search

celestiaorg / celestia-node

Celestia Data Availability Nodes
Apache License 2.0
934 stars 931 forks source link

das: Better metric for showing sampled range now that SamplingWindow is enabled #3201

Open renaynay opened 9 months ago

renaynay commented 9 months ago

Right now, sampled chain head refers to the tip of the sampled chain which, while accurate, does not reflect that the light node may not have sampled everything before the SamplingWindow.

There should be some sort of metric that reflects where the light node began sampling potentially - or some way to know whether a light node has actually triggered a sampling job over block at height N.

Needs to be fleshed out a bit more.

(from discussions with @jcstein)

jcstein commented 9 months ago

I think #3196 should accomplish part of this, but may be wrong. cc @ramin

ramin commented 9 months ago

@jcstein this wont quite be solved by #3196, this is issue is regarding telemetry of metrics we'd instrument in the code and which node operators could optionally collect and visualize to show metrics about the operation of a running node

renaynay commented 6 months ago

Now that https://github.com/celestiaorg/celestia-node/issues/3194 is resolved, this one needs to be as well.

walldiss commented 6 months ago

It should instead be a metric on the pruner. DASer is unaware of where or whether the latest pruned height is. In fact, it already exists as prnr_last_pruned (the pruner's highest pruned height). Therefore, the range of stored samples can be easily determined between the DASer head and this metric.