶端也會不斷重試這種服務。參見「第4圖」之“保護，，狀態服 19 1323844 務凊求處理流程示意圖。在這種情況下，無論是否需要時間準備 的服務都將得到很好的接管。 在任一服務進入就緒狀態（步驟116)後，會同時判斷 * 是否還有處於“保護”狀態之服務（步驟118)，若沒有，則整個 系統设置為就緒狀態（步驟122);否則，前進至步驟1〇8， 對其匕處於保護〃狀態之接管服務執行步驟丨⑽至步驟122。 重複上述步驟，直至完成所有服務之接管，使其均處於A就緒,， 籲 狀二即整個糸統處於就緒狀態，因此這時對服務之訪問請 求將按,¾弟5圖」之就緒”狀態服^務請求處理流程示意圖之 流程進行處理。 如第5圖」所示，主機系統接收對服務端口之訪問請求（步 驟3〇2)，並絲錢至城服_此關請求進行處理（步驟 3〇=。這也是正常狀態的處理流程，系統在大部分時間處於這個 狀態:，此時ί有請求資料包被丟棄的處理。—旦整個系統設置為 者☆狀Ί上述網際協定接管步驟、服務接管及訪問請求資 料包丢棄之步驟則不再使用，岐由接管主機自動處理，直到再 次發生轉婦時由上述步驟互相作収成安全的故障切換。 j過上述說明可知，本發明不健證了具有快物換特性的 11^快輕供’同時保證了不驗速提供服務情況下客戶端 ==連財中斷。此外，不僅保證了·速提供服務情 _於個1緒錢時提供服務，絲證了將錄麵的服務隼 令於一個⑧可祕主機,㈣上時多餘務的可靠性。在以上= 20 下執打服務輯接管時，能夠帶給用戶 的多服務系統及良好的用戶體驗。 >、的無_、透明切換 阳定=發:以前述之較佳實施方式揭露如上，然其並非用以 之V二= 或之技術人員應當意識到在不脫離本發明_ 。月專利補所揭*之本發明之範圍和精神之情況下 閏=屬本發明之專利保護範圍之内。關於本發明5 '、瘦乾圍明芩考所附之申請專利範圍。 " 【圖式簡單說明】 第/圖為一典型高可用性雙主機叢集系統之結構示意圖； 弟2圖為本發明之多域服務接管系統； =3圖為本翻之多域服務接管方法之步驟流程圖； 第^圖為]保護”狀態服務之訪問請求處理流程圖；以及 第5圖為〃就緒〃狀態服務之訪問請求處理流程圖。 【主要元件符號說明】 10主機系統 12 主機 14 .主機 20網際協定位址接管模組 22 服務接管模組 24資源準備模組 26 請求處理模組 步驟102設置系統為保護狀態 步驟104設置所有服務為保護狀態 21 1323844 步驟106接管公共網際協定位址 步驟108是否需要準備服務接管環境？ 步驟110立即接管此服務 步驟112執行服務接管之資源環境準備 步驟114接管此服務 步驟116設置此服務為就緒狀態 步驟118是否仍存在保護狀態之服務？ 步驟120提供對外服務 步驟122設置系統為就緒狀態 步驟202接收到服務訪問請求 步驟204服務是否處於就緒狀態？ 步驟206丟棄服務之訪問請求資料包 步驟208發送至相應服務進行處理 步驟302接收到服務訪問請求 步驟304發送至相應服務進行處理 22For the m-mode service access request packet discarding has multiple implementations, the easiest way to use the Unix/Limix platform is to use the ring (four) / this job sink implementation, for example, the following age can be (four) discard all sent to the "network small computer system Interface, service request #iptables -A INPUT -p tcp -dport 3260 -j DROP, where 3260 is the service of iSCSI. For non-protected state service, ie when the service is in the ready state, then φ takes ' The access request discard operation for this service, that is, the protection of the service is canceled, but - the service is required to process the access request. For example, the instruction to cancel the access request packet discard is: #iptables -D INPUT -p tcp -dport 3260 -j DROP #iptables -A INPUT -p tcp -dport 3260 -j ACCEPT The above two processes will remove the protection, protection status, so that the system can receive and process the network to the network computer system that was discarded in the previous step. Service request. 17 丄JZJOBandou: The heart of κ' is just to give the above operation - the general is not used for mosquitoes and private insurance Wei Wei, (4) New knowledge can be secrets On behalf of the U-Technology Deco site (step hall). About the Internet Protocol address t technology 'for example, see the open source project _, u brew Virtuals called the international agreement to take over the generation of drinking. Then, you can take over each Service. Li Tong & for a number of external services, for the need to ... = bribery, the system can provide immediately. Therefore, the judgment needs to take over: the service is: utrr tube Wei (step _), if not f, _卩Take this ~11G). For example, provide management Wei and provide services for the network Wei, this; = this __ 'face _ health increase after the supply can be _ = step 1G8 towel 需要 no need to prepare services to f environment, Can be based on the take-up service _ (8) heart. Touch 'If the service provided is related to the storage space content, = network small computer system interface (ISCSI), age transfer age (yang), super / text read delivery agreement (HTTP) , network frequency (four), system (10), the message block: use the network remaining t (four) system (SMB / CIFS) and storage or file related to the second two services - then take time to prepare the service environment; on the contrary, for pure network Service, ^ such as pure network service dynamic host group _ set (10) (8), domain name Service DNS), etc., does not require time to prepare the service environment. 剞·^文所达's services for certain hardware resources, such as network mini-mi-face;}# case communication agreement, etc., due to the need for service takeover Environmental preparation 18 is therefore not immediately available, environmental preparation steps (steps are detailed. So proceed to Spicy 112, perform service takeover resource U2). Specific steps for environmental preparation will be given below when taking the environmental preparation for different service takeovers, taking over the environment Prepare for the type of service with the right ^ very time-consuming software and hardware preparation is mainly with the hard Wei waiting time can be == standby _ waiting for a super paste to be more _ or change the way. It is only necessary to use the service program on the scale crane or the machine host. For the services that need to be agreed with the hardware resources, in addition to ensuring that the take-over service after the series is picked up, it is necessary to use the Laiguan Lang and the pre-failure access to the P host space. No _ Household access (4) will change, resulting in the