Oracle International Corporation

Compaq Computer Corporation, Inc.

Sameer Arun Joshi

Alok Kumar Srivastava

Surojit Chatterjee

Prasad V Bagal

Jonathan Creighton

Harish Nandyala

Richard L Long

Rajiv Wickremesinghe

Shie-rei Huang

Ken Lee

Angelo Pruscino

Hanlin Daniel Chien

Raj Kumar

Kannabran Viswanathan

David Cavazos Woo

Suman R Bezawada

Emily Ronshien Su

Sha Chang

Ricardo Rey Diez

Michael Zoll

Sitaram V Lanka

Johannes Klein

Albert C Gondi

Roger J Hansen

Samarjeet Singh Tomar

Fulu Li

Mohsin Beg

Ravi Ranjan

Vikram Rai

WeiHsin Gu

Harendra Kumar Mishra

Shrihari Amarendra Bhat

'Semi-supervised' machine learning relies on less human input than a supervised algorithm to train a machine learning algorithm to perform entity recognition (NER). Starting with a known entity value or known pattern value for a specific entity type, phrases in a training data corpus are identified that include the known entity value. Context-value patterns are generated to match selected phrases that include the known entity value. One or more context-value patterns may be validated based on human input. The validated patterns identify additional entity values. A subset of the additional entity values may also be validated based on human input. Occurrences of validated entity values may be labeled in the training corpus. Sample phrases from the labeled training dataset may be extracted to form a reduced-size training set for a supervised machine learning model which may be further used in production to label data for any named entity recognition application.

System and method to generate a labeled dataset for training an entity detection system

Duplicate bug report detection using machine learning algorithms and automated feedback incorporation is disclosed. For each set of bug reports, a user-classification of the set of bug reports as including duplicate bug reports or non-duplicate bug reports is identified. Also for each set of bug reports, correlation values corresponding to a respective feature, of a plurality of features, between bug reports in the set of bug reports is identified. Based on the user-classifications and the correlation values, a model is generated to identify any set of bug reports as including duplicate bug reports or non-duplicate bug reports. The model is applied to classify a particular bug report and a candidate bug report as duplicate bug reports or non-duplicate bug reports.

Duplicate bug report detection using machine learning algorithms and automated feedback incorporation

A multi-node cluster is configured for credential management. A method commences by retrieving a super-user credential from a credential record stored in a location accessible to the cluster, then propagating the super-user credential to a set of nodes in the multi-node cluster. A credential creating processes is invoked on at least some of the set of nodes. Application-level credential access can be implemented in a multi-cluster environment by carrying-out an exchange that passes credentials between a first cluster and a second cluster over a secure channel. A protocol is observed whereby one or more applications running on the first cluster receive new credentials for accessing the second cluster from the credential serving process after the credential creating process creates the new credential.

Automatic creation and management of credentials in a distributed environment

One embodiment of the present invention provides a system for facilitating an upgrade of a cluster of servers in the presence of one or more inaccessible nodes in the cluster. During operation, the system upgrades a version of a distributed software program on each of a plurality of nodes in the cluster. The system may detect that one or more nodes of the cluster are inaccessible. The system continues to upgrade nodes in the cluster other than the one or more nodes that were detected to be inaccessible, in which upgrading involves installing and activating a newer version of the distributed software on the nodes being upgraded. The system then upgrades an acting version of the cluster.

Forcibly completing upgrade of distributed software in presence of failures

A method, system, and computer program product for restoring blocks of data stored at a corrupted data site using two or more mirror sites. The method commences by receiving a trigger event from a component within an application server environment where the trigger event indicates detection of a corrupted data site. The trigger is classified into at least one of a plurality of trigger event types, which trigger event type signals further processing for retrieving from at least two mirror sites, a first stored data block and a second stored data block corresponding to the same logical block identifier from the first mirror site. The retrieved blocks are compared to determine a match value, and when the match value is greater than a confidence threshold, then writing good data to the corrupted data site before performing consistency checks on blocks in physical or logical proximity to the corrupted data site.

Detection of logical corruption in persistent storage and automatic recovery therefrom

One embodiment of the present invention provides a system for facilitating an upgrade of a cluster of servers in the presence of one or more inaccessible nodes in the cluster. During operation, the system receives an instruction to upgrade an older version of a distributed software program installed on each of a plurality of nodes forming the cluster. The system installs and activates a newer version of software on a subset of the plurality of nodes, the subset having fewer nodes than the entire cluster. The system receives an instruction to forcibly upgrade the cluster. Subsequently, the system upgrades the cluster so that the subset of nodes operate with the newer version of software, in which each node remains a member of the cluster during the upgrade.

Disclosed is a method, mechanism, and computer usable medium for managing and accessing static and shard data. In one approach, described is a method and mechanism for implementing group dependent keys (GDKs) in a computing system, in which the GDK is visible to all members of a distributed system, but its value(s) and subtree(s) could be different for different groups. Members of each group see the same view of the value and subtree of a GDK. Also disclosed is a method, mechanism, and computer usable medium for implementing group dependent links (GDLs) in a computing system. According to one approach, a data transformation function is used to coordinate changes to different versions of shared data.

Method and mechanism for managing and accessing static and dynamic data

Techniques are described herein for synchronizing cluster time. According to one technique, a master node is appointed in a cluster. Other 'slave' nodes periodically synchronize their clocks with the master node. To synchronize its clock with the master node, a slave node sends a timestamped message to the master node, which also timestamps the message and sends the message back to the slave node, which then timestamps the message again. Based on the timestamps, the slave node is able to determine the difference between the master node's clock's time and slave node's clock's time, compensating for the message travel time between master node and slave node. Depending on various circumstances, and based on the determined difference, the slave node adjusts its clock so that the time indicated by the slave node's clock at least begins to approach more closely the time indicated by the master node's clock.

Time synchronization in cluster systems

A method and apparatus is provided for determining the most probable cause of a problem observed in a complex multi-host system. The approach relies on a probabilistic model to represent causes and effects in a complex computing system. However, complex systems include a multitude of independently operating components that can cause temporary anomalous states. To reduce the resources required to perform root cause analysis on each transient failure, as well as to raise the confidence in the most probable cause of a failure that is identified by the model, inputs to the probabilistic model are aggregated over a sliding window of values from the recent past.

Fault identification of multi-host complex systems with timesliding window analysis in a time series

A method and mechanism for managing data to resolve incompatibilities between different versions of data is disclosed. According to one approach, incompatibilities to private data is managed using group/member dependent key(s)/link(s), static shared data is managed using group/member dependent key(s)/link(s), and dynamic shared data is managed using data transformation function(s).

Method and mechanism for managing incompatible changes in a distributed system

Techniques are provided for managing replicated data items in a cluster of shared nothing nodes. The techniques include synchronously propagating changes to one or more replicas, while asynchronously propagating changes to the remaining replicas. Because the replicas to which the change is synchronously replicated may be a relatively small percentage of the total number of replicas, the overhead associated with performing an update is reduced. The process of selecting which nodes are involved in the synchronous update operation may take into account a variety of factors, including which replicas are available at the time the update operation is being performed.

Replicating data across the nodes in a cluster environment

Less-restrictive techniques are provided for ensuring that replicated-data systems will never provide out-of-date version of data items. A replicated-data system maintains a version number, a membership group identifier, and a membership count, with each replica of a data item. These values are maintained in such a way as to allow the replicated-data system to reliably satisfy some read requests even though half, or less than half, of the replicas of the data item are available.

Growing community of inventors

San Jose, CA, United States of America

Sameer Arun Joshi