Gitiles
Code Review
Sign In
asterix-gerrit.ics.uci.edu
/
asterixdb
/
caf0833b518a5c2dba3a48f85a7c9203ab1ac865
caf0833
Merge fullstack_asterix_stabilization into fullstack_hyracks_result_distribution.
by madhusudancs@gmail.com
· 12 years ago
1373c07
Move print newline back to PrinterBasedWriter because not all writers need this functionality.
by madhusudancs@gmail.com
· 12 years ago
d4140ee
Implement a method to return valid file references.
by madhusudancs@gmail.com
· 12 years ago
cbcd73b
Style fix in result writer operator.
by madhusudancs@gmail.com
· 12 years ago
c3bd7c3
Implement a basic memory manager to cache results from jobs and use the writer and reader to store and distribute the results.
by madhusudancs@gmail.com
· 12 years ago
7b604e0
Implement interfaces for result writing and reading.
by madhusudancs@gmail.com
· 12 years ago
e611d23
Add support to provide the result distribution buffer memory budget as command line option to the node controller.
by madhusudancs@gmail.com
· 12 years ago
f58b75d
Implement a class for storing the state of the results.
by madhusudancs@gmail.com
· 12 years ago
88b4d32
Supply only frame size to NetworkOutputChannel not the whole context.
by madhusudancs@gmail.com
· 12 years ago
f5cfdff
Change almost all the integration tests to use ResultWriterOperatorDescriptor for result distribution.
by madhusudancs@gmail.com
· 12 years ago
d6066cb
Add the additional datasetIPAddress parameter required for the NCConfig in all the tests and implement the basic infrastructure required to run hyracks integration tests using ResultWriterOperatorDescriptor.
by madhusudancs@gmail.com
· 12 years ago
248afcb
Style/Formatting fixes.
by madhusudancs@gmail.com
· 12 years ago
94c3683
Provide the functionality to set byte array size in the frame output stream and its parent classes.
by madhusudancs@gmail.com
· 12 years ago
7acfa1e
Implement a result serializer, its factories and providers for test to serialize the results based on record descriptor instead of printer factory in test framework.
by madhusudancs@gmail.com
· 12 years ago
994601e
Add DistributeResultOperator implementation that was missed.
by madhusudancs@gmail.com
· 12 years ago
b198c57
Implement a frame tuple accessor class for results which contains just a single serialized string of records without any fields.
by madhusudancs@gmail.com
· 12 years ago
20403e2
ResultWriterOperator should provide the record descriptor of its input buffers at run time to serializer and it shouldn't be statically provided to the factory provider at static query compilation time.
by madhusudancs@gmail.com
· 12 years ago
1daf6a6
Separate the result reader out of the client class which sets up connection.
by madhusudancs@gmail.com
· 12 years ago
16c7c77
Separate out WriteOperator and create a new DistributeResultOperator for result distribution.
by madhusudancs@gmail.com
· 12 years ago
1617d6c
Implement a default result serializer factory provider in Algebricks.
by madhusudancs@gmail.com
· 12 years ago
296655c
Merge recent set of changes from fullstack_asterix_stabilization to fullstack_hyracks_result_distribution.
by madhusudancs@gmail.com
· 12 years ago
a837e39
Adding the import that got lost somewhere :P
by madhusudancs@gmail.com
· 12 years ago
7a4d551
Merge fullstack_asterix_stabilization into fullstack_hyracks_result_distribution.
by madhusudancs@gmail.com
· 12 years ago
00f2ad0
Fix abstract integration test to include serializer init.
by madhusudancs@gmail.com
· 12 years ago
1374e75
Define an init method for the serializer and call it in result writer's open method.
by madhusudancs@gmail.com
· 12 years ago
c03819b
Update Java version to 1.7 in the pom.xml files that were added in this branch.
by madhusudancs@gmail.com
· 12 years ago
a5ed515
Merge fullstack_asterix_stabilization into fullstack_hyracks_result_distribution.
by madhusudancs@gmail.com
· 12 years ago
c9a8229
Decrement the number of available frames after each read and check for the available frames count each time.
by madhusudancs@gmail.com
· 12 years ago
ca27294
Fix the tests and code to support server side serialization.
by madhusudancs@gmail.com
· 12 years ago
46f3897
Fix the tests to use the new serializer names.
by madhusudancs@gmail.com
· 12 years ago
1822817
Refactor ResultWriterOperatorDescriptor to use the new interface names and use the rewritten FrameOutput with different arguments to be passed.
by madhusudancs@gmail.com
· 12 years ago
fbb43fb
Move the FrameOutputStream class to utils directory from io directory.
by madhusudancs@gmail.com
· 12 years ago
ba89d0b
Remove the usage of serialized descriptor.
by madhusudancs@gmail.com
· 12 years ago
76c0791
The tuple printer should not print newline, the caller can do it by itself.
by madhusudancs@gmail.com
· 12 years ago
e4551ab
Remove the usage of "Appender" from result serializer interfaces because we do not want those implementations to ByteBuffer frames anymore.
by madhusudancs@gmail.com
· 12 years ago
ffc49ae
To be able to serialize the results on the server side obtain the printer factories for the results and pass it to the result handling metadataprovider run time.
by madhusudancs@gmail.com
· 12 years ago
7656bf4
Provide the network address of the result distribution TCP endpoint for scheduler tests.
by madhusudancs@gmail.com
· 12 years ago
e5b8fab
Merge fullstack_asterix_stabilization into fullstack_hyracks_result_distribution.
by madhusudancs@gmail.com
· 12 years ago
698f3fa
Add the new parameter to node controller info which takes result distribution tcp endpoint.
by madhusudancs@gmail.com
· 12 years ago
dca02da
Update hyracks api version to 0.2.3-SNAPSHOT.
by madhusudancs@gmail.com
· 12 years ago
58f5056
Merge fullstack_asterix_stabilization into fullstack_hyracks_result_distribution branch.
by madhusudancs@gmail.com
· 12 years ago
af06318
The dataset API only gets the serialized descriptor.
by madhusudancs@gmail.com
· 12 years ago
e9b8956
Revert hyracks api version back to 0.2.2 in the hope that it will be changed to 0.2.3 during full stack merge.
by madhusudancs@gmail.com
· 12 years ago
311bfb3
Implement the result handling method in hivestrix metadataprovider that returns nothing.
by madhusudancs@gmail.com
· 12 years ago
3daffb7
Implement a null returning dataset partition manager get method for tests.
by madhusudancs@gmail.com
· 12 years ago
f8187e1
Fix an issue that arose during the merge of missing to merge a conflicted hunk.
by madhusudancs@gmail.com
· 12 years ago
b1d369e
Implement an interface to represent all objects that can be JSON serialized.
by madhusudancs@gmail.com
· 12 years ago
d788caf
Close the handle of the result storing file after it is completely read.
by madhusudancs@gmail.com
· 12 years ago
9a7bd81
Pass job and result set id to dataset partition writer too.
by madhusudancs@gmail.com
· 12 years ago
df98536
Include result set id in the string identifier corresponding to a job.
by madhusudancs@gmail.com
· 12 years ago
4d84931
Add a missed line to report the starting of a job.
by madhusudancs@gmail.com
· 12 years ago
a1801d9
Remove unnecessary suffix "Function" from the method a couple of get method names.
by madhusudancs@gmail.com
· 12 years ago
3b7514b
Implement the client side of the RPC chain to fetch the job status from the client.
by madhusudancs@gmail.com
· 12 years ago
1180edf
Implement support for determining and reporting job status.
by madhusudancs@gmail.com
· 12 years ago
5a98ef7
Add support for storing result status for a job in the dataset directory record.
by madhusudancs@gmail.com
· 12 years ago
49a6af0
Remove the unnecessary wait for getting the monitors since the client architecture has changed.
by madhusudancs@gmail.com
· 12 years ago
d072edf
Pass the IP address of the result distributor while starting up Hyracks NC.
by madhusudancs@gmail.com
· 12 years ago
bc7dfa8
Update the tests to use the rewritten HyracksDataset API.
by madhusudancs@gmail.com
· 12 years ago
7990356
Implement the support in Algebricks to instantiate a result writer operator.
by madhusudancs@gmail.com
· 12 years ago
801ad55
Completely rewrite the HyracksDataset client API class to use the new serializer and fetch the records buffer by before.
by madhusudancs@gmail.com
· 12 years ago
6d90e3e
Style and documentation fixes.
by madhusudancs@gmail.com
· 12 years ago
0b2aca3
Add the support to pass serialized record descriptor to the dataset directory service which was missed in the previous commit.
by madhusudancs@gmail.com
· 12 years ago
78f8c00
Remove the job id parameter to fetch the dataset directory service info.
by madhusudancs@gmail.com
· 12 years ago
37b7d29
Implement the client side of RPCs to give the clients the ability to retrieve serialized record descriptor.
by madhusudancs@gmail.com
· 12 years ago
a66e9dc
Rewrite the dataset directory service.
