Hello team, we want to use <Pool-Based Instance As...
# troubleshooting
r
Hello team, we want to use Pool-Based Instance Assignment in our Pinot Cluster. Since it requires to add Helix InstanceConfig to each server. Is there any way that we can populate the Helix InstanceConfig via some config at server startup. The reason behind this ask is, we want to avoid any manual task in our prod cluster in case if we are scaling servers or replenishing the existing servers. We could only find below ways to add Helix InstanceConfig: • Via controller UI (manual) • Via pinot-admin.sh (needs to be run after server is up) • REST API (needs to be run after server is up) We could not find any method that can be given as server configuration and it gets picked automatically when server boots up. Please suggest. Also, we have deployed Pinot in k8s using official Helm Chart. And we want to maintain 2 or 3 pools of servers. What would be your recommendation, should we mange each server pool as different statefulset or all should be part of single statefullset only.
m
Those are the ways to set up, currently
r
Thanks @Mayank, Also via REST API, we could only see the way to update tags, how to update the pool configuration via it? Any doc around that?
y
There are controller APIs
GET /instances/{instanceName}
and
PUT /instances/{instanceName}
that allow retrieving and updating instance configs.
m
That's tehe REST API right?
y
Yeah, I was answering this question:
Also via REST API, we could only see the way to update tags, how to update the pool configuration via it
The request body for
PUT /instances/{instanceName}
can contain the pool configuration like:
Copy code
{
  "pools": {
    "DefaultTenant_OFFLINE": 1
  }
}
m
Ah, thanks
r
Thanks Yash.