Troubleshoot issues based on the error message. Common errors result from failed actions, timeouts, multiple entries, bad and expired certificates, invalid characters, syntax parsing, malformed responses, and exceeded deadlines.
This topic describes common messages that may appear when installing and running Consul. Errors usually they indicate an issue in your network or in your server's configuration. Refer to the [Troubleshooting Guide][troubleshooting] for help resolving error messages that do not appear on this page.
Your server has multiple active network interfaces. Consul needs to know which interface to use for local LAN communications. Add the [`bind`][bind] option to your configuration.
-> **Tip**: If your server does not have a static IP address, you can use a [go-sockaddr template][go-sockaddr] as the argument to the `bind` option, e.g. `"bind_addr": "{{GetInterfaceIP \"eth0\"}}"`.
There is a syntax error in your configuration file. If the error message doesn't identify the exact location in the file where the problem is, try using [jq] to find it, for example:
If the Consul client and server are on the same LAN, then most likely, a firewall is blocking connections to the Consul server.
If they are not on the same LAN, check the [`retry_join`][retry_join] settings in the Consul client configuration. The client should be configured to join a cluster inside its local network.
These error messages indicate a general performance problem on the Consul server. Make sure you are monitoring Consul telemetry and system metrics according to our [monitoring guide][monitoring]. Increase the CPU or memory allocation to the server if needed. Check the performance of the network between Consul nodes.
On a busy cluster, the operating system may not provide enough file descriptors to the Consul process. You will need to increase the limit for the Consul user, and maybe the system-wide limit as well. A good guide for Linux can be found [here][files].
Or, if you are starting Consul from `systemd`, you could add `LimitNOFILE=65536` to the unit file for Consul. You can see our example unit file [here][systemd].
Our RPC protocol requires support for a TCP half-close in order to signal the other side that they are done reading the stream, since we don't know the size in advance. This saves us from having to buffer just to calculate the size.
If a host does not properly implement half-close you may see an error message `[ERR] consul: Failed to close snapshot: write tcp <source>-><destination>: write: broken pipe` when saving snapshots. This should not affect saving and restoring snapshots.
This has been a [known issue](https://github.com/docker/libnetwork/issues/1204) in Docker, but may manifest in other environments as well.
This indicates that you have ACL enabled in your cluster, but you aren't passing a valid token. Make sure that when creating your tokens that they have the correct permissions set. In addition, you would want to make sure that an agent token is provided on each call.
Make sure that your Consul clients and servers are using the correct certificates, and that they've been signed by the same CA. The easiest way to do this is to follow [our guide][certificates].
If you generate your own certificates, make sure the server certificates include the special name `server.dc1.consul` in the Subject Alternative Name (SAN) field. (If you change the values of `datacenter` or `domain` in your configuration, update the SAN accordingly.)
You have installed an Enterprise version of Consul. If you are an Enterprise customer, [provide a license key][license] to Consul before it shuts down. Otherwise, install the open-source Consul binary instead.
-> **Note:** Enterprise binaries can be identified on our [download site][releases] by the `+ent` suffix.
You may receive a `RESOURCE_EXHAUSTED` error from the Consul server if the maximum number of read or write requests per second have been reached. Refer to [Set a global limit on traffic rates](/consul/docs/agent/limits/set-global-traffic-rate-limits) for additional information. You can retry another server unless the number of retries is exhausted. If the number of retries is exhausted, you should implement an exponential backoff.
The RESOURCE_EXHAUSTED RPC response is translated into a `429 Too Many Requests` error code on the HTTP interface.
The server may respond as `UNAVAILABLE` if it is the leader node and the global write request rate limit is reached. The solution is to apply an exponential backoff until the leader has capacity to serve those requests.
The `UNAVAILABLE` RPC response is translated into a `503 Service Unavailable` error code on the RPC requests sent through HTTP interface.
