Computer event log overwriting intermediate events

• US 20040250182A1
• Filed: 06/04/2003
• Published: 12/09/2004
• Est. Priority Date: 06/04/2003
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Lyle, Stephen B., Bouchier, Paul Henry
• Granted Patent: US 7,143,317 B2
• Time in Patent Office: Days
• US Class Current: 714/723
• CPC Class Codes: G06F 11/0787 Storage of error reports, e.g. persistent data storage, storage using memory protection

Abstract

A service processor for a server system includes an event log that, once full, stores recent events by overwriting events of intermediate age so that the information required to diagnose both cascade errors and hangs are preserved. This contrasts with bottom-up buffers that discard recent events when full and with circular buffers that discard the oldest events when full. The event log can be reset by moving an exception region, that is, a region that is not overwritten by recent events. Alternatively, a partial reset can initialize an exception region (e.g., a bottom-up sublog), while a circular region or sublog continues to operate without being reset.

Claims

1. An event-log system comprising:

an event log for recording events generated by a data-processing system;

an event-log manager that, when said event log is full, retains the oldest events stored in said event log while overwriting intermediate events with more recent events.

2. An event-log system as recited in claim 1 wherein said event-log manager stores said oldest events in an excepted region of said event log and stores said intermediated events and said recent events in an unexcepted region of said event log, said event-log manager storing events cyclically in said unexcepted region.

3. An event-log system as recited in claim 2 wherein said event-log manager can move said excepted region and said unexcepted region so that said oldest events stored in said excepted region before said move are then in said unexcepted region so that they can be overwritten by further recent events.

4. An event-log system as recited in claim 1 wherein said event log includes a bottom-up sublog and a circular sublog.

5. An event-log system as recited in claim 4 wherein said event-log manager partially resets said event log by resetting said bottom-up sublog while not resetting said circular sublog.

6. An event-log system as recited in claim 4 wherein said event-log manager dedicates an address space to said circular sublog that is larger than the capacity of said circular sublog so that each event storage location of said event sublog is assigned different addresses at different times.

7. An event-log system as recited in claim 6 further comprising system-management software that asserts an address dedicated to said circular sublog, said system-management software, in the case it is informed said address is not currently assigned to an event-storage location in said circular sublog, determining that event logs have been sent to said circular sublog and have subsequently become inaccessible.

8. An event-log method for recording events generated by a data-processing system, said method comprising:

filling an event log with old events and intermediate events; and

as recent events are received, retaining said old events while overwriting said intermediate events with said recent events.

9. An event-log method as recited in claim 8 wherein said old events are stored in an excepted region of said event log and said intermediate and recent events are stored in an unexcepted region of said event log, said recent events being writing to said unexcepted region in a circular manner.

10. An event log method as recited in claim 9 further comprising:

moving said excepted region and said unexcepted region so that said old events are in said unexcepted region; and

overwriting said old events with said recent events while retaining intermediate events.

11. An event-log method as recited in claim 8 wherein said old events are stored in a bottom-up sublog and said intermediate events and said recent events are stored in a circular sublog.

12. An event-log method as recited in claim 11 further comprising partially resetting said event log by resetting said bottom-up sublog while not resetting said circular sublog.

13. An event-log method as recited in claim 11 further comprising system management software addressing logs within and without said circular log.

14. An event-log system as recited in claim 13 wherein said system management software determines whether event logs had been sent to the log and had subsequently become inaccessible due to being pushed out of the circular log before they had been read by SMS.

15. An event-log system comprising:

a circular event log having a number of event log locations;

a log manager for managing said circular event log, said log manager dedicating an address space to said circular event log manager with more addresses than said number of event log locations, said log manager assigning different addresses at different times to each of said event log locations.

16. An event-log system as recited in claim 15 further comprising system-management software that transmits addresses to said log manager, said log manager, when it receives an address dedicated to said circular event log but not currently assigned to one of said log locations, returning a “

not available”

indication to said software, said software using said “

not available”

indication to determine that an event formerly stored in said log has been overwritten by some more recent event.