(POOLER-123) Implement a max TTL

[sbeaulie]

Before this change, we could checkout a vm and set the lifetime to a
very high number which would esssentially keep the vm running forever.
Now implementing a config setting max_lifetime_upper_limit which enforces
a maximum lifetime in hours both for initial checkout and extending a
running vm

[highb]

Hmmm... seems that when I spin this up with max_lifetime_upper_limit: 9000 set in the config I can no longer change the lifetime at all? Here's how it works normally against prod:

➜  vmpooler git:(POOLER-123) floaty get centos-7-64 --service vm
- diffuse-pizza.delivery.puppetlabs.net (centos-7-64)
➜  vmpooler git:(POOLER-123) floaty modify diffuse-pizza.delivery.puppetlabs.net --service vm --lifetime 10

Successfully modified VM diffuse-pizza.delivery.puppetlabs.net.
Use `floaty list --active` to see the results.
➜  vmpooler git:(POOLER-123) floaty list --active --service vm

Your VMs on vmpooler.delivery.puppetlabs.net:
- diffuse-pizza.delivery.puppetlabs.net (centos-7-x86_64, 0.01/10 hours)

And here's with the changes running locally:

➜  vmpooler git:(POOLER-123) floaty list --active --service vmlocal
Your VMs on localhost:
- amateur-voyage.delivery.puppetlabs.net (centos-7-x86_64, 0.08/2 hours)
➜  vmpooler git:(POOLER-123) floaty modify amateur-voyage.delivery.puppetlabs.net --lifetime 8999 --trace

Successfully modified VM amateur-voyage.delivery.puppetlabs.net.
Use `floaty list --active` to see the results.
➜  vmpooler git:(POOLER-123) floaty list --active --service vmlocal
Your VMs on localhost:
- amateur-voyage.delivery.puppetlabs.net (centos-7-x86_64, 0.09/2 hours)

[sbeaulie]

I should check what api command floaty uses? Did you put max_lifetime_upper_limit as a child of the config: key, and not top scope?

[highb]

Like this? :config: site_name: '<b>vmpooler-provisioner-highb</b>.delivery.puppetlabs.net' domain: 'delivery.puppetlabs.net' logfile: '/var/log/vmpooler.log' clone_target: 'acceptance4' task_limit: 10 vm_lifetime: 1 vm_lifetime_auth: 2 vm_checktime: 1 max_tries: 3 retry_factor: 10 create_folders: true backend_weight: 'acceptance1': 5 'acceptance2': 6 'acceptance3': 6 purge_unconfigured_folders: true experimental_features: true migration_limit: 10 max_lifetime_upper_limit: 9000

[highb]

As for floaty, I think it just PUTS to the vm endpoint https://github.com/briancain/vmfloaty/blob/master/lib/vmfloaty/pooler.rb#L54

  def self.modify(verbose, url, hostname, token, modify_hash)
    raise TokenError, 'Token provided was nil. Request cannot be made to modify vm' if token.nil?

    modify_hash.keys.each do |key|
      raise ModifyError, "Configured service type does not support modification of #{key}." unless %i[tags lifetime disk].include? key
    end
...
  response = conn.put do |req|
      req.url "vm/#{hostname}"
      req.body = modify_hash.to_json
    end

[highb]

Ah, I found the issue. This key is a string, not a symbol. When I switched that out, it works.

[sbeaulie]

good catch. That was an issue with testing this as well, since we set the parameters and I wasn't sure if it was a symbol or a string. Thanks!

[highb]

I've switched out the symbol for a string in the config hash reference and added a little better error messaging so a user knows what actually went wrong. Also, vmfloaty didn't correctly handle when the PUT came back with ok: false so I've put up a PR to fix that up a bit. puppetlabs/vmfloaty#55

[highb]

Oh and this little line was helpful for me to ensure that redis wasn't persisting between testing runs, as it was getting a strange no space left on device error. My computer has plenty of room, so I'm not sure where that is coming from all of a sudden.

[sbeaulie]

check the docker settings for how much space is allowed for all containers? I had to bump mine up a few times.

[highb]

Ah, I didn't even know that was a setting. Thanks!

[sbeaulie]

Thanks for the commits @highb

[kevpl]

lgtm, I'm especially glad to see we'll be including the message with an error return.

this will still have to be released, I'm not planning on doing this on merge.