Parallel file system and method for multiple node file access

• US 6,023,706 A
• Filed: 07/11/1997
• Issued: 02/08/2000
• Est. Priority Date: 07/11/1997
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: McNabb, Daniel Lloyd, Shmueli, Boaz, Schmuck, Frank B., Wyllie, James Christopher
• Primary Examiner(s): Lintz, Paul R.
• Assistant Examiner(s): Shah, Sanjiv
• Application Number:
  US08/893,865
• Time in Patent Office: 942 Days
• Field of Search: 707/200, 707/8, 707/10, 707/201, 707/202
• US Class Current: 1/1
• CPC Class Codes: G06F 16/1858 Parallel file systems, i.e. file systems supporting multiple processors

Abstract

A computer system having a shared disk file system running on on multiple computers each having their own instance of an operating system and being coupled for parallel data sharing access to files residing on network attached shared disks. Methods are provided for use as a parallel file system in a shared disk environment by use of a scalable directory service for the system with a stable cursor, a segmented allocation map. Dynamic prefetch and cached balance pools for multiple accesses improve the system. Extended file attributes are used for implementation of Access Control Lists in a parallel file system. Improvements to caching and cache performance developments balance pools for multiple accesses. A metadata node manages file metadata for parallel read and write actions. For our system, tokens are used for metadata node selection and identification, and we have enhanced token modes for controlling file size, as well as smart caching of byte range tokens using file access patterns and a byte range lock algorithm using a byte range token interface. Locking techniques reduce the overhead of a token manager which is also used in the file system recovery if a computer participating in the management of shared disks becomes unavailable or failed.

Claims

1. In a system employing a file system for storing file data and having a file structure that includes metadata with distributed communicating computer nodes which have access to the file system, the method comprising:

providing a file system with a file structure residing on one or more disks accessible by multiple file systems running on separate computer nodes, said file system allowing a computation that uses the file structure to be broken up into multiple pieces that can be run in parallel on separate ones of said computer nodes, and

providing a metadata node for each file of the file structure with a mechanism to control metadata in a distributed file system wherein every node which accesses a file and needs to read or write metadata information to the metadata node is accommodated, and wherein the metadata node keeps the information about the file'"'"'s metadata and acts as a smart cache between the disk and all nodes that access the file.

2. In a system employing a file system according to claim 1, wherein in situations when the metadata node ceases to serve as a metadata node during transition to enable smooth operation and recovery, these situations need to be handled and recovery operations including dynamic takeover occurs.

3. In a system employing a file system according to claim 2, wherein of three situations when a metanode ceases to operate as a metanode;

the two are asynchronous and other nodes are not aware of this immediately, as when a metanode fails, or when a metanode closes the file or flushes it from it'"'"'s cache, while third one is synchronous and all the nodes are aware of the takeover as when another node needs to become the metanode for the file.

4. In a system employing a file system according to claim 1, wherein in asynchronous operations, the first node that tries to access an old metanode detects an error as when the node crashed, in which case is gets a communication error, or the old node decided not to be the metanode any more, in which case the node gets an appropriate error from the old metanode, then in both of the asynchronous operation cases the node to become a metanode requests an appropriate token for its use in accessing files from a token manager.

5. In a system employing a file system according to claim 4, wherein

when a request for an appropriate token is received by said token manager from a node to become a metanode, if there is no other metanode, the requester will become the new metanode and other nodes that subsequently try to access the old metanode will also go through the same process, but will fail to acquire an the appropriate token and a query to the token manager will reveal the new metanode.

6. In a system employing a file system according to claim 5, wherein

upon a change in a metanode every computer node of the file system eventually knows whether it was the one to become the new metanode or whether the metanode changed.

7. In a system employing a file system according to claim 5, wherein if a node became a metanode, it reads the most recent metadata from disk, while if a node learns that its metanode has been changed changed, the node will re-send its own metadata to the new metanode using a version number for each such update.