Page MenuHomePhabricator

Create tool that displays real-time load balancer health status per pool/node
Open, MediumPublic

Description

The ongoing OpenSearch migration (T388610) requires a lot of changes to our load balancer pools.

We have accidentally triggered alerts by forgetting to depool nodes before a reimage. We might also have nodes failing health checks because they haven't properly joined the cluster, or for any number of reasons. A real-time view of each node's health status would help us triage these issues quickly, hopefully before they generate alerts for other teams.

Creating this ticket to:

  • Find the best source of this information:
  • (@Volans pointed me to this alerts dashboard)
  • @dcausse pointed me to this PyBal dashboard which does almost everything we want
  • Use the information to create a tool (probably a dashboard) that displays real-time (or as near as is practical) information on node status per pool.

Event Timeline

bking triaged this task as Medium priority.
bking renamed this task from Create tool that displays real-time load balancer health status per pool/host to Create tool that displays real-time load balancer health status per pool/node.May 19 2025, 2:14 PM
bking updated the task description. (Show Details)
bking updated the task description. (Show Details)
bking updated the task description. (Show Details)
bking added a subscriber: Volans.
bking added a subscriber: dcausse.

@gmodena , @Lydia_Pintscher and I have been discussing something similar in Slack .

Config-master's pools.json has the information we need; we just need a convenient way to get it into our dashboards. Long-term, we could do that with the grafana json datasource plugin , but that would have to be approved by Observability. Absent that, we'd probably just link it from a description panel (assuming we can put HTML in descriptions) and/or link from a Wikitech doc.