Project

General

Profile

Bug #9899

Subscription Indexing Issue When Deleting Manifest

Added by Justin Sherrill about 8 years ago. Updated over 4 years ago.

Status:
Closed
Priority:
Normal
Category:
Web UI
Target version:
Difficulty:
Triaged:
Yes
Bugzilla link:
Fixed in Releases:
Found in Releases:
Red Hat JIRA:

Description

Cloned from https://bugzilla.redhat.com/show_bug.cgi?id=1205378
Description of problem:

After deleting a manifest, several subscriptions remained in Content > Red Hat Subscriptions.

Version-Release number of selected component (if applicable):

6.1.0 Snap 7 Compose 3

How reproducible:

Seems fairly easy to hit.

Steps to Reproduce:
1. Import the attached manifest. 8 pools are visible in the web UI, 11 are actually in the Candlepin db. (suspect this is unmapped guest pools being hidden)

2. Register a virt guest and consume a subscription. In my case I used one of the temporary unmapped guest subscriptions.

3. Delete the manifest.

4. Navigate to Content > Red Hat Subscriptions

Actual results:

In my case there were 4 of the 11 original pools remaining, however Candlepin has no pools in it's database.

Expected results:

All pools gone.

Additional info:

I believe this data is originating from Katello's data index, which is recalculated when pool events are received.

In this case I can confirm that 11 pool deleted events were sent by candlepin, and I can see that all 11 arrived in gutterball, so presumably they made it to katello as well.

Running 'foreman-rake katello:reindex' corrects the issue.

I also have a similar case reporting that entitlement counts are wrong after entitlements are deleted, which I am very suspicious is related:

https://bugzilla.redhat.com/show_bug.cgi?id=1202896

Feels like there may be some kind of issue in the event processing code? Perhaps when multiple events are received at once, could multiple index attempts be trampling each other?

In testing I think I've also seen the counts not reflecting entitlements on systems, just showing me 0 for all pools, even though the guest was consuming a physical entitlement (which wouldn't be hidden). I checked and CPU was quiet so it doesn't appear indexing is running... Running reindex again fixed this issue.

Attempted to delete the manifest a second time, and this time all 8 pools remain visible in the web UI, almost seems like no events are moving now. I can still confirm they are making it to qpid and being delivered to gutterball.

Associated revisions

Revision 7ddbb554 (diff)
Added by Justin Sherrill about 8 years ago

fixes #9899 - reindex all org subscriptions during manifest destroy

Revision c278246a
Added by Justin Sherrill about 8 years ago

Merge pull request #5142 from jlsherrill/9899

fixes #9899 - reindex all org subscriptions during manifest destroy

History

#1 Updated by The Foreman Bot about 8 years ago

  • Status changed from New to Ready For Testing
  • Target version set to 68
  • Pull request https://github.com/Katello/katello/pull/5142 added
  • Pull request deleted ()

#2 Updated by Eric Helms about 8 years ago

  • Legacy Backlogs Release (now unused) set to 23
  • Triaged changed from No to Yes

#3 Updated by Justin Sherrill about 8 years ago

  • Status changed from Ready For Testing to Closed
  • % Done changed from 0 to 100

Also available in: Atom PDF