Method, electronic device, and computer program product for training failure analysis model

1. A method for training a failure analysis model, the failure analysis model being used to determine a failure cause of a storage system, and the method comprising:

obtaining at least one set of log files including multiple preprocessed log files, the at least one set of log files including a marked failure cause of the storage system, and preprocessed log files in the multiple preprocessed log files including one or more potential failure causes of the storage system and scores associated with the potential failure causes;

predicting a failure cause of the storage system according to the failure analysis model and based on the potential failure causes and the scores in the multiple preprocessed log files; and

updating parameters of the failure analysis model based on a probability that the predicted failure cause is the marked failure cause;

wherein each of at least a subset of preprocessed log files comprises a plurality of fields for respective ones of a failure type, a log file type, an identifier of a storage node of the storage system, and a score for the corresponding preprocessed log file; and

wherein a failure feature is determined for use in the failure analysis model at least in part by aggregating the scores for multiple ones of the preprocessed log files based on one or more of the failure type, the storage node identifier and the log file type.

2. The method according to claim 1 , wherein predicting a failure cause of the storage system comprises:

aggregating the scores in the multiple preprocessed log files based on failure types of the potential failure causes to generate failure features, the failure feature at least indicating a failure type and the aggregated score for the corresponding failure type; and

predicting a failure cause of the storage system according to the failure analysis model and based on the failure features.

3. The method according to claim 2 , wherein aggregating the scores in the multiple preprocessed log files comprises:

aggregating the scores based on the failure types of the potential failure causes and log file types associated with the potential failure causes and included in the preprocessed log files of the multiple preprocessed log files.

4. The method according to claim 3 , wherein aggregating the scores in the multiple preprocessed log files further comprises:

aggregating the scores based on the failure types of the potential failure causes, the log file types, and storage node identifiers associated with the log file types and included in the preprocessed log files of the multiple preprocessed log files.

5. The method according to claim 1 , further comprising:

using the parameters to update the failure analysis model based on the probability being less than a threshold.

6. The method according to claim 1 , wherein obtaining at least one set of log files further comprises:

dividing the multiple preprocessed log files by multiple time windows; and

obtaining the divided multiple preprocessed log files within time windows in the multiple time windows as the at least one set of log files.

7. The method according to claim 1 , wherein the failure feature is determined for use in the failure analysis model at least in part by aggregating the scores for two or more of the preprocessed log files that have the same failure types, the same log files types and the same storage node identifiers.

8. An electronic device, comprising:

a processor; and

a memory coupled to the processor, the memory having instructions stored therein, wherein the instructions, when executed by the processor, cause the electronic device to perform actions for training a failure analysis model, the failure analysis model being used to determine a failure cause of a storage system, the actions comprising:

obtaining at least one set of log files including multiple preprocessed log files, the at least one set of log files including a marked failure cause of the storage system, and preprocessed log files in the multiple preprocessed log files including one or more potential failure causes of the storage system and scores associated with the potential failure causes;

predicting a failure cause of the storage system according to the failure analysis model and based on the potential failure causes and the scores in the multiple preprocessed log files; and

updating parameters of the failure analysis model based on a probability that the predicted failure cause is the marked failure cause;

wherein each of at least a subset of preprocessed log files comprises a plurality of fields for respective ones of a failure type, a log file type, an identifier of a storage node of the storage system, and a score for the corresponding preprocessed log file; and

wherein a failure feature is determined for use in the failure analysis model at least in part by aggregating the scores for multiple ones of the preprocessed log files based on one or more of the failure type, the storage node identifier and the log file type.

9. The electronic device according to claim 8 , wherein predicting a failure cause of the storage system comprises:

aggregating the scores in the multiple preprocessed log files based on failure types of the potential failure causes to generate failure features, the failure feature at least indicating a failure type and an aggregated score for the corresponding failure type; and

predicting a failure cause of the storage system according to the failure analysis model and based on the failure features.

10. The electronic device according to claim 9 , wherein aggregating the scores in the multiple preprocessed log files comprises:

aggregating the scores based on the failure types of the potential failure causes and log file types associated with the potential failure causes and included in the preprocessed log files of the multiple preprocessed log files.

11. The electronic device according to claim 10 , wherein aggregating the scores in the multiple preprocessed log files further comprises:

aggregating the scores based on the failure types of the potential failure causes, the log file types, and storage node identifiers associated with the log file types and included in the preprocessed log files of the multiple preprocessed log files.

12. The electronic device according to claim 8 , wherein the actions further comprise:

using the parameters to update the failure analysis model based on the probability being less than a threshold.

13. The electronic device according to claim 8 , wherein obtaining at least one set of log files further comprises:

dividing the multiple preprocessed log files by multiple time windows; and

obtaining the divided multiple preprocessed log files within time windows in the multiple time windows as the at least one set of log files.

14. A computer program product that is tangibly stored on a non-transitory computer-readable storage medium and comprises machine-executable instructions, wherein the machine-executable instructions, when executed by a machine, cause the machine to perform a method for training a failure analysis model, the failure analysis model being used to determine a failure cause of a storage system, the method comprising:

obtaining at least one set of log files including multiple preprocessed log files, the at least one set of log files including a marked failure cause of the storage system, and preprocessed log files in the multiple preprocessed log files including one or more potential failure causes of the storage system and scores associated with the potential failure causes;

predicting a failure cause of the storage system according to the failure analysis model and based on the potential failure causes and the scores in the multiple preprocessed log files; and

updating parameters of the failure analysis model based on a probability that the predicted failure cause is the marked failure cause;

wherein each of at least a subset of preprocessed log files comprises a plurality of fields for respective ones of a failure type, a log file type, an identifier of a storage node of the storage system, and a score for the corresponding preprocessed log file; and

wherein a failure feature is determined for use in the failure analysis model at least in part by aggregating the scores for multiple ones of the preprocessed log files based on one or more of the failure type, the storage node identifier and the log file type.

15. The computer program product according to claim 14 , wherein predicting a failure cause of the storage system comprises:

aggregating the scores in the multiple preprocessed log files based on failure types of the potential failure causes to generate failure features, the failure feature at least indicating a failure type and the aggregated score for the corresponding failure type; and

predicting a failure cause of the storage system according to the failure analysis model and based on the failure features.

16. The computer program product according to claim 15 , wherein aggregating the scores in the multiple preprocessed log files comprises:

aggregating the scores based on the failure types of the potential failure causes and log file types associated with the potential failure causes and included in the preprocessed log files of the multiple preprocessed log files.

17. The computer program product according to claim 16 , wherein aggregating the scores in the multiple preprocessed log files further comprises:

aggregating the scores based on the failure types of the potential failure causes, the log file types, and storage node identifiers associated with the log file types and included in the preprocessed log files of the multiple preprocessed log files.

18. The computer program product according to claim 14 , further comprising:

using the parameters to update the failure analysis model based on the probability being less than a threshold.

19. The computer program product according to claim 14 , wherein obtaining at least one set of log files further comprises:

dividing the multiple preprocessed log files by multiple time windows; and

obtaining the divided multiple preprocessed log files within time windows in the multiple time windows as the at least one set of log files.

20. The computer program product according to claim 14 , wherein the failure feature is determined for use in the failure analysis model at least in part by aggregating the scores for two or more of the preprocessed log files that have the same failure types, the same log files types and the same storage node identifiers.