Method for counting instructions for logging and replay of a deterministic sequence of events

• US 7,774,647 B2
• Filed: 01/24/2006
• Issued: 08/10/2010
• Est. Priority Date: 01/28/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Vertes, Marc P.
• Primary Examiner(s): Maskulinski; Michael C
• Application Number:
  US11/814,302
• Publication Number:
  US 20090119549A1
• Time in Patent Office: 1,659 Days
• Field of Search: None
• US Class Current: 714/15
• CPC Class Codes: G06F 11/3414 Workload generation, e.g. scripts, playback
• CPC Class Codes: G06F 11/3433 for load management allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029
• CPC Class Codes: G06F 11/3476 Data logging G06F11/14, G06F11/2205 take precedence
• CPC Class Codes: G06F 2201/88 Monitoring involving counting

Abstract

This invention relates to a transparent and non-intrusive method for monitoring and managing the running of tasks executed in one or more computer processors, in particular in multi-processor systems with a parallel architecture. It proposes a system and method for managing a computer task, termed target, during a given execution period, termed activity period (SchJ, SchR), within a computer system, in a computer processor provided with means of monitoring or estimating performance and including a counter (PMC) with a given possible error in plus or minus, termed relative error, this process comprising

• • on the one hand, an evaluation of a number of executed instructions (NR, NJ) up to at least one given point of said activity period, using said counter; and
  • on the other hand, a generation of data, termed signature (SGJ, SGR), read or calculated from the state of the processor or computer system and corresponding to at least one given point of said activity period.

Claims

1. A method for managing a computer task, termed target, during a given execution period, termed activity period, within a computer system, in a computer processor provided with means of monitoring or estimating performance and including a counter with a given possible error in plus or minus, termed relative error, this process comprising:

an evaluation of a number of executed instructions up to at least one given point of said activity period, using said counter; and

a generation of data, termed signature, determined from the state of the computer system and corresponding to at least one given point of said activity period;

wherein the method logs the use of a processor by storing in an ordered fashion, in at least one file termed processor log, logging data representing the logging of a succession of activity periods of a plurality of tasks executed in said processor, said logging data comprising for each of the activity periods a datum identifying the executed task; and

wherein the method carries out, from the logging data, a replay of the logged period by managing a task, termed replayed task, executed by a processor, termed replay processor, within a replay computer system, starting from a state corresponding to a state of a logged system at the start of the logged period, the method also comprising;

from the start of the replay period, a monitored execution phase executing a number of instructions, evaluated by the counter, less than or equal to the supposed value of the number of instructions of the logged period, reduced by the relative error of said counter;

a confirmation phase comprising an iteration of the following steps;

interruption of the execution of the replay task at a given point in the replay; and

comparative test between the logged signature and the signature corresponding to the replay interruption point.

2. The method according to claim 1, wherein the logged signature includes a datum termed log pointer, representing the value of the instruction pointer of the logged task at the end of the logged period,

the method also comprising a setting up of a breakpoint on the replay instruction corresponding to the logged pointer datum.

3. The method according to claim 1, wherein the monitoring of the execution of the replay task comprises an overflow of the counter, initialised beforehand in order to determine a number of instructions having to be executed from the start of the replay period and the overflow of which causes an interruption of the replay task.

4. The method according to claim 3, wherein the counter is initialised at the start of the replay period with a certain margin, in order to overflow sufficiently early in order that, in case of a latency period existing between the overflow of the counter and the interruption of the task which has caused it, the number of instructions executed by the replay task is less than or equal to the supposed value of the number of instructions of the logged period, reduced by the relative error of said counter.

5. The method according to claim 1, wherein the confirmation phase comprises a security step signalling an error if the number of replayed instructions exceeds the supposed value of the number of logged instructions increased by a given number of instructions.

6. The method according to claim 1, wherein the method carries out, in a replay processor, a replay of the use of a logged processor, by using an ordered set of logging data of a succession of activity periods logged in said logged processor.

7. The method according to claim 1, wherein the method carries out a logging of a succession of attributed exclusive accesses pertaining to a shared resource, termed target resource, accessed by a plurality of logged tasks, this method also transmitting or storing at least one file termed resource log, comprising at least one identification of the successive tasks having obtained these exclusive accesses.

8. The method according to claim 7,

wherein the method manages the functioning of at least two application tasks,

implemented within a system software managing by sequential activation the execution of said tasks in a computer system, endowed with a parallel structure comprising calculation means capable of executing a number of application tasks simultaneously in at least two arithmetic units, these two application tasks accessing to at least one shared resource, the method comprising:

a logging of a first succession of activation periods of one of these tasks in a first arithmetic unit; and

a logging of a second succession of activation periods of one of these tasks in a second arithmetic unit;

and a logging of a succession of attributions, to a so-called accessing task among said tasks in response to a request for access to said target resource, of an access termed exclusive to said target resource, with such an attribution excluding any access to said target resource by another of these tasks during the entire rest of the activation period of the accessing task immediately after said request for access;

the method also comprising a combination, in an ordered structure termed replay serialisation, of logging data representing successions of activation periods in each of the arithmetic units, combined with logging data representing the succession of attributed exclusive accesses, so as to maintain the order of succession of the activation periods within each task and vis-à

-vis said shared resource.

9. The method according to claim 8, wherein the replay serialisation data is used in a replay computer system for replaying the logging of the successions of activation periods of the tasks in the first and second arithmetic units.

10. The method according to claim 9, wherein the method comprises a virtualisation, within the replay computer system, of at least part of the software resources accessible, during the logging, to the tasks logged.

11. The method according to claim 9, wherein the method carries out a replication, termed active, of the functioning of a logged application comprising at least two tasks, executed over at least one node with parallel structure, termed primary node, of a network of computers and accessing at least one shared resource, this replication comprising a replay, in at least one replay application over a replay system, of a replay serialisation originating from logging data transmitted on-the-flow from the primary node to the secondary node when generated.

12. The method according to claim 11, wherein the method carries out a reliabilization of an application comprising at least two tasks, executed over at least one node with parallel structure, termed primary node, of a network of computers and accessing at least one shared resource, this reliabilization also comprising a switching of service, from the primary node to at least one secondary node instead of the primary node, initiated on detection of a failure or of a given event within the primary node.

13. The method according to claim 11, wherein the method carries out an adjustment of the work load within a network of computers executing over at least one secondary node an active replica of a target application executed over a primary node, this load balancing comprising a switching to the active replica of at least part of the service provided by the managed target application.