[
https://issues.apache.org/jira/browse/CASSANDRA-21478?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Shlomit updated CASSANDRA-21478:
--------------------------------
Test and Documentation Plan:
Testing:
- Build + checkstyle green: .build/sh/ai-build (clean + jar) and
ant checkstyle checkstyle-test.
- New in-JVM dtest GuardrailSSTablesPerReadTest (7 tests): single-partition
and
range, warn and fail paths, internal-read-not-aborted, runtime threshold
change, and compaction-clears-condition. All pass.
- Updated GuardrailsConfigCommandsTest; existing SSTableReadLogsQueryTest
still
passes after relocating the single-partition diagnostic log.
- Manually validated end-to-end on a live single node: client warning and
abort
for both read paths, retained "has read N SSTables" log, and the startup
deprecation warning for the old sstables_per_read_log_threshold yaml key.
Documentation:
- CHANGES.txt entry.
- NEWS.txt: new guardrail, plus deprecation of sstables_per_read_log_threshold
and the StorageService JMX get/setSSTablesPerReadLogThreshold methods.
- conf/cassandra.yaml documents the new warn/fail thresholds.
Status: Patch Available (was: Open)
> Add a new guardrail around the number of SSTables per read
> ----------------------------------------------------------
>
> Key: CASSANDRA-21478
> URL: https://issues.apache.org/jira/browse/CASSANDRA-21478
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Feature/Guardrails
> Reporter: Caleb Rackliffe
> Assignee: Shlomit
> Priority: Normal
> Fix For: 6.x, 7.x
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> While there is some prior art on this for SAI, which warns and then fails
> queries that touch too many SSTable indexes, the normal non-index read path
> does not have any guardrail for extremely aberrant cases where perhaps
> compaction falls behind and a small percentage of reads touch hundreds of
> SSTables. It might be nice for the server to be louder about this.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
