Capsule migration option, take 2

[wzzrd]

So after a long pause, here's attempt two to bring in a capsule migration option into bootstrap.py.

Because the change is fairly large, here's the breakdown, to start with the nitty-gritty stuff:

• changes were required to basic functions, like put_json and get_bootstrap_rpm. The put_json function needed to accept a payload to update a host record, and the get_bootstrap_rpm needed to have options to override both whether or not to remove existing katello-ca-consumer RPMs, and whether or not to unregister when forcing the installation of new katello-ca-consumer RPMs. Unregistering, needless to say, from a migration PoV, would be bad.
• I also have added an update_host_capsule_mapping function to DRY when updating the proxy_ids for a system.
• Some obvious changes have been made to the parts of the script that provide help and figure out what to do with all the options.
• The part that does the actual migration is fairly long, that can't be helped much, I'm afraid. What happens is as follows:
  • get the new katello-ca-consumer from the target capsule and figure out the API_PORT, capsule_id, host_id and the features the capsule offers
  • optionally (depending on whether the --hostgroup and --location parameters are passed) switch to new location, hostgroup
  • optionally (depending on puppet and foreman in options.skip) change puppet_proxy_id and puppet_ca_proxy_id
  • optionally (depending on foreman in options.skip and the features the new capsule offers) change openscap_proxy_id
  • switch content_source_id
  • re-enable rhsmcertd
  • optionally, depending on options.skip, reconfigure Puppet

Remarks:

• Right now, I check for foreman and puppet in options.skip when changing puppet_proxy_id, puppet_ca_proxy_id and re-configuring Puppet. I don't think it makes sense to reconfigure Puppet if we skip foreman, because submitted reports will create / update an existing foreman host anyway. So if we skip puppet or foreman, we don't touch anything Puppet related. Agreed?
• We cannot check whether or not a host has an openscap_proxy_id configured through the API. Therefore, what I do is:
  • if we do not skip foreman and the capsule offers OpenSCAP, we set the openscap_proxy_id. After all, if we are staying in the same hostgroup and that hostgroup has OpenSCAP, we want to switch capsules. If we switch hostgroups to a hostgroup that has OpenSCAP, we want to switch capsules. If our hostgroup does not have OpenSCAP, setting the openscap_proxy_id doesn't matter.
  • if there is no openscap feature on the new capsule, we don't touch openscap_proxy_id, but warn the user. Same reasons as above.

Sample usage is provided in README.md. Tested on RHEL6 and RHEL7 w/ Satellite 6.2.12. RHEL5 untested, but I'm not using any functionality in Python that wasn't available on Python 2.4 (I think).

[wzzrd]

Soooo..... the bare excepts that Travis still complains about have been in there for months and have your prints all over them, @evgeni ;)

Not sure what to do about those...

[evgeni]

@wzzrd ah, this is new in flake8 3.5.0. Let's see how we can fix that.

[wzzrd]

@evgeni, @sideangelside, so, any comments on how to move this forward, i.e. get this merged?

[wzzrd]

As for the Capsule / Proxy discussion: I'm not against naming everything by the upstream names, but the problem is that there is a million places where that is not the case right now.

To be honest, looking at get_api_port(), that will only work on Satellite or Capsule servers, because it relies on parsing/etc/rhsm/rhsm.conf. That file might not even exist on pure upstream Foreman / Foreman Proxy, as I understand it. I'm not saying that is right, I'm just saying this issue needs a closer look.

[wzzrd]

• DRYd the Puppet message and noop run by putting them in a function
• Created a function to change hostgroup / location
• Created get_capsule_features function
• Removed superfluous check for foreman_fqdn

Please review code, but wait with potential merge until I do a final functional that I cannot do here.

[wzzrd]

@evgeni ping

[wzzrd]

Test done w/ RHEL7 on Satellite 6.2.12 machine and ditto Capsule. One last change incorporated, works fine otherwise: 6e8d9bee966a5dbf75463da9ea24c82a9e5f0661

[evgeni]

/etc/rhsm/rhsm.conf exists on Katello clients, which is good enough for us :)

[wzzrd]

@evgeni @sideangleside are yall waiting for me to 's/capsule/smart_proxy/g' or is there something else?

[wzzrd]

@evgeni I've manually rebased on current master this morning because there were some conflicts

[evgeni]

@wzzrd puppet4 support and sed-fixes are at https://github.com/evgeni/katello-client-bootstrap/tree/capsule-migration-option-new, feel free to pull these in

[wzzrd]

Pulled in latest master and changes from your fork, @evgeni

[evgeni]

Thanks @wzzrd! I think if we update the README to say Katello and not Satellite and give the new Capsule a example.com domain as with the rest of the examples, this is good to go.

@sideangleside you also happy with that?

[wzzrd]

Done.

[sideangleside]

I finally got time to review this. No issues to report from me. Other than squashing the commits, we can merge this PR. Thanks for the pretty big feature @wzzrd

[wzzrd]

Squashed and pushed @sideangleside @evgeni