feat(core): Support for partitions #343

New issue

Closed

opened 2026-03-28 04:23:40 +00:00 by mfreeman451 · 0 comments

mfreeman451 commented

2026-03-28 04:23:40 +00:00

Owner

Imported from GitHub.

Original GitHub issue: #994
Original author: @mfreeman451
Original URL: https://github.com/carverauto/serviceradar/issues/994
Original created: 2025-06-20T03:50:19Z

We are currently building device ids like this:

ip-address:agent_id:poller_id

Besides the obvious issue that would present itself with IPv6 addressing, if we find the same device but it is being accessed through a different agent_id or poller_id, it will show up as a duplicate.

We need to introduce the concept of partitions to replace the current device_id naming strategy, we could probably just do "partition:ip address" and be able to achieve the same goal, and also support IPv6 by looking at everything after the first :.

Maybe the most obvious place to configure the 'partition' ID, would be at the poller level, where we define the partition in the poller.json config, and then we enrich messages we receive from the agent before we send them up to the serviceradar-core for further processing.

Imported from GitHub. Original GitHub issue: #994 Original author: @mfreeman451 Original URL: https://github.com/carverauto/serviceradar/issues/994 Original created: 2025-06-20T03:50:19Z --- We are currently building device ids like this: * ip-address:agent_id:poller_id Besides the obvious issue that would present itself with IPv6 addressing, if we find the same device but it is being accessed through a different agent_id or poller_id, it will show up as a duplicate. We need to introduce the concept of partitions to replace the current device_id naming strategy, we could probably just do "partition:ip address" and be able to achieve the same goal, and also support IPv6 by looking at everything after the first `:`. Maybe the most obvious place to configure the 'partition' ID, would be at the poller level, where we define the `partition` in the poller.json config, and then we enrich messages we receive from the agent before we send them up to the serviceradar-core for further processing.