Hi

I get this when creating database with the module:

Error: /Stage[main]/Main/Influxdb::Database[librenms]/Exec[create_database_librenms]: Could not evaluate: Could not find command 'influx'

Works in second run though.

Puppet looks like this:

class { 'influxdb': }

influxdb::database { 'librenms':
  ensure  => present,
}

I tried adding a require on the database for the class, but I got the same result, so I'm probably missing some resource that isn't explicitly referred to here.

Yum influxdb now delivers influxdb v2 as latest and module fails to install it (influx -execute quit command does not exists in v2).
A new parameter to specify the version to install would be nice to have.

I tried out the current git revision to enable https, but the module failed to start InfluxDB due to a misconfiguration with graphite:

class { 'influxdb':
  admin_enable   => true,
  admin_username => $influxdb_admin_username,
  admin_password => $influxdb_admin_password,
  domain_name    => 'whatever', # affects ssl cert paths
  https_enable   => true
}

Logs:

-- Logs begin at Tue 2019-09-10 08:45:40 UTC, end at Tue 2019-09-10 10:45:40 UTC. --
Sep 10 09:00:39 hostdata-01 systemd[1]: Started InfluxDB is an open-source, distributed, time series database.
Sep 10 09:00:40 hostdata-01 influxd[18551]: run: invalid graphite config: no measurement in template `server`. To generate a valid configuration file run `influxd config > influxdb.generated.conf`
Sep 10 09:00:40 hostdata-01 systemd[1]: influxdb.service: Main process exited, code=exited, status=1/FAILURE
Sep 10 09:00:40 hostdata-01 systemd[1]: influxdb.service: Failed with result 'exit-code'.
Sep 10 09:00:41 hostdata-01 systemd[1]: influxdb.service: Service hold-off time over, scheduling restart.
Sep 10 09:00:41 hostdata-01 systemd[1]: influxdb.service: Scheduled restart job, restart counter is at 1.
Sep 10 09:00:41 hostdata-01 systemd[1]: Stopped InfluxDB is an open-source, distributed, time series database.

Removing the [[graphite]] section in influxdb.conf makes the service work.

Hi,
Would it be possible to make the module compatible with Puppet 8x ?

For example, currently :

Error: Could not retrieve catalog from remote server: Error 500 on SERVER: Server Error: Evaluation Error: Unknown variable: '::operatingsystem'. (file: /etc/puppetlabs/code/environments/flp_puppet8/modules/influxdb/manifests/params.pp, line: 37, column: 8) on node cladev-ubu2004.adm.fr.clara.net
Warning: Not using cache on failed catalog
Error: Could not retrieve catalog; skipping run


Thanks.

I use no graphite. But even this module generates graphite-enabled=false the influxdb considers the config beeing invalid and won't start.

I use master branch of the module

The signing key used by the packages for the Yum repository has changed: (https://www.influxdata.com/blog/linux-package-signing-key-rotation/).

This module will setup the influx.repo file using the old key.

Quick fix is to change templates/influxdb.repo.erb

gpgkey = https://repos.influxdata.com/influxdata-archive_compat.key

Though maybe that element could also be parameterized with that URL as a default.

The 2020-03-13 merge of 5bdf872 adds support for changing TSM engine settings. The values of these settings should be in double quotes in the templated config in case they use suffixes (k, m, g for *-memory-size; m, h for *-duration).

This bug is reproducible in the default config (tested on Ubuntu 18.04): influxd fails to start with log message "run: parse config: Near line 76 (last key parsed 'data'): expected a top-level item to end with a newline, comment, or EOF, but got 'm' instead". Line 76 is

cache-snapshot-write-cold-duration = 10m

Changing that to

cache-snapshot-write-cold-duration = "10m"

fixes the problem.

I have a trivial patch to do this for all four relevant settings, which I will commit tagged to this issue and open a PR for shortly.

Error: Error: /Stage[main]/Influxdb/Exec[wait_for_influxdb_to_listen]: Could not evaluate: Could not find command 'influx'

Verified that the command is not there

 # dpkg -L influxdb | grep influx
 /etc/influxdb
 /etc/influxdb/influxdb.conf
 /etc/init.d/influxdb
 /etc/logrotate.d/influxdb
 /lib/systemd/system/influxdb.service
 /usr/bin/influxd
 /usr/share/doc/influxdb
 /usr/share/doc/influxdb/NEWS.Debian.gz
 /usr/share/doc/influxdb/changelog.Debian.amd64.gz
 /usr/share/doc/influxdb/changelog.Debian.gz
 /usr/share/doc/influxdb/changelog.gz
/usr/share/doc/influxdb/copyright
/usr/share/man/man8/influxd.8.gz
/var/lib/influxdb
/var/log/influxdb

It seems like the version from the Debian repo was installed and not that from the Influx repo.

