Skip to content

Starting index already present on disk bugs out the S3 download #329

@sarthakn7

Description

@sarthakn7

Starting index on replica from a backup when the index is already downloaded before bugs out S3 download with the log Not all bytes were read from the S3ObjectInputStream being infinitely printed.

Command run: curl -XPOST --fail localhost:12001/v1/start_index -d "{\"indexName\": \"test_idx\",\"mode\": \"REPLICA\",\"primaryAddress\": \"localhost\",\"port\": 11001,\"restore\": {\"serviceName\": \"nrtsearch-test\",\"resourceName\": \"test_idx\"}}";

Logs:

[INFO ] 2021-06-07 09:41:29.527 [GrpcLuceneServerExecutor-4-thread-6] ArchiverImpl - Downloading resource test_idx for service nrtsearch-test version test_idx_2021-05-05_uswest2_data to directory /home/sarthakn/nrtsearch_dev/replica_files/archiver/test_idx_data/test_idx_2021-05-05_uswest2_data
[INFO ] 2021-06-07 09:41:29.564 [GrpcLuceneServerExecutor-4-thread-6] ArchiverImpl - Full object size: 39585368645
[INFO ] 2021-06-07 09:41:29.660 [GrpcLuceneServerExecutor-4-thread-6] ArchiverImpl - Object parts: 4719
[INFO ] 2021-06-07 09:41:29.940 [GrpcLuceneServerExecutor-4-thread-6] ArchiverImpl - Object streaming started...
[INFO ] 2021-06-07 09:41:29.943 [GrpcLuceneServerExecutor-4-thread-6] ArchiverImpl - Directory /home/sarthakn/nrtsearch_dev/replica_files/archiver/test_idx_data/test_idx_2021-05-05_uswest2_data already exists, not re-downloading from Archiver
Jun 07, 2021 9:41:29 AM com.amazonaws.services.s3.internal.S3AbortableInputStream close
WARNING: Not all bytes were read from the S3ObjectInputStream, aborting HTTP connection. This is likely an error and may result in sub-optimal behavior. Request only the bytes you need via a ranged GET or drain the input stream after use.
Jun 07, 2021 9:41:29 AM com.amazonaws.services.s3.internal.S3AbortableInputStream close
WARNING: Not all bytes were read from the S3ObjectInputStream, aborting HTTP connection. This is likely an error and may result in sub-optimal behavior. Request only the bytes you need via a ranged GET or drain the input stream after use.
...

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions