Hi,

When Download and unarchive node_exporter into temporary location is skipped, Move node_exporter binary into place is still executed because a skipped task still counts as success. See ansible/ansible#25027 (comment) for reference.

This causes the Move node_exporter binary into place task to fail if the download job was skipped and the temporary file doesn't exist anymore.

PLAY [all] *************************************************************************************************************

TASK [Gathering Facts] *************************************************************************************************
ok: [host.example.org]

TASK [geerlingguy.node_exporter : Check current node_exporter version.] ************************************************
ok: [host.example.org]

TASK [geerlingguy.node_exporter : Download and unarchive node_exporter into temporary location.] ***********************
skipping: [host.example.org]

TASK [geerlingguy.node_exporter : Move node_exporter binary into place.] ***********************************************
fatal: [host.example.org]: FAILED! => {"changed": false, "msg": "Source /tmp/node_exporter-1.3.1.linux-amd64/node_exporter not found"}

The Determine latest GitHub release (local) call is subject to GitHub's default rate limits of 60 calls per hour (unauthenticated).

There is a workaround for this by pinning version to avoid this call but was wondering if other options should be explored such as a user passing in a API token so they can use their own limits?

I am using packer to make an image that includes the node exporter in a chroot, therefore the liveness check at the end is failing. Please consider adding an option to skip this check.

Referring to this task:

There are some issues around this check and the node_exporter_host variable. I can PR a fix, but want to think a little more what the best option to offer is.

The role's current default is to use localhost. One issue is that localhost will make node_exporter only listen on IPv4. If no --web.listen-address option is specified, the default is actually ":9100" (so an empty hostname and will listen on all interfaces):

ts=2022-08-31T02:23:56.106Z caller=node_exporter.go:199 level=info msg="Listening on" address=:9100

and this also listens on IPv4 and IPv6 (notice the tcp6 and :::9100 listener):

$ netstat -an | grep 9100
tcp6       0      0 :::9100                 :::*                    LISTEN    

Specifying the empty string for node_exporter_host will make the uri check in this task fail as http://:9100/ is not valid. Other valid example of --web.listen-address options are:

• --web.listen-address=0.0.0.0:9100. # alternate way to specify "listen on all interfaces"
• --web.listen-address='[::1]:9100' # Listen on IPv6 & IPv4 localhost

Maybe an easy workaround is to provide a role variable that makes this check optional?

Setting the package architecture at runtime, instead of defining it in a variable, might make the role easier to use. If you're interested, I can make a pull request for this feature.

In certain environments you can have users created with specific UIDs.
By setting the node_exporter user as system: true. It will use a dynamic
system uid instead.

I am passing in the var node_exporter_options: '--web.listen-address=:19100' which works fine but the task Verify node_exporter is responding to requests fails as the url is hardcoded. Could this be changed to a var instead ? Verified the task passes when I update the port to match what I pass in

By clicking on the cog beside About and adding hacktoberfest topic to the repo, any contributors can gain points for hacktoberfest. See more info here: https://hacktoberfest.digitalocean.com/

Thanks!

Is it possible to implement checksum verification for node exporter releases? Perhaps not by default but optionally in case users want to define an explicit version or use a custom value for node_exporter_download_url.

If considered a good idea, I can see if I can make a simple implementation of this feature.

Hi, I get this error

TASK [ansible-role-node_exporter : Verify node_exporter is responding to requests.] *******************************************************************************************************************************************************fatal: [mobile-admin.backend.ab.internal]: FAILED! => {"changed": false, "content": "", "elapsed": 30, "failed_when_result": true, "msg": "Status code was -1 and not [200]: Connection failure: timed out", "redirected": false, "status": -1, "url": "http://localhost:9100/"}
If I do restart node_exporter on the target server, the error does not appear after that

2 Issues..

Issue (1) ..Changing the node_options_does not get picked up unless we change the version

Steps To Reproduce

• Install version 1.6.1 of node-exporter

 - role: geerlingguy.node_exporter
      tags: node_exporter
      vars:
        node_exporter_version: 1.6.1
        node_exporter_options: '--collector.systemd"'

• Change the node_exporter_options

 - role: geerlingguy.node_exporter
      tags: node_exporter
      vars:
        node_exporter_version: 1.6.1
        node_exporter_options: '--collector.textfile.directory=/var/lib/node_exporter/textfile_collector --collector.systemd"'

• Rerun the playbook - the new options do not get pickedup

A workaround is to downgrade to 1.6.0 deploy to pick up the new node_exporter_options, and then redeploy 1.61.

However this leaves package 1.6.0 on disk in the /tmp folder so the redeployment is skipped leaving you with 1.6.0

• Removing the downloaded packages from /tmp allows you to redeploy v1.6.1 again

I use ansible==2.9.11.

