Normal behavior for Conjur Standbys

We’re attempting to set up a healthy cluster and have configured and synced our standbies. We want to confirm that it’s normal for standbies to:

  1. Return a 502 when accessed via UI
  2. To return the following errors on health check:
    Version:0.9 StartHTML:0000000105 EndHTML:0000000972 StartFragment:0000000216 EndFragment:0000000940 “services”: {
    “possum”: “unknown error - Failed to open TCP connection to localhost:5000 (Address family not supported by protocol - socket(2) for “localhost” port 5000)”,
    “ui”: “unknown error - Failed to open TCP connection to localhost:3000 (Address family not supported by protocol - socket(2) for “localhost” port 3000)”,
    “ok”: false
    }

Any help or insight would be appreciated! Thank you in advance.

Hi NRSayed,

A standby is not serving any requests, and thus the services (aside from pg) are not running. Thus the port binding errors you’re seeing in the logs. We return a 502 response code so when you setup your load balancer in front of the cluster, the load balancer will only get a 200 response from the active master and can properly route traffic to it and not the standbys. So yes, both of these behaviors are expected for standbys.

Nate

2 Likes