# apt policy influxdb
influxdb:
  Installed: 1.6.4-1+b22
  Candidate: 1.8.3-1
  Version table:
     1.8.3-1 500
        500 https://repos.influxdata.com/debian buster/stable amd64 Packages
 *** 1.6.4-1+b22 500
        500 http://ftp.stratoserver.net/pub/linux/debian buster/main amd64 Packages
        100 /var/lib/dpkg/status

After upgrading the influx package wiht # apt upgrade everything looks good:

Notice: /Stage[main]/Influxdb/Exec[wait_for_influxdb_to_listen]/returns: executed successfully

There is notify => Exec['apt_update'] in place so it looks like notify and/or Exec are not blocking the require => Class['influxdb::repos'], for package { $package_names:... in init.pp so InfluxDB is installed before the packages in the Influx repo are known to apt.

I haven't found a proper solution for this. I will submit a MR but it is ugly by all means and I hesitated to even upload it. Maybe somebody can come up with a better approach.

Would it be possible to add a resource for creating influxdb retention policies?

Ansible has an influxdb_retention_policy module that has this functionality.

I'm currently using this module to setup a Telegraf database for system metrics, but I'm concerned the database will fill the disk after enough time passes.

Currently it's not possible to enable FLUX queries. Would be nice to add it.
https://docs.influxdata.com/influxdb/v1.7/administration/config/#flux-enabled-false

It is possible to end up with an old version of influxdb on debian/ubuntu if an apt update isn't run between adding the repo and installing the package. This is quite common actually.

I had a quick look around and the puppet telegraf module fixes this by adding:

    Class['apt::update'] -> Package[$telegraf::package_name]

Right after defining the apt::source resource.

The dependencies for this module require an outdated version of puppetlabs/apt ( >= 4.5.1 <= 7.1.0), making it difficult to install since the current version is 7.6.0

This module (v1.5.0) creates a /etc/apt/sources.list.d/influxdb.list with a hard-coded release name 'buster' even though my machine is running 'bullseye':

cat influxdb.list
# This file is managed by Puppet. DO NOT EDIT.
# influxdb
deb https://repos.influxdata.com/debian buster stable

Instead of hard-coding the value I think you should use e.g. $::lsbdistcodename
(In my opinion this is also better than using 'stable')

Also:
If you browse to https://repos.influxdata.com/debian/ you see instructions to generate a trusted gpg key to go with the apt source.

Also:
The hard-coded apt source is missing the 'main' dir:
deb https://repos.influxdata.com/debian buster stable
should look like this:
deb https://repos.influxdata.com/debian <release> main

Also:
The current version (v1.5.0) seem to try to install an old gpg key (05CE15085FC09D18E99EFB22684A14CF2582E0C5) and it also use the deprecated 'apt-key' command.
Error: Execution of '/usr/bin/apt-key adv --no-tty --keyserver keyserver.ubuntu.com --recv-keys 05CE15085FC09D18E99EFB22684A14CF2582E0C5' returned 2: Warning: apt-key is deprecated. Manage keyring files in trusted.gpg.d instead
The module should install the new key directly to /etc/apt/trusted.gpg.d/influxdata-archive_compat.gpg

unix-socket-enabled is not configurable. It's a fixed setting commented out in influxdb.conf.erb

  # Enable http service over unix domain socket
  # unix-socket-enabled = false

MR will follow but there is a related topic to discuss. The default setting for http_bind_socket is /var/run/influxdb.sock This is the same default influxdb itself uses, but it is not working for most installations. Usually the influxdb user cannot write to /var/run/ so fails to create the socket and exits with run: open server: open service: listen unix /var/run/influxdb.sock: bind: permission denied

I doubt that influxdb would accept a braking change to change the default. Because this puppet module did not support sockets so far, this could be changed here without breaking existing installations.

This is a decision of keeping defaults consistent vs. providing working defaults. What is your opinion?

Since you have a Puppetfile tools like librarian-puppet will try to force-install stdlib 4.25.0 and apt 4.5.1 even though in the metadata.json is says otherwise. This breaks our deployment.

By default in this module InfluxDB runs as the user influxDB. If you enable HTTPS and set the private key using your company's domain name (which is required with the template influxdb.conf.erb with the following line) then the key has to be in /etc/ssl/private/ which by default only root can access.
https-private-key = "/etc/ssl/private/<%= @domain_name %>.key"

I think the best option would be to add a new parameter that is a full path to the private key so users can set a private key that influxdb can access and then change the template line to something like
https-private-key = "<%= @ssl_private_key %>"

latest version on puppetforge is 1.3.2 but in the repo its 1.3.4 :)

Installing this module fails when using a modern puppetlabs/apt module since the include_src option changed syntax. I removed the option in my installation since a new apt module also defaults to false.

Not sure what version you want to target of puppetlabs/apt. The current "version_requirement": ">= 1.5.0 <= 7.0.0" in metadata.json is confusing since the latest released version is only 4.5.1.