Method and apparatus for distributed locking of shared data, employing a central coupling facility

A method for locking data distributed in a processing complex, said processing complex comprising:

• shared data comprising shared resources;
• one or more processors coupled to all others of said processors and to said shared data:
• and two or more data managers each controlling access to said shared data for one or more transactions;
• said method comprising the steps of:
• a) one of said data managers issuing a lock request for a shared resource having a shared resource name, triggered by a request by a requesting transaction for said shared data, and transmitting said lock request to a tailored lock manager (TLM), said lock request comprising a requested ownership state and said shared resource name;
• b) said TLM using a hashing algorithm to convert said lock request to a request to lock a hash table entry, and transmitting a converted lock request to a system lock manager (SLM);
• c) said SLM checking said converted lock request against an entry in a local hash table in a local one of said one or more processors for local ownership of said entry with said requested ownership state;
• d) if said step of checking indicated local ownership with said requested ownership state, setting an indication of said requested ownership state of said shared resource and granting said lock request;
• e) if said step of checking did not indicate local ownership with said requested ownership state, communicating a message comprising said lock request to a central coupling facility, said central coupling facility responding to said message by:
• i) indicating that said lock request may be granted if a control structure in said central coupling facility so indicates;
• ii) indicating that lock contention exists if said control structure in said central coupling facility so indicates.

A method of recovering from a failure involving a first lock manager function associated with a first data manager function in a processing complex, said processing complex comprising:

• shared data comprising shared resources;
• one or more processors coupled to all others of said processors and to said shared data;
• and two or more data managers, belonging to a sharing group, each controlling access to said shared data for one or more transactions;
• said method comprising the steps of:
• a) storing record data comprising a first resource name for a first resource owned exclusively by said first lock manger in a control structure in a central coupling facility before said failure;
• b ) detecting said failure;
• c) notifying a second lock manager function, belonging to said sharing group, of said failure;
• d) said second lock manager function retrieving said stored record data from said central coupling facility and indicating the setting of a local RETAIN lock for said first resource;
• e) indicating release of ownership of said first resource, by said first lock manager, in said control structure.

A locking apparatus for managing access to shared data by a plurality of transactions said access to said shared data controlled by two or more data managers executing in a processing complex comprising one or more processors, said shared data comprising shared resources, each of said processors coupled to all others of said processors and to said shared data, said apparatus comprising:

• a) a plurality of lock managers (LM's) each executing on one of said processors, each of the LM's responsive to a serialization request, by an associated one of said two or more data managers, to serialize an access attempt for one of said shared resources, each of said LM's comprising:
• i) a tailored lock manager (TLM) means for formulating a locking request for said shared resource in response to said serialization request;
• ii) a system lock manager (SLM) means for providing local and first global locking services, responsive to said locking request, said SLM comprising local hash table means for local indication of status for said shared resource, and global table means for distributed contention resolution;
• b) a structured external storage facility (SES) connected to each SLM, said SES comprising a SES lock manager (SESLM) means for providing second global locking services, said SESLM means comprising global lock table means for global indication of status of said shared resource, and further comprising list means for recording modify information for recovering portions of said shared data;
• c) a wait indicator having a first value if said associated one of said two or more data managers will wait for delayed completion of said serialization request, and having a second value if said associated one of said two or more data managers will not wait for delayed completion of said serialization request;
• d) means for asynchronous notification of delayed completion of said serialization request if said wait indicator has said second value.