Inaccurate Bucket Intervals for celestia_consensus_block_interval_seconds_bucket Metric #1307
Labels
ice-box
issues are automatically assigned this label until they are planned.
introspection
T:Bug
Type: Bug (confirmed)
Bug Report
Setup
CometBFT version: N/A
Have you tried the latest version: yes
ABCI app: N/A
Environment: N/A
What happened?
The
block_interval_seconds
metric uses the default Prometheus histogram buckets which do not align with the expected block times for the Celestia blockchain. As the typical block time is approximately 12 seconds, using the default bucket ranges leads to an inaccurate representation of the block interval times.What did you expect to happen?
The histogram for the
block_interval_seconds
metric should use buckets that align closely with the typical block times observed in Celestia, providing a detailed and accurate representation of block intervals.How to reproduce it
block_interval_seconds
metric under normal operation.+Inf
histogram buckets, and no blocks get counted in any of the buckets below 10.Logs
n/a
dump_consensus_state
outputn/a
Anything else we need to know
The proposed trivial fix involves adjusting the histogram bucket ranges to more appropriately reflect the observed block times in Celestia. Suggested bucket configuration is as follows:
This setup provides finer granularity around the expected median block time of approximately 12 seconds, improving monitoring and analysis capabilities.
Tasks
The text was updated successfully, but these errors were encountered: