You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the exporter doesn't expose stats regarding the indexing pressure. This information would be quite valuable when it comes to having the ability to automatically detect when one ore many instances are unable to accept indexing requests and consequently returning 429 errors. These statics can be obtained from the Node Stats API, specifically with indexing_pressure metric.
For example, some of these useful resulting metrics would include:
I haven't considered the most appropriate structure with regards to labels although it should be mostly similar to the logic used in the current nodes info collector. You could potentially go one step further and have a configuration flag to enable each possible metric exposed by the node stats API.
The text was updated successfully, but these errors were encountered:
Currently, the exporter doesn't expose stats regarding the indexing pressure. This information would be quite valuable when it comes to having the ability to automatically detect when one ore many instances are unable to accept indexing requests and consequently returning 429 errors. These statics can be obtained from the Node Stats API, specifically with
indexing_pressure
metric.For example, some of these useful resulting metrics would include:
nodes.<NODE_ID>.indexing_pressure.memory.total.primary_rejections
nodes.<NODE_ID>.indexing_pressure.memory.total.replica_rejections
I haven't considered the most appropriate structure with regards to labels although it should be mostly similar to the logic used in the current nodes info collector. You could potentially go one step further and have a configuration flag to enable each possible metric exposed by the node stats API.
The text was updated successfully, but these errors were encountered: