跳到主要内容

Troubleshooting

Runner prints error 404​

How to fix​

Getting 401 error code from Runner

Gathering Facts issue for localhost​

The issue can occur on Semaphore UI installed via Snap or Docker.

4:10:16 PM
TASK [Gathering Facts] *********************************************************
4:10:17 PM
fatal: [localhost]: FAILED! => changed=false

Why this happens​

For more information about localhost use in Ansible, read this article Implicit 'localhost'.

Ansible tries to gather facts locally, but Ansible is located in a limited isolated container which doesn't allow this.

How to fix this​

There are two ways:

  1. Disable facts gathering:
- hosts: localhost
  gather_facts: False
  roles:
    - ...
  1. Explicitly set the connection type to ssh:
[localhost]
127.0.0.1 ansible_connection=ssh ansible_ssh_user=your_localhost_user

panic: pq: SSL is not enabled on the server​

This means that your Postgres doesn't work by SSL.

How to fix this​

Add option sslmode=disable to the configuration file:

	"postgres": {
		"host": "localhost",
		"user": "postgres",
		"pass": "pwd",
		"name": "semaphore",
		"options": {
			"sslmode": "disable"
		}
	},

fatal: bad numeric config value '0' for 'GIT_TERMINAL_PROMPT': invalid unit​

This means that you are trying to access a repository over HTTPS that requires authentication.

How to fix this​

  • Go to Key Store screen.
  • Create a new key Login with password type.
  • Specify your login for GitHub/BitBucket/etc.
  • Specify the password. You can't use your account password for GitHub/BitBucket, you should use a Personal Access Token (PAT) instead of it. Read more here.
  • After creating the key, go to the Repositories screen, find your repository and specify the key.

Git clone or pull fails intermittently​

Task logs may show messages like Git pull failed (...), retrying in 2s followed by either success or a final failure after several attempts.

Why this happens​

The git server (GitHub, GitLab, Bitbucket, or a self-hosted instance) was temporarily unreachable, returned a transient HTTP error, or the network between Semaphore and the server had a brief outage. Semaphore retries clone and pull operations automatically before failing the task.

How to fix this​

  1. Transient outages: Usually resolve on their own. Semaphore retries up to git_attempts times (default 4) with exponential backoff between attempts.
  2. Frequent failures: Increase the attempt budget in your configuration:
{
  "git_attempts": 8
}

Or with an environment variable:

export SEMAPHORE_GIT_ATTEMPTS=8
  1. Immediate, consistent failures: Retries will not help. Check repository URL, branch name, access keys, and network connectivity from the Semaphore server or runner host.

See Git operations for details on git_client and git_attempts.

Bash script output is missing or incomplete​

A Bash task finishes successfully but the log shows little or no output from echo, printf, or other commands — especially when the script exits quickly.

Why this happens​

Semaphore captures stdout and stderr from shell commands while they run. Very short scripts can finish before all buffered output is read, so the last lines may be dropped from the task log.

How to fix this​

  1. Upgrade: Recent Semaphore versions drain process output before marking a task complete. Update server and runners if you are on an older release.
  2. Flush output in the script when you need guaranteed delivery:
#!/bin/bash
echo "Starting deploy"
echo "Done" >&2

For critical diagnostics, write to a file inside the repository workspace and cat it at the end of the script. 3. Avoid silent early exit: Use set -euo pipefail and explicit error messages so failures are visible even when output is brief.

unable to read LDAP response packet: unexpected EOF​

Most likely, you are trying to connect to the LDAP server using an insecure method, although it expects a secure connection (via TLS).

How to fix this​

Enable TLS in your config.json file:

...
"ldap_needtls": true
...

LDAP Result Code 49 "Invalid Credentials"​

You have the wrong password or binddn.

How to fix this​

Use ldapwhoami tool and check if your binddn works:

ldapwhoami\
  -H ldap://ldap.com:389\
  -D "CN=/your/ldap_binddn/value/in/config/file"\
  -x\
  -W

It will ask interactively for the password and should return code 0 and echo out the DN as specified.

You also can read the following articles:

LDAP Result Code 32 "No Such Object"​

Coming soon.