Anthony Woods
Anthony Woods
This change also ensures that OnIdle() is only called when MaxRTT is reached and no packet has been received.
This all sounds excellent to me. I dont see any obvious issues in the suggested changes and to me it looks like it will solve most of the caching inefficiencies...
> when importing, we could just restart MT, @replay can we add an API method to clear the cache? It would also be really handy for resolving any cache issues...
In dockerized environments, running commands or looking at package versions inside a container is not always an option.
We also need to send new aggregated points through Kafka rather then adding directly to a chunk. This will enable us to rebuild the unsaved aggregated chunks at startup by...
> crng should have the ability to publish internal metrics to a dedicated endpoint on tsdb-gw i much prefer this method of exposing crng performance stats over trying to intercept...
For kafka, cassandra based on our own experience we can make the following suggestions **Kafka:** minimum 3 nodes. 8core 32GB ram, 1TB SSD disk. With 3-way replication, each node can...
looks like this is the same problem as #304
In order to be able to collect certain metrics, the agent runs with "hostNetwork: true". Due to using the hostNetwork , the pod running the agent will use the DNS...
This is a DUP of https://github.com/grafana/kubernetes-app/issues/38