can't purge cron resources

[vchepkov]

while attempting removing 'unmanaged' cron entries, module recognizes those, but fails to remove

how to reproduce:

# crontab -u user -l
0 0 * * * /bin/echo test

puppet resource reports it as 'unmanaged'

# puppet resource cron 
cron { 'unmanaged:/bin/echo_test-1':
  ensure   => 'present',
  command  => '/bin/echo test',
  hour     => ['0'],
  minute   => ['0'],
  provider => 'crontab',
  target   => 'user',
  user     => 'user',
}

Code to apply:

purge { 'cron':
  if => [ 'user', '==', 'user' ],
}

Result of puppet apply:

Notice: /Stage[main]/Main/Purge[cron]/ensure: ensure changed 'purgable' to 'purged'
Notice: /Stage[main]/Main/Cron[unmanaged:/bin/echo_test-1]/ensure: removed

But cron entry remains intact.

Thanks,
Vadym

[lelutin]

Hello,

I've just hit this while trying to use the module.

I'm able to get cronjobs purged for the root user using a snippet similar to @vchepkov 's, but any for other user the cronjobs get marked as getting removed but they just stay there in the crontab file like what is described by OP.

[lelutin]

I'm trying to figure out what's present in the catalog, but with with a request of the following type against puppetdb, I'm not seeing a resource for the cronjob that's not getting removed:

catalogs { certname = "node-name.fqdn" }

nothing in the output is mentioning "unmanaged:" or "echo" (which is part of the test cronjob that should get removed).

I don't know why that resource is not in the catalog but visibly getting processed by the agent

[lelutin]

ah, sorry for my confusion in the previous message. I've read the type code and I can see the resources are injected by the purge type. if I look at the reports (instead of the catalogs) for that same machine, I can see that puppet claims to have removed the resource:

      {
        "status": "success",
        "timestamp": "2021-08-20T12:33:12.795-04:00",
        "resource_type": "Cron",
        "resource_title": "unmanaged:/usr/bin/echo_blah_toisssi-1",
        "property": "ensure",
        "corrective_change": null,
        "new_value": "absent",
        "old_value": "present",
        "message": "removed",
        "file": null,
        "line": null,
        "containment_path": [
          "Stage[main]",
          "Profile::Cron",
          "Cron[unmanaged:/usr/bin/echo_blah_toisssi-1]"
        ],
        "containing_class": "Profile::Cron"
      },

as stated before though, it's still present in the crontab file for the www-data user.

[vchepkov]

It's being awhile, anyone was able to find a solution?

[vchepkov]

I have inspected the res object in generate function and I found that

res[:user] -> nil
res[:target] -> "root"

I believe both are wrong and should be set to an actual user

[vchepkov]

FYI, module fails to remove previously managed entries, i.e. those that have Puppet Name: prefix