Common errors

This section covers the most common errors you might encounter when connecting resources with Cloudflare Tunnel. If you do not see your issue listed below, refer to the troubleshooting FAQ, view your Tunnel logs, or contact Cloudflare Support.

I see cloudflared service is already installed.

If you see this error when installing a remotely-managed tunnel, ensure that no other cloudflared instances are running as a service on this machine. Only a single instance of cloudflared may run as a service on any given machine. Instead, we recommend adding additional routes to your existing tunnel. Alternatively, you can run sudo cloudflared service uninstall to uninstall cloudflared.

I see An A, AAAA, or CNAME record with that host already exists.

If you are unable to save your tunnel’s public hostname, choose a different hostname or delete the existing DNS record. Check the DNS records for your domain from the Cloudflare dashboard ↗.

Tunnel credentials file doesn’t exist or is not a file.

If you encounter the following error when running a tunnel, double check your config.yml file and ensure that the credentials-file points to the correct location. You may need to change /root/ to your home directory.

cloudflared tunnel run

2021-06-04T06:21:16Z INF Starting tunnel tunnelID=928655cc-7f95-43f2-8539-2aba6cf3592d
Tunnel credentials file '/root/.cloudflared/928655cc-7f95-43f2-8539-2aba6cf3592d.json' doesn't exist or is not a file

My tunnel fails to authenticate.

To start using Cloudflare Tunnel, a super administrator in the Cloudflare account must first log in through cloudflared login. The client will launch a browser window and prompt the user to select a hostname in their Cloudflare account. Once selected, Cloudflare generates a certificate that consists of three components:

• The public key of the origin certificate for that hostname
• The private key of the origin certificate for that domain
• A token that is unique to Cloudflare Tunnel

Those three components are bundled into a single PEM file that is downloaded one time during that login flow. The host certificate is valid for the root domain and any subdomain one-level deep. Cloudflare uses that certificate file to authenticate cloudflared to create DNS records for your domain in Cloudflare.

The third component, the token, consists of the zone ID (for the selected domain) and an API token scoped to the user who first authenticated with the login command. When user permissions change (if that user is removed from the account or becomes an admin of another account, for example), Cloudflare rolls the user’s API key. However, the certificate file downloaded through cloudflared retains the older API key and can cause authentication failures. The user will need to login once more through cloudflared to regenerate the certificate. Alternatively, the administrator can create a dedicated service user to authenticate.

I see an error: x509: certificate signed by unknown authority.

This means the origin is using a certificate that cloudflared does not trust. For example, you may get this error if you are using SSL/TLS inspection in a proxy between your server and Cloudflare. To solve this:

• Add the certificate to the system certificate pool.
• Use the --origin-ca-pool flag and specify the path to the certificate.
• Use the --no-tls-verify flag to stop cloudflared checking the certificate for a trust chain.

I see an error 1033 when attempting to run a tunnel.

An error 1033 indicates your tunnel is not connected to Cloudflare’s edge. First, run cloudflared tunnel list to see whether your tunnel is listed as active. If it isn’t, check the following:

1. Make sure you correctly routed traffic to your tunnel (step 5 in the Tunnel guide) by assigning a CNAME record to point traffic to your tunnel. Alternatively, check this guide to route traffic to your tunnel using load balancers.
2. Make sure you run your tunnel (step 6 in the Tunnel guide).

For more information, here is a comprehensive list of Cloudflare 1xxx errors.

I see ERR_TOO_MANY_REDIRECTS when attempting to connect to an Access self-hosted app.

This error occurs when cloudflared does not recognize the SSL/TLS certificate presented by your origin. To resolve the issue, set the origin server name parameter to the hostname on your origin certificate. Here is an example of a locally-managed tunnel configuration:

ingress:
  - hostname: test.example.com
    service: https://localhost:443
    originRequest:
      originServerName: test.example.com

cloudflared access shows an error websocket: bad handshake.

This means that your cloudflared access client is unable to reach your cloudflared tunnel origin. To diagnose this, you should look at the cloudflared tunnel logs. A very often root cause is that the cloudflared tunnel is unable to proxy to your origin (e.g. because the ingress is mis-configured, or the origin is down, or because the origin HTTPS certificate cannot be validated by cloudflared tunnel). If cloudflared tunnel has no logs, it means Cloudflare Edge is not even able to route the websocket traffic to it.

There are a few different possible root causes behind the websocket: bad handshake error:

• Your cloudflared tunnel is either not running or not connected to Cloudflare Edge.

• WebSockets are not enabled. To enable them, go to dash.cloudflare.com > Network.

• Your Cloudflare account has Universal SSL enabled but your SSL/TLS encryption mode is set to Off (not secure). To resolve:

  1. On the Cloudflare dashboard for your zone, go to SSL/TLS > Overview.
  2. Ensure that your SSL/TLS encryption mode is set to either Flexible, Full or Full (strict).

• Your requests are blocked by Super Bot Fight Mode. To resolve, make sure you set Definitely automated to Allow in the bot fight mode settings.

• Your SSH or RDP Access application has the Binding Cookie enabled. To disable the cookie, go to Access > Applications and edit the application settings.

My tunnel randomly disconnects.

Long-lived connections initiated through the Cloudflare Zero Trust platform, such as SSH sessions, can last up to eight hours. However, disruptions along the service path may result in more frequent disconnects. Often, these disconnects are caused by regularly scheduled maintenance events such as data center, server, or service updates and restarts. If you believe these events are not the cause of disconnects in your environment, collect the relevant WARP logs and Tunnel logs and contact Support.

Tunnel connections fail with SSL error.

If cloudflared returns error error="remote error: tls: handshake failure", check to make sure the hostname in question is covered by a SSL certificate. If using a multi-level subdomain, an advanced certificate may be required as the Universal SSL will not cover more than one level of subdomain. This may surface in the browser as ERR_SSL_VERSION_OR_CIPHER_MISMATCH.

Tunnel connections fail with Too many open files error.

If your Cloudflare Tunnel logs returns a socket: too many open files error, it means that cloudflared has exhausted the open files limit on your machine. The maximum number of open files, or file descriptors, is an operating system setting that determines how many files a process is allowed to open. To increase the open file limit, you will need to configure ulimit settings on the machine running cloudflared.

I see failed to sufficiently increase receive buffer size in my cloudflared logs.

This buffer size increase is reported by the quic-go library ↗ leveraged by cloudflared ↗. You can learn more about the log message in the quic-go repository ↗. This log message is generally not impactful and can be safely ignored when troubleshooting. However, if you have deployed cloudflared within a unique, high-bandwidth environment then buffer size can be manually overridden for testing purposes.

To set the maximum receive buffer size on Linux:

1. Create a new file under /etc/sysctl.d/:

sudo vi 98-core-rmem-max.conf

2. In the file, define the desired buffer size:

net.core.rmem_max=2500000

3. Reboot the host machine running cloudflared.

4. To validate that these changes have taken effect, use the grep command:

sudo sysctl -a | grep net.core.rmem_max

net.core.rmem_max = 2500000