Hi As discussed, we can meet up next week to have a session with the support team (CC / ).

Hi , I have opened to remove extra stats. Hoping to see quick improvement on mortimer tooling side.

Unfortunately, moving away from indexer_stats.log is not a simple. Currently prometheus is configured to vary the stat collection frequency for high cardinality stats. That does not work well when debugging indexing CBSEs. If it was simple, we would have taken it up in 7.0 timeframe.

Hi , I am again requesting that you review the above stats and do not add any new aggregated metrics to Prometheus if they can be derived from existing ones.

If you don't remove them now then they will most likely be deprecated when I do a metrics review later this year, causing disruption to anyone who has started to rely on them.

CC /

Repro steps->

1. Create a 2 node cluster with n1ql,kv and index services

2. Load some data and create some indexes and run queries on them

3. using the end point to verify the stats curl -u Administrator:password http://<your_ip>:8091/_prometheus/api/v1/query?query=index_total_rows_scanned

Verify for all the below stats-

num_indexes
num_storage_instances
avg_resident_percent
avg_mutation_rate
avg_drain_rate
avg_disk_bps
total_data_size
total_disk_size
memory_used_storage
memory_total_storage
total_requests
total_rows_returned
total_rows_scanned

Verified on 7.2.0 5195 

Closing the ticket

Build couchbase-server-7.2.0-5156 contains indexing commit aaa1143 with commit message:
: Expose aggregated node level stats via prometheus