by madhusudancs@gmail.com
· 12 years ago
dac14cf
Pass the serialized record descriptor all the way upto dataset directory service in the RPC chain.
by madhusudancs@gmail.com
· 12 years ago
a1e56ce
Teach result writer operator about the record descriptor.
by madhusudancs@gmail.com
· 12 years ago
4bb03e9
Revert the hyracks version to 0.2.2 for all the packages.
by madhusudancs@gmail.com
· 12 years ago
646a03f
Enforce ordering in exposing the directory records for partitions to the clients if ordering constraint is imposed by the ResultWriterOperatoryDescriptor.
by madhusudancs
· 12 years ago
b44e20d
Provide the ordering specification to result writer operator descriptor in SortMergeTest.
by madhusudancs
· 12 years ago
a4478a2
Use a slightly enhanced datastructure to store whether the results should be ordered or not.
by madhusudancs
· 12 years ago
75fd920
Add a parameter for the ResultWriterOperatorDescriptor to support ordered result distribution.
by madhusudancs
· 12 years ago
98214bf
Remove the unused IDatasetInputChannel interface definition file.
by madhusudancs
· 12 years ago
499aaf1
We do not have a need to define a frame size statically within the dataset now because we obtain that from the job specification.
by madhusudancs
· 12 years ago
f9332dd
Use IInputChannel instead of IDatasetInputChannel in the monitor definitions.
by madhusudancs
· 12 years ago
459d110
Construct the DatasetClientContext object in the HyracksDataset constructor and pass it to the result channel's open method.
by madhusudancs
· 12 years ago
b641a67
Implement a client side context from IHyracksCommonContext.
by madhusudancs
· 12 years ago
bcb2b0a
Extend IDatasetInputChannelMonitor from IInputChannelMonitor rather than creating a new one altogether.
by madhusudancs
· 12 years ago
82d17f7
Reimplement the DatasetNetworkInputChannel using the IInputChannel interface instead of creating a new interface defintion.
by madhusudancs
· 12 years ago
18143eb
Refactor IInputChannel.open method to take IHyracksCommonContext as the parameter instead of IHyracksTaskContext.
by madhusudancs
· 12 years ago
3b2d55b
Just use null as values to localAddress and listener for non listening TCP endpoints instead of copy pasting the constructor method.
by madhusudancs
· 12 years ago
c11728d
Fix a formatting issue with the comment.
by madhusudancs
· 12 years ago
3385489
Update the sort merge test to use the new API.
by madhusudancs
· 12 years ago
4da79e4
Use DatasetDirectoryRecords instead of NetworkAddresses everywhere we access that information in the dataset stack.
by madhusudancs
· 12 years ago
a5b45f5
Add support for registering the result set id with the ResultWriterOperatorDescriptor.
by madhusudancs
· 12 years ago
d5769b0
Support adding result set ids to the job specification and retrieving it.
by madhusudancs
· 12 years ago
41da2db
Implement a class that holds a directory record.
by madhusudancs
· 12 years ago
9c7a006
Implement ResultSetId class to represent result sets.
by madhusudancs
· 12 years ago
544eb63
We should print the buffer before recycling it
by madhusudancs
· 12 years ago
93f3bb9
Synchronize reads and writes to not let the reader attempt to read the file before it is written by the writer.
by madhusudancs
· 12 years ago
5b029cf
Formatting fixes.
by madhusudancs
· 12 years ago
78eb0c2
Make all the necessary changes or additions to report the data transfer info across the new result distribution socket for the node controller.
by madhusudancs
· 12 years ago
83f33fb
Update the tests to set the IPAddress for result distribution socket.
by madhusudancs
· 12 years ago
b6cd6db5
Update all the imports with the new locations and fix the method calls with argument changes.
by madhusudancs
· 12 years ago
117703a
Complete the implementation of basic DatasetPartitionWriter class to work with the network layer.
by madhusudancs
· 12 years ago
1b6cfde
Update the ResultWriterOperatorDescriptor to work with the new network layer implementation.
by madhusudancs
· 12 years ago
44faf47
Update the implementation of DatasetPartitionManger along with its interface to support the network communication completely.
by madhusudancs
· 12 years ago
586abd7
Create a separate datasetNetworkManager with its own socket (ipaddress and port)
by madhusudancs
· 12 years ago
07c1e8e
Allow TCPEndpoints without listeners by overloading the required methods.
by madhusudancs
· 12 years ago
b7fa154
Teach root pom.xml about the new projects and update the poms in other projects to include the required projects as dependencies.
by madhusudancs
· 12 years ago
Next »