Automatic root cause analysis and diagnostics engine

• US 20050120273A1
• Filed: 11/14/2003
• Published: 06/02/2005
• Est. Priority Date: 11/14/2003
• Status: Active Grant

• Current Assignee: ServiceNow Incorporated
• Original Assignee: Microsoft Corporation
• Inventors: Hudson, William Hunter, Meng, Yi, LeVine, Eric, Sharma, Kshitiz K., Pease, Geoff, Fink, Reiner, Vachon, Andre, Bliss, Andrew L., Shan, Jing, Maffeo, Gerald
• Granted Patent: US 7,191,364 B2
• Time in Patent Office: Days
• US Class Current: 714/38
• CPC Class Codes: G06F 11/0748 in a remote unit communicating with a single-box computer node experiencing an error/fault remote testing G06F11/2294
• CPC Class Codes: G06F 11/079 Root cause analysis, i.e. error or fault diagnosis in a hardware test environment G06F11/22; in a software test environment G06F11/36
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: G06F 11/366 using diagnostics G06F11/0703 takes precedence
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

A large portion of software bugs are related to hangs. Analyzing and diagnosing hang-related bugs involves capturing data from a hung program and extracting attributes from the captured data. Extracting attributes from the capture data provides a scheme to determine relevant characteristics of the hang. Moreover, the extracted attributes may be compared to known issues and, based on that comparison, a bug may be classified as known or unknown. Alternatively, triage may be performed on the client computing device in order to determine the potential cause of the hang event. Once the potential cause of the hang event has been determined, troubleshooting steps may be performed on the client computing device to quarantine it. Ultimately, if the hang-inducing bug is known, a user may be provided with a solution to the bug. Alternatively, if the bug is unknown, implementations of the invention send the captured data to be analyzed and fixed by the software'"'"'s provider.

Claims

1. A method of troubleshooting software hangs on a computing device, the method comprising:

capturing data associated with a hang;

extracting attributes associated with the hang; and

comparing the extracted attributes to a database of issues to troubleshoot the hang.

2. The method of claim 1 further comprising:

packaging the captured data into a file; and

assigning the packaged file an identification value for tracking the hang.

3. The method of claim 2 wherein the identification value comprises a hash value associated with a call stack.

4. The method of claim 1, wherein comparing the extracted attributes further comprises:

identifying the hang; and

providing a user with a solution to the hang, if the solution is available.

5. The method of claim 1, wherein capturing data associated with a hang further comprises extending a schema by using a data capture program extension.

6. The method of claim 1, wherein extracting attributes to diagnose the hang further comprises extending an attribute extraction schema through the use of an attribute plugin.

7. The method claim 1, wherein the database of issues comprises data to represent at least one hang event.

8. The method of claim 1, further comprising

performing on the computing device the comparison of extracted attributes to the database of issues;

assigning the extracted attributes a value based on a history of hang events;

determining a potential culprit for the hang event based on the assigned values;

performing troubleshooting steps to quarantine the potential culprit.

9. The method of claim 8, wherein the potential culprit comprises one of a file, module, process, thread, block of code, instruction, or the like.

10. The method of claim 8, wherein performing troubleshooting steps to quarantine the potential culprit comprises renaming a file.

11. The method of claim 8, further comprising updating the history of hang events.

12. A system for troubleshooting software hangs comprising:

a data capture program to capture data associated with a hang on a computing device;

a diagnostic data file to save the captured data associated with the hang;

a packaged file to wrap up the diagnostic data file with at least an identification value to identify the hang;

a packaged file repository operating to receive the packaged file;

a datamining utility operating to extract an attribute from the packaged file into a first attribute structure; and

a bucket database operating to maintain a second attribute structure and to generate a value based on a comparison of the first attribute structure to the second attribute structure.

13. The system of claim 12, wherein the value of the comparison of the first attribute structure to the second attribute structure comprises a known value.

14. The system of claim 13, wherein the known value is a solution to the hang.

15. The system of claim 12, wherein the data capture program comprises Watson technologies.

16. The system of claim 12, wherein the datamining utility comprises a component of the computing device.

17. A method of operating a diagnostic data file from a client computing device, comprising:

upon the occurrence of a hang event associated with a program at the client computing device, initiating a capture event to collect data associated with the hang event;

making a data image of the collected data from the capture event;

mining the data image to extract attributes of the hang event; and

structuring the extracted attributes to a format usable by an analysis tool.

18. The method of claim 17, wherein structuring the extracted attributes to a format usable by an analysis tool comprises generating a decision tree.

19. The method of claim 17, wherein the data image of the collected data comprises a mini-dump.

20. The method of claim 17, further comprising analyzing the extracted attributes to determine whether the extracted attributes identify a known bug:

if so, providing a solution to solve the hang event;

else, if the extracted attributes do not identify a known bug;

sending the data image of the collected data to be diagnosed;

upon discovery of a cause of the hang event, fixing the cause of the hang event; and

updating the analysis tool to look for the cause of the hang event.