You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When running Consul Connect using a managed proxy, the bind address for the proxy is set to what's advertised to the cluster members as the address to reach the agent with:
bind=10.120.0.254
will have the proxy bound to 10.120.0.254.
However, when -bind is not used, but rather -client and -advertise, the proxy listen address defaults to 0.0.0.0, which will never be reachable by any health check or client request.
After discussing this with @banks he mentioned that it would be better to default to 127.0.0.1 instead, or the first usable private IP(v4|v6) address, as done with the other bind assignment statements.
Reproduction Steps
Steps to reproduce this issue, eg:
Run a cluster with a single server and agent
Bind the agent's client interface to 0.0.0.0, -advertise to the cluster with the first IP(v4|v6) address
Not fixed yet but there is a work-around: if you manually specify the proxy-bind address you can make it what you want. This is only an issue if you rely on it default to the agent's bind address and that happens to be 0.0.0.0.
Note that it's not possible until we fix this to have proxy bind to 0.0.0.0 as it's health check will fail. The fix is trivial though and should be out within a few weeks.
Overview of the Issue
When running Consul Connect using a managed proxy, the bind address for the proxy is set to what's advertised to the cluster members as the address to reach the agent with:
will have the proxy bound to
10.120.0.254
.However, when
-bind
is not used, but rather-client
and-advertise
, the proxy listen address defaults to0.0.0.0
, which will never be reachable by any health check or client request.After discussing this with @banks he mentioned that it would be better to default to
127.0.0.1
instead, or the first usable private IP(v4|v6) address, as done with the other bind assignment statements.Reproduction Steps
Steps to reproduce this issue, eg:
0.0.0.0
,-advertise
to the cluster with the first IP(v4|v6) addressConsul info for both Client and Server
Client info
Server info
Operating system and Environment details
Running inside a container using Alpine Linux 3.7 and the latest binaries from
releases.hashicorp.com
.Log Fragments
The text was updated successfully, but these errors were encountered: