Skip to content

Conversation

@michaelsembwever
Copy link
Member

marianne-manaog and others added 3 commits December 7, 2022 11:50
default LatencySelector write(double percentile) { return (read, write) -> write.get(percentile); }
default LatencySelector maxReadWrite(double percentile) { return (read, write) -> max(read.get(percentile), write.get(percentile)); }
}

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Was this line break removal intentional?

{
ContentionStrategy strategy = parseStrategy("min=0ms,max=10ms,random=uniform").strategy;
double total = 0;
int count = 1000;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mocking sleep and time would allow measuring if it the calculations are precisely correct.

It's fine for now though.

@michaelsembwever michaelsembwever changed the base branch from trunk to cassandra-4.1 December 7, 2022 19:39
@aweisberg
Copy link
Contributor

Committed as f0ad7eadbeb3208e08a9339881931222fdab253b

@aweisberg aweisberg closed this Dec 9, 2022
michaeljmarshall pushed a commit to michaeljmarshall/cassandra that referenced this pull request Dec 17, 2025
…che#2044)

Adds the metrics in IndexContext and the tree view of the query's Plan
to the log messages produced for slow SAI queries.

There is a new system property named cassandra.sai.slow_query_log.execution_info_enabled
to toggle this feature, which is enabled by default. This property can be changed
dynamically at any time, without a restart.

When SAI-specific information is added to the slow query logger, the log messages for
slow queries can take up almost 10x the (uncompressed) disk space. In the worst of the
worst cases, with all queries being permanently slower than 500ms, and at least 50
different queries, reporting every 5 seconds (the default), the slow query logger can
produce around 120MB of text per day and instance. With the detailed logging, it would
be 1.14GB per day.
michaelsembwever pushed a commit to thelastpickle/cassandra that referenced this pull request Jan 7, 2026
… logger (apache#2044)

Adds the metrics in IndexContext and the tree view of the query's Plan
to the log messages produced for slow SAI queries.

There is a new system property named cassandra.sai.slow_query_log.execution_info_enabled
to toggle this feature, which is enabled by default. This property can be changed
dynamically at any time, without a restart.

When SAI-specific information is added to the slow query logger, the log messages for
slow queries can take up almost 10x the (uncompressed) disk space. In the worst of the
worst cases, with all queries being permanently slower than 500ms, and at least 50
different queries, reporting every 5 seconds (the default), the slow query logger can
produce around 120MB of text per day and instance. With the detailed logging, it would
be 1.14GB per day.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants