Details
-
New Feature
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
26.10
-
None
-
MXS-SPRINT-282, MXS-SPRINT-283
-
Description
he cache filter reports almost nothing about its own behaviour,
- why the cache makes queries slower,
- why the hit rate is low,
- and whether the cache slows MaxScale itself.
Cloud is asking for 12 metrics from the cache filter.
All of them are counters, histograms or gauges at decision points that already exist in the filter.
COUNTERS
|
cache.queries 1 statements the filter examined |
cache.not_cached{reason} 1 dropped before any lookup, by reason |
cache.lookup_results{result} 1 outcome of every lookup |
cache.puts{result} 1 outcome of every store attempt |
cache.uncached_evictions 2 evictions from the known-fast set |
|
|
HISTOGRAMS (buckets or quantiles, not averages)
|
cache.lookup_duration 1 ask the store -> answer in hand |
cache.put_duration 1 store a value; on the client's path |
cache.backend_duration 1 server execution time |
cache.filter_duration 1 routing-thread time inside the filter |
cache.value_bytes 2 result size |
|
|
GAUGES
|
cache.backend_up 1 store reachable |
max_uncached_query_count 2 add to the existing uncached_queries |
values (items, hits, misses)
|
For more details see https://mariadbcorp.atlassian.net/browse/MCDEV-4561
Attachments
Issue Links
- is duplicated by
-
MCDEV-4561 Loading...