Bug #5255
closed
Calling version destroy takes 20+ minutes
Added by David Davis about 10 years ago.
Updated almost 6 years ago.
Description
It looks like it's deleting repo packages during the plan phase. Also the deletion is really slow because it's using old code instead of the new code (which does batch updates).
- Priority changed from Normal to High
Total time was 23 minutes. I'm going to remove the hook in the ElasticSearch repo module and remove the orphan code from clear_packages_index.
I created a separate dynflow action for updating the content in ElasticSearch when a repo gets deleted. Apparently the code was relying on the before_destroy hook.
To test this:
Create a content view with RHEL 6 RPMs. Publish it. Remove the version via the CLI. Notice that the CLI times out. It should only take a matter of seconds to create a dynflow task. However, it's actually deleting the ES content in the plan phase.
- Blocks Support #5179: Get a baseline for content view performance added
- Subject changed from Calling version destroy takes 10+ minutes to Calling version destroy takes 25+ minutes
- Subject changed from Calling version destroy takes 25+ minutes to Calling version destroy takes 20+ minutes
- Status changed from New to Closed
- % Done changed from 0 to 100
Applied in changeset katello|commit:1cf128184cf205e6d534ad213f9f8c5584b60bda.
- Triaged changed from No to Yes
- translation missing: en.field_release set to 13
Also available in: Atom
PDF
Updated by 
Updated by