elsmorian
commented
4 years ago Just a suggestion - the cluster page can be a great resource for an overview of the health of your nodes, so the ability to see the whole list of nodes across the board can be really helpful. We have a smaller number of nodes (~20) so maybe the page can either automatically scale when the number of nodes starts to get too many, or there could be some setting to turn it on explicitly for users of smaller clusters?
Project board link
The current "cluster" page doesn't give much valuable information and is too bloated when dealing with large clusters (it gets heavy on the browser). We need to rework this screen and display a more compact list of clusters, without displaying the whole list of nodes throughout the fleet.We also need to provide additional informations if the sidecar mode is being used, as we have metrics being polled to the database, and easily available even for large clusters. Displaying the number of running/pending compactions, p99 latencies, ... would be interesting. It would also be good to grab the load average of the servers, store them in the db and use the information to display some cluster health metrics.
┆Issue is synchronized with this Jira Story by Unito