Fix MiqWorker service_base_name to match unit file prefix #21149
Conversation
agrare
requested review from
Fryguy,
gtanzillo,
jrafanie and
kbrock
as code owners
| describe ".service_base_name" do | ||
| before { MiqWorkerType.seed } | ||
|
|
||
| it "every worker has a matching systemd target and service file" do |
There was a problem hiding this comment.
Really want to make this a providers_common spec but we'll need a good way to determine which workers come from which engine.
This is how the container_common_spec is testing all worker deployment name lengths
There was a problem hiding this comment.
Do we? It can't hurt to check all workers in all engines. If someone changes a particular engine, then the probability they break their own engine is higher than breaking others, especially since we won't merge on red.
agrare
force-pushed
the
fix_systemd_service_names
branch
from
7204ae3
to
8e186ca
Compare
| [service_file, target_file] | ||
| end | ||
|
|
||
| expect(expected_units).to match_array(found_units) |
There was a problem hiding this comment.
Just realized this should have been the other way around...
expect(found_units).to match_array(expected_units)
but it's fine.
There was a problem hiding this comment.
@Fryguy I think this was actually correct originally. I swapped this in the follow-up PR but now that I'm running one switching this around prints missing unit files as "extra" rather than "missing". Nevermind there were issues with the worker seed for subclassed workers, thought it was the missing unit files but was actually missing worker classes
After we renamed the systemd services to have "manageiq-" and "manageiq-providers-" prefixes the service_base_name wasn't updated to match the new names.
Also adds a spec test to ensure all workers have systemd unit files and that there are no extra unit files without workers.
Depends on: