[Labs-l] displaying puppet status for each host

Petr Bena benapetr at gmail.com
Thu Jan 30 16:05:35 UTC 2014


On Thu, Jan 30, 2014 at 2:57 PM, Ryan Lane <rlane32 at gmail.com> wrote:
> On Thu, Jan 30, 2014 at 2:06 AM, Petr Bena <benapetr at gmail.com> wrote:
>>
>> Hi,
>>
>> This is something I am now trying to set up on my own servers but it
>> would be useful on labs too.
>>
>> I recently started using puppet on my servers too (I like it, unlike
>> on labs where I rather hate it - because using puppet when you need to
>> wait weeks / months for every simple config change to get merged and
>> applied is true nightmare of every sysadmin), however what I am having
>> troubles with now, is how to check what the puppet status of each node
>> is, other than sshing there and checking puppet agent logs.
>>
>
> Why start off with puppet now? Why not go with salt or ansible and save
> yourself the trouble of eventually migrating?
>

I never heard of salt and such. Is that even better than puppet?

>>
>> Is there some nice web GUI or something like that, which would display
>> status of every node, errors, warnings etc. So that I could have easy
>> to reach overview of all nodes managed by puppet? I think this would
>> be extremely helpful on labs as well. Nagios can display puppet
>> freshness, but that isn't very much.
>>
>
> Puppet dashboard exists, but I think it's enterprise only now? There's also
> a fork of it, but it basically requires puppetdb, which requires puppet 3
> (which I'd hope you're using, since you're starting from scratch).
>

No, the thing I downloaded is called puppet dashboard, is open source
(i cloned it from github), requires only mysql and sort of works.
However, minimal installation of this board uses 140mb of ram, typical
installation about 500mb (with 4 + 1 worker processes).

>>
>> Imagine a nagios-like website where you would have a list of all nodes
>> maintained by puppet, with detail configuration information, which
>> manifests are used on each node and if they were successfully applied,
>> when etc. Is there anything like that? Can we set it up on labs? Can
>> someone tell me how to do that or link me to some resources? Thanks
>>
>
> This has been done in labs for a while now. Check out the Manage instances
> view.
>

Yes I know about this of course, but that only allows you to change
the configuration, it doesn't parse the puppet reports, so you have
basically no feedback from node. You can set a class there, but you
can never check whether and how it was applied on node.



More information about the Labs-l mailing list