failure to provide services. Therefore, this warship must be in the first place to provide the real service. When the takeover service agreed with the hardware resource is the archive service, it is necessary to provide the service domain before the barrier is issued. Floor ^ storage space; when the takeover service is a block device access service, it is prepared to be the same block device as the service access block device before the service host fails. Take over the service after the resource environment preparation for the takeover service is completed (step 114) 'The service then enters the "Ready," state (step n6) and provides external services normally (step 120). These time-consuming services take a long time to prepare resources, but because they are always in the dual "protection of systems and services, the system accesses all requests for access to the service through the Internet Protocol address, that is, the Internet Protocol. The data packet will be discarded, so there will be no message of service rejection, and the client will continually retry the service. See "4," "Protection, State Service 19 1323844. In this case, the service will be well taken regardless of whether or not the time is required to be taken. After any service enters the ready state (step 116), it is simultaneously judged whether * there is still a service in the "protected" state (step 118). If not, the entire system is set to the ready state (step 122); otherwise, proceed to step 1〇8, and perform the steps 10(10) to 122 for the takeover service in the protected state. Repeat the above steps until the completion The takeover of all services, so that they are all in A, the second is that the entire system is in a state of readiness, so the visit to the service at this time The request will be processed according to the process of the process flow diagram of the ready-to-use service request processing. As shown in FIG. 5, the host system receives an access request to the service port (step 3〇2), and Silk money to the city service _ this request to process (step 3 〇 =. This is also the normal state of the process, the system is in this state most of the time:, at this time ί there is a request for the packet is discarded. - the whole The system is set to ☆ The above steps of the Internet Protocol takeover step, service takeover and access request packet discarding are no longer used, and are automatically processed by the takeover host until the next step occurs. Failover. j After the above description, the present invention does not prove that the 11^ fast and light supply with the fast change feature ensures that the client is not interrupted by the service. · Speed to provide service _ _ provide services when the money is 1 semester, the silk card will be ordered to serve on a 8 secret host, (4) the reliability of the redundant service. Under the above = 20 When the takeover takes over, it can bring the user a multi-service system and a good user experience. >, no _, transparent switching yangding = hair: the above preferred embodiment discloses the above, but it is not used for V two The technical staff of the present invention should be aware that it is within the scope of the invention of the present invention without departing from the scope and spirit of the present invention. The scope of the patent application attached to the examination of the company is as follows: "Simple description of the diagram] The first diagram is a schematic diagram of the structure of a typical high-availability dual-host cluster system; the second diagram is the multi-domain service takeover system of the present invention; 3 is a flow chart of the steps of the multi-domain service takeover method; the second figure is a flow chart of the access request processing of the protection status service; and the fifth figure is a flow chart of the access request processing of the ready status service. [Main component symbol description] 10 host system 12 host 14 host 20 internet protocol address takeover module 22 service takeover module 24 resource preparation module 26 request processing module step 102 set system to protect state step 104 set all services as Protection Status 21 1323844 Step 106 Takes over the Public Internet Protocol Address Step 108 Do you need to prepare a service takeover environment? Step 110 immediately takes over the service Step 112 Performs resource takeover for service takeover Step 114 takes over the service Step 116 sets the service to the ready state Step 118 Is there still a protected state service? Step 120 provides an external service. Step 122 sets the system to a ready state. Step 202 receives a service access request. Step 204 Is the service in a ready state? Step 206 discards the service access request data packet. Step 208 sends the corresponding service to the processing. Step 302 receives the service access request. Step 304 sends the corresponding service to the processing.