LatencyMetrics processing is incorrect in certain Cassandra versions #560
Assignees
Comments
Closed
emerkle826
added a commit
that referenced
this issue
Oct 28, 2024
* [CHANGE] [#556](#556) Update Management API dependencies to address CVEs * [CHANGE] [#547](#547) Deprecate Cassandra 3.11 support * [FEATURE] [#551](#551) Add Cassandra 5.0.2 to the build matrix * [FEATURE] [#549](#549) Add DSE 6.9.3 to the build matrix * [ENHANCEMENT] [#552](#552) Improve "liveness" probe implementation * [BUGFIX] [#553](#553) Fix CassandraTaskExports metric filtering to make it work with 5.0.x Major compactions * [BUGFIX] [#560](#560) Fix LatencyMetrics bucketing on 4.1 and 5.0 as their reservoir stores the data in microseconds, not nano (unlike 3.11 and 4.0) * [BUGFIX] [#562](#562) Fix version detection
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We set our buckets to accommodate for nanoseconds in latency metrics, however this isn't true anymore in newer versions of Cassandra, which are actually storing the dropwizard with microseconds.
Thus, we need to modify the timerFiller with nanosecond buckets for 3.11/4.0, but microseconds for 4.1/5.0.
┆Issue is synchronized with this Jira Story by Unito
┆Issue Number: MAPI-77
The text was updated successfully, but these errors were encountered: