wwj6591812 opened a new pull request, #10922:
URL: https://github.com/apache/paimon/pull/10922

   ### Purpose
   
   Existing BLOB fetch metrics expose outcomes and successful byte totals, but 
do not show partial payload reads/writes or transfer latency. Add a 
`blobTransfer` metric group with payload byte counters, completed/failed copy 
counters, cumulative elapsed time and a bounded millisecond histogram.
   
   - Classify the reader API into five fixed `sourceKind` values: `inline`, 
`http`, `file`, `custom` and `unknown`. Classification does not inspect or 
expose URIs, paths, hostnames, credentials, storage locations or network routes.
   - Preserve existing NULL fallback, exception propagation and the BLOB file 
format. The optional transfer reporter has a default implementation; its 
runtime failures do not replace the data-processing outcome.
   - Close all metric groups even if cleanup fails, release partially 
registered groups, and prevent late reports from reopening a closed collector.
   - Document that payload copies are distinct from committed snapshots and 
physical network traffic, including partial-write limitations, reused-source 
timing and counter/quantile aggregation.
   
   This PR targets master. A release-line backport can follow the community's 
normal process after review.
   
   ### Tests
   
   JDK 8; a clean nine-module reactor passed with 135 tests, zero failures, 
errors or skips: common 24, format 104, core 7.
   
   ```shell
   mvn -B -pl paimon-common,paimon-core,paimon-format spotless:apply
   mvn -B -pl paimon-core,paimon-format -am clean test -DwildcardSuites=none 
-Dtest=BlobTransferSourceTest,BlobFetchMetricsTest,BlobFormatWriterTest,BlobFileFormatTest,VideoFileFormatTest,BlobTest,ReusingBlobRefStreamProviderTest
 -Dsurefire.failIfNoSpecifiedTests=false -DfailIfNoTests=false 
-Dgit-commit-id.skip=true -Dmaven.gitcommitid.skip=true
   git diff --check
   ```
   
   Coverage includes standard/custom/transformed readers, partial read/write 
failures, telemetry failures preserving successful writes and original I/O 
exceptions, group reuse, bounded histograms, late reports, failed registration 
and cleanup failures. Existing BLOB format, array/map, video and source-reuse 
tests pass. The metric reference was updated; the complete documentation site 
was not built locally.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to