System and method for dynamically varying low level file system operation timeout parameters in network systems of variable bandwidth

• US 5,699,511 A
• Filed: 10/10/1995
• Issued: 12/16/1997
• Est. Priority Date: 10/10/1995
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Waldron, Theodore Clayton III, Yellepeddy, Krishna Kishore, Ward, Richard Byron, Porcaro, Thomas Joseph
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Palys, Joseph E.
• Application Number:
  US08/540,431
• Time in Patent Office: 798 Days
• Field of Search: 395/185.08, 395/182.02, 395/184.01, 395/200.13, 395/200.19, 395/200.11, 395/200.12, 395/550, 395/186, 395/187.01, 395/559, 395/182.22, 371/33, 371/32, 371/62, 371/61
• US Class Current: 714/55
• CPC Class Codes: H04L 43/00 Arrangements for monitoring or testing data switching networks

Abstract

An improved file system apparatus and method for minimizing the length of time a client system waits before declaring a data communication link disconnected. The apparatus and method dynamically modify a file system request time-out value based on the actual length of time required to service each file system request. In one embodiment, a time-out value is determined for each request type based on the actual response time and a buffer time for each request type. The response timer is based on readings from a system clock therefore operating as a low overhead process. A monitoring system periodically tests the server to ensure that a physical connection still exists.

Claims

1. A computer implemented process for detecting network failure while avoiding false detection in a network system connecting a source device to one or more target devices, said network system operable over any one of a plurality of communication links each having variable communication bandwidth and being subject to intermittent non-failure disconnection, the process comprising the steps of:

receiving a minimum and a maximum time-out value for each user of one of said one or more target device;

setting said network service request time-out period equal to said maximum time-out value for said one of said one or more target devices;

repeating the following steps for each of a plurality of network service requests to said one of said one or more target devices;

issuing a network service request over said communications link;

signalling network failure if said network service request is not satisfied by said target device within said time-out period;

measuring network service request time if said network service request is satisfied; and

modifying said time-out period for said user in response to said network service request time of each satisfied service request by performing the following steps;

setting said time-out period to the minimum time-out value if said network service request time is less than or equal to said minimum time-out value;

setting said time-out period to the lesser of said network service request time plus a service request buffer interval or said maximum time-out value, if said network service request time is greater than said minimum time-out value.

2. The process of claim 1 wherein the source device contains a system clock, and wherein the step of measuring network service request time comprises the steps of:

reading said system clock and storing a first system clock value in a storage area;

reading said system clock to determine a second system clock value upon successful completion of said network service request before the end of the time-out period; and

determining network service request time as the difference between said second system clock value and said first system clock value.

3. The process of claim 1, wherein the step of signalling network failure comprises the steps of:

initializing an independent timer with said time-out period;

starting said independent timer when said network service request is issued;

cancelling said independent timer if said network service request is satisfied before said independent timer completes the time-out period; and

cancelling the network service request, cancelling said independent timer, and signalling network failure if said independent timer completes the time-out period before the network service request is satisfied.

4. The process of claim 1 wherein said network services request can be any one of a plurality of network service request types and wherein said service request buffer value and said time-out period is stored and applied independently for each of said network service request types.

5. The process of claim 1 wherein said network service requests are low-level file system requests.

6. The process of claim 1, further comprising the step of:

setting the source device to a disconnected state in response to the signalling of network failure.

7. The process of claim 6 further comprising the steps of:

testing said network for connected state prior to issuing a network service request;

periodically testing for connected state during any period in which said source device is in said disconnected state.

8. The process of claim 7, further comprising the steps of:

setting the source device to a quiescent state in response to a target device failure to acknowledge the network service request after a predetermined number of tries; and

sending a signal from said target device to said source device upon reconnection.

9. The process of claim 6 further comprising the step of:

satisfying network service requests from a source device cache when said source device is in said disconnected state.

10. A computer program product for use with distributed computer system connected to a network system, said computer program product comprising:

a computer usable medium having computer readable program code means embodied in said medium for causing detection of network failure with minimal delay while avoiding false failure detection in a network system connecting a source device to one or more target devices, said network system operable over any one of a plurality of communication links each having variable communication bandwidth and being subject to intermittent non-failure disconnection, said computer usable mediums further having;

computer program product means for causing a computer system to receive a minimum and a maximum time-out value for each of said one or more target devices for a user;

computer program product means for causing a computer system to set said network service request time-out period equal to said maximum time-out value of said one of said one or more target devices;

computer program product means for causing a computer system to repeat the following steps for each of a plurality of network service requests to said one of said one or more target devices;

computer program product means for causing a computer system to issue a network service request over said communications link;

computer program product means for causing a computer system to signal network failure if said network service request is not satisfied within said time-out period;

computer program product means for causing a computer system to measure network service request time if said network service request is satisfied; and

computer program product means for causing a computer system to modify said time-out period for said user in response to said network service request time for each satisfied request, said means including computer program product means for causing a computer system to set said time-out period to the minimum time-out value if said network service request time is less than or equal to said minimum time-out value;

computer program product means for causing a computer system to set said time-out period to the lesser of said network service request time plus a service request buffer interval or said maximum time-out value, if said network service request time is greater than said minimum time-out value.

11. The computer program product of claim 10 wherein the source device contains a system clock, and wherein the computer program product means for causing a computer system to measure network service request time comprises:

computer program product means for causing a computer system to read said system clock and storing a first system clock value in a storage area;

computer program product means for causing a computer system to read said system clock to determine a second system clock value upon successful completion of said network service request before the end of the time-out period; and

computer program product means for causing a computer system to determine network service request time as the difference between said second system clock value and said first system clock value.

12. The computer program product of claim 10, wherein the computer program product means for causing a computer system to signal network failure comprises:

computer program product means for causing a computer system to initialize an independent timer with said time-out period;

computer program product means for causing a computer system to start said independent timer when said network service request is issued;

computer program product means for causing a computer system to cancel said independent timer if said network service request is satisfied before said independent timer completes the time-out period; and

computer program product means for causing a computer system to cancel the network service request, cancel said independent timer, and signal network failure if said independent timer completes the time-out period before the network service request is satisfied.

13. The computer program product of claim 10 wherein said network services request can be any one of a plurality of network service request types and wherein said service request buffer value and said time-out period is stored and applied independently for each of said network service request types.

14. The computer program product of claim 10 wherein said network service requests are low-level file system requests.

15. The computer program product of claim 10, further comprising:

computer program product means for causing a computer system to set the source device to a disconnected state in response to the signalling of network failure.

16. The computer program product of claim 15 further comprising:

computer program product means for causing a computer system to test said network for connected state prior to issuing a network service request;

computer program product means for causing a computer system to periodically test for connected state whenever said network during any period in which said source device is in said disconnected state.

17. The computer program product of claim 16, further comprising:

program product means for causing a computer system to set the source device to a quiescent state in response to a target device failure to acknowledge the network service request after a predetermined number of tries; and

computer program product means for causing a computer system to send a signal from said target device to said source device upon reconnection.

18. The computer program product of claim 15 further comprising:

computer program product means for causing a computer system to satisfy network service requests from a source device cache when said source device is in said disconnected state.