I don't know since which version of node_exporter this is the case, but when using this role with version 1.1.2 the version details retrieved are written to stderr instead of stdout.

This results in always running the subsequent task ( Download and unarchive node_exporter into temporary location.) with the condition:

when: >
node_exporter_version_check.stdout is not defined
or node_exporter_version not in node_exporter_version_check.stdout

Also, the above condition is missing in task Move node_exporter binary into place. When nothing is downloaded to /tmp the folder won't be found with copy.

The task of "Determine latest GitHub release" that delegates the call to localhost is causing an unbalanced number of tasks.
e.g.
PLAY RECAP *********************************************************************
influxdb.themachine.art : ok=21 changed=0 unreachable=0 failed=0 skipped=3 rescued=0 ignored=0
prometheus.themachine.art : ok=22 changed=0 unreachable=0 failed=0 skipped=3 rescued=0 ignored=0

you can see that influxdb.themachine.art only has 21 oks as opposed to prometheus.themachine.art that has 22
it causes an ambiguity and uncertainty, it is also preventing me from running automations to check that ansible is running properly.

is it possible to have a configuration where I choose not to delegate to localhost or even not to run this task at all?

Thanks,
Roy

Hello! I am relatively new to Ansible.

In my playbook:

---
- name: Install node_exporter
  hosts: local
  vars_files:
    - "{{ inventory_dir }}/vars/vars.yaml"
  become: true

  roles:
  - role: geerlingguy.node_exporter

I store vars_files in "{{ inventory_dir }}/vars/vars.yaml" path. The file is encrypted and instantly decrypted on any playbook run via .vault_password.txt file which contain password to decrypt vars,yaml file.

Using your role gives error:

fatal: [<SOME IP ADDRESS> -> localhost]: FAILED! => {
    "msg": "'root_password' is undefined"
}

I checked logs with -vvv option when launch the playbook, and found behaviour that is strange for me:

TASK [geerlingguy.node_exporter : Configure latest version] **************************************
task path: /home/<SOME USER>/.ansible/roles/geerlingguy.node_exporter/tasks/main.yml:8
Read vars_file '{{ inventory_dir }}/vars/vars.yaml'
included: /home/<SOME USER>/.ansible/roles/geerlingguy.node_exporter/tasks/config-version.yaml for <SOME IP ADDRESS>
skipping vars_file '{{ inventory_dir }}/vars/vars.yaml' due to an undefined variable
Read vars_file '{{ inventory_dir }}/vars/vars.yaml'
skipping vars_file '{{ inventory_dir }}/vars/vars.yaml' due to an undefined variable

TASK [geerlingguy.node_exporter : Determine latest GitHub release (local)] ***********************
task path: /home/<SOME USER>/.ansible/roles/geerlingguy.node_exporter/tasks/config-version.yaml:2
fatal: [<SOME IP ADDRESS> -> localhost]: FAILED! => {
    "msg": "'root_password' is undefined"
}

While other playbooks catch variables file correctly, this one does not when using your role.

Where is the problem, exactly?

My playbook has just this

- role: geerlingguy.node_exporter

I've not overridden any of the default variable values.

Installation fails with

TASK [geerlingguy.node_exporter : Check current node_exporter version.] **********************************************************************************************************************************************************************
ok: [---removed---] => {"changed": false, "cmd": "/usr/local/bin/node_exporter --version", "failed_when_result": false, "msg": "[Errno 2] No such file or directory: b'/usr/local/bin/node_exporter'", "rc": 2, "stderr": "", "stderr_lines": [], "stdout": "", "stdout_lines": []}

TASK [geerlingguy.node_exporter : Download and unarchive node_exporter into temporary location.] *********************************************************************************************************************************************
ok: [---removed---] => {"changed": false, "dest": "/tmp", "gid": 0, "group": "root", "handler": "TgzArchive", "mode": "01777", "owner": "root", "size": 4096, "src": "/home/talonx/.ansible/tmp/ansible-tmp-1679037032.2916012-18457-279738429074661/node_exporter-0.16.0.linux-amd64.tarwia9p17m.gz", "state": "directory", "uid": 0}

TASK [geerlingguy.node_exporter : Move node_exporter binary into place.] *********************************************************************************************************************************************************************
fatal: [---removed---]: FAILED! => {"changed": false, "msg": "Source /tmp/node_exporter-0.18.1.linux-amd64/node_exporter not found"}

PLAY RECAP ***********************************************************************************************************************************************************************************************************************************
---removed---               : ok=3    changed=0    unreachable=0    failed=1    skipped=0    rescued=0    ignored=0   

It seems to be attempting to download 0.16.0 but trying to unpack 0.18.1

NAME="Debian GNU/Linux"
VERSION_ID="11"
VERSION="11 (bullseye)"
VERSION_CODENAME=bullseye
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"

Ansible role version is 1.0.0