[Rearch] Combine worker messages 'sync_config' and sync_active_role' into a single 'sync_config' message. #15597
Conversation
Fryguy
changed the title
|
Marking as WIP because of the TODOs and because of the discussion we had |
app/models/miq_worker/runner.rb
Outdated
| _log.info("#{log_prefix} Synchronizing active roles...") | ||
| opts = _args.extract_options! | ||
| sync_active_roles(opts[:roles]) | ||
| _log.info("#{log_prefix} Synchronizing active roles complete...") |
There was a problem hiding this comment.
As discussed, let's just move this into the sync_config method itself (or perhaps let's just merge the two methods into message_sync_config and get rid of sync_config itself.
There was a problem hiding this comment.
Also, please verify the order of when sync_active_roles should be called with respect to reloading the config
gtanzillo
force-pushed
the
rearch-combine-config-and-roles-messages
branch
from
6d8b07d
to
84696b0
Compare
gtanzillo
changed the title
…ngle 'sync_config' message. Preparing for running workers in pods, we won't be able to pass messages to workers over a Drb connection. Combining the messages into one that does both every time will simplify the implementation when we use a different mechanism.
…to `sync_config` method Removed all callers of `sync_active_roles`
|
This pull request is not mergeable. Please rebase and repush. |
gtanzillo
force-pushed
the
rearch-combine-config-and-roles-messages
branch
from
84696b0
to
c801bdb
Compare
|
@Fryguy This one is ready to go. I did a lot of testing with this and determined that it doesn't really matter what order the roles and config are synchronized. I left it with the roles going first followed by the config because that's the order it was originally done when both things needed synchronization. |
|
Checked commits gtanzillo/manageiq@580b0ab~...c801bdb with ruby 2.2.6, rubocop 0.47.1, and haml-lint 0.20.0 |
Preparing for running workers in pods, we won't be able to pass messages to workers over a Drb connection. Combining
the messages into one that does both every time will simplify the implementation when we use a different mechanism.
In this case, we can simplify the logic in #15471 where memcached is new mechanism
/cc @Fryguy