OCI HDFS connector in read-ahead mode doesn't emit `bytesRead` input metric

[oracle]

OCI HDFS connector in read-ahead mode doesn't emit `bytesRead` input metric

[mricken]

Thank you for your fix, we'll work on making a new release that includes your contribution.

[kvirund]

The problem is with read implementations of the com.oracle.bmc.hdfs.store.BmcReadAheadFSInputStream: they do not call the super.read and do not change statistics on their own.

[kvirund]

Implementations should be changed to:

    @Override
    public int read() throws IOException {
        LOG.debug("{}: Reading single byte at position {}", this, filePos);
        if (dataPos == -1) {
            fillBuffer();
        }
        if (dataPos == -1) {
            return -1;
        }
        filePos++;
        int result = Byte.toUnsignedInt(data[dataCurOffset++]);
+       this.statistics.incrementBytesRead(1);
        if (atEndOfBuffer()) {
            clearBuffer();
        }
        return result;
    }

and

    @Override
    public int read(byte[] buffer, int offset, int length) throws IOException {
        LOG.debug("{}: Reading {} bytes at current position {}", this, length, filePos);
        if (dataPos == -1) {
            fillBuffer();
        }
        if (dataPos == -1) {
            return -1; // EOF
        }
        int n = Math.min(length, dataMax - dataCurOffset);
        System.arraycopy(data, dataCurOffset, buffer, offset, n);
        dataCurOffset += n;
        filePos += n;
        if (atEndOfBuffer()) {
            if (n != length) {
                LOG.debug("{}: Short Read; exhausted buffer", this);
            }
            clearBuffer();
        }
+       this.statistics.incrementBytesRead(n);
        return n;
    }

[kvirund]

Tested it by referencing patched implementation in fs.oci.io.read.custom.stream.

[anton-gorev-oracle]

PR: #78

[mricken]

This was released in version 3.3.4.0.0.0. Closing.