Database replication error recovery based on supervised learning

1. A system, comprising:

a database system implemented using a server system, the database system configurable to cause:

training a predictive model using a set of training data;

performing database replication processing in a cloud computing environment such that one or more log files are generated, the one or more log files including text generated during the database replication processing;

obtaining the text from the one or more log files;

transforming the text into a plurality of vectors, each vector representing an instance of an error that occurred during database replication processing, wherein transforming the text includes:

segmenting the text into sentences;

segmenting the sentences into tokens such that segmented sentences are generated;

tagging the tokens of the segmented sentences as parts of speech such that tagged sentences are generated;

detecting entities in the tagged sentences;

detecting relations among the entities; and

converting the relations into vectors;

classifying, using the predictive model, one or more errors based on the vectors as a recoverable error such that one or more recoverable errors are identified;

determining, for the one or more recoverable errors based, at least in part, on corresponding error sub-types, one or more recovery jobs associated with database replication processing in the cloud computing environment; and

automatically invoking the one or more recovery jobs, the one or more recovery jobs including re-executing at least a portion of the database replication processing.

2. The system of claim 1 , the database system further configurable to cause:

classifying, using the predictive model, at least a second error as an unrecoverable error based on at least one of the vectors;

notifying a user that the unrecoverable error has occurred.

3. The system of claim 1 , wherein transforming the text into a plurality of vectors comprises:

segmenting the text into sentences;

segmenting the sentences into tokens such that segmented sentences are generated;

tagging the tokens of the segmented sentences as parts of speech such that tagged sentences are generated;

detecting entities in the tagged sentences;

detecting relations among the entities; and

converting the relations into vectors.

4. The system of claim 1 , wherein classifying the errors based on vectors comprises applying the predictive model to the vectors.

5. The system of claim 4 , the database system further configurable to cause:

splitting the vectors into training data and testing data;

training the predictive model with the training data;

testing the predictive model with the testing data;

evaluating metrics obtained from the training and testing; and

tuning error classification predictive model parameters of the predictive model based, at least in part, on the evaluated metrics.

6. The system of claim 5 , wherein the predictive model performs binary classification on the vectors.

7. The system of claim 1 , wherein classifying the errors includes classifying the errors into sub-types of errors; and

determining at least one recovery job for each sub-type of the recoverable errors.

8. A computer-implemented method comprising:

training a predictive model using a set of training data;

performing database replication processing in a cloud computing environment such that one or more log files are generated, the one or more log files including text generated during the database replication processing;

obtaining the text from the one or more log files;

transforming the text into a plurality of vectors, each vector representing an instance of an error that occurred during database replication processing, wherein transforming the text includes:

segmenting the text into sentences;

segmenting the sentences into tokens such that segmented sentences are generated;

tagging the tokens of the segmented sentences as parts of speech such that tagged sentences are generated;

detecting entities in the tagged sentences;

detecting relations among the entities; and

converting the relations into vectors;

classifying, using the predictive model, one or more errors based on the vectors as a recoverable error such that one or more recoverable errors are identified;

determining, for the one or more recoverable errors based, at least in part, on corresponding error sub-types, one or more recovery jobs associated with database replication processing in the cloud computing environment; and

automatically invoking the one or more recovery jobs, the one or more recovery jobs including re-executing at least a portion of the database replication processing.

9. The computer-implemented method of claim 8 , further comprising

classifying, using the predictive model, at least a second error as an unrecoverable error based on at least one of the vectors;

notifying a user that the unrecoverable error has occurred.

10. The computer-implemented method of claim 8 , wherein transforming the text into a plurality of vectors comprises:

segmenting the text into sentences;

segmenting the sentences into tokens;

tagging the tokens of the segmented sentences as parts of speech such that tagged sentences are generated;

detecting entities in the tagged sentences;

detecting relations among the entities; and

converting the relations into vectors.

11. The computer-implemented method of claim 8 , wherein classifying one or more errors based on the vectors comprises applying the predictive model to the vectors.

12. The computer-implemented method of claim 11 , further comprising:

splitting the vectors into training data and testing data;

training the predictive model with the training data;

testing the predictive model with the testing data;

evaluating metrics obtained from the training and testing; and

tuning error classification predictive model parameters of the predictive model based, at least in part, on the evaluated metrics.

13. The computer-implemented method of claim 8 , wherein the predictive model performs binary classification on the vectors.

14. The computer-implemented method of claim 8 , further comprising:

classifying the errors into sub-types of errors; and

for each sub-type of the recoverable errors, determine at least one recovery job.

15. A tangible, non-transitory computer-readable storage medium having instructions stored thereon which, when executed by a processing device, cause the processing device to:

train a predictive model using a set of training data;

perform database replication processing in a cloud computing environment such that one or more log files are generated, the one or more log files including text generated during the database replication processing;

obtain the text from the one or more log files;

transform the text into a plurality of vectors, each vector representing an instance of an error that occurred during database replication processing, wherein transforming the text includes:

segmenting the text into sentences;

segmenting the sentences into tokens such that segmented sentences are generated;

tagging the tokens of the segmented sentences as parts of speech such that tagged sentences are generated;

detecting entities in the tagged sentences;

detecting relations among the entities; and

converting the relations into vectors;

classify, using the predictive model, one or more errors based on the vectors as a recoverable error such that one or more recoverable errors are identified;

determine, for the one or more recoverable errors based, at least in part, on corresponding error sub-types, one or more recovery jobs associated with database replication processing in the cloud computing environment; and

automatically invoke the one or more recovery jobs, the one or more recovery jobs including re-executing at least a portion of the database replication processing.

16. The tangible, non-transitory computer-readable storage medium of claim 15 , having instructions stored thereon which, when executed by the processing device, cause the processing device to:

classify, using the predictive model, at least a second error as an unrecoverable error based on at least one of the vectors;

notify a user that the unrecoverable error has occurred.

17. The tangible, non-transitory computer-readable storage medium of claim 15 , wherein instructions to transform the text comprise instructions stored thereon that, in response to execution by the processing device, cause the processing device to:

segment the text into sentences;

segment the sentences into tokens such that tokenized sentences are generated;

tag the tokens of the tokenized sentences as parts of speech such that tagged sentences are generated;

detect entities in the tagged sentences;

detect relations among the entities; and

convert the relations into vectors.

18. The tangible, non-transitory computer-readable storage medium of claim 15 , wherein instructions to classify one or more errors apply the predictive model to the vectors.

19. The tangible, non-transitory computer-readable storage medium of claim 18 , further comprising instructions stored thereon that, in response to execution by the processing device, cause the processing device to:

split the vectors into training data and testing data;

train the predictive model with the training data;

test the predictive model with the testing data;

evaluate metrics obtained from the training and testing; and

tune predictive model parameters of the predictive model based, at least in part, on the evaluated metrics.

20. The tangible, non-transitory computer-readable storage medium of claim 19 , wherein the predictive model performs binary classification on the vectors.

21. The tangible, non-transitory computer-readable storage medium of claim 15 , further comprising instructions stored thereon that, in response to execution by the processing device, cause the processing device to:

classify the errors into sub-types of errors; and

determine, for each sub-type of the recoverable errors, at least one recovery job.