Issue #2559
closed
Empty pkglist for the erratum may be generated which may cause yum not to find applicable packages
Status:
CLOSED - CURRENTRELEASE
Description
If none of the packages listed in the original erratum are in the repo, such erratum will be published with empty pkglist.
That does not make any sense - no packages to update anyway.
The case when the same erratum is in multiple repos is more complicated and is worse for the yum to handle.
If yum finds the erratum with the same id and collection name for pkglist in N repos, it chooses the first pkglist and that's it.
If the first one will turn out to be an empty one, yum will show no packages for update in the scope of the particular erratum.
Our users can easily end up with one or more empty pkglists for the erratum because of the way Pulp stores and then generates updateinfo.xml.
- Description updated (diff)
- Priority changed from Normal to High
- Sprint/Milestone set to 32
- Severity changed from 2. Medium to 3. High
- Triaged changed from No to Yes
- Sprint/Milestone changed from 32 to 33
- Status changed from NEW to ASSIGNED
- Assignee set to daviddavis
- Status changed from ASSIGNED to NEW
- Assignee deleted (
daviddavis)
- Status changed from NEW to ASSIGNED
- Assignee set to ttereshc
- Status changed from ASSIGNED to POST
- Status changed from POST to MODIFIED
Is the following statement correct?
If an {S,D,}RPM is absent from a repository, then errata referencing that RPM shall also be absent.
I'm curious, because we have at least three test cases that just broke in exactly this fashion. One of the broken tests is ErratumPkgListCountTestCase. The test does the following:
1. Create an RPM repository with a feed and distributor, and sync the repository.
2. Remove the "gorilla" package from the repository.
3. Publish the repository.
4. Examine the published repository's updateinfo.xml file.
The source repository has four errata, and one of the errata references the "gorilla" package. In the repository published by Pulp, only three errata are present, with the one referencing the "gorilla" package missing.
Is the following statement correct?
If an {S,D,}RPM is absent from a repository, then errata referencing that RPM shall also be absent.
Partially.
If this RPM was the only RPM in the erratum pkglist, then yes, such erratum should be absent from `updateinfo.xml`.
If there are more RPMs in the erratum pkglist, then such erratum should be published, just without those RPMs absent from repo.
OK. I think I understand what you're saying, but I want to make sure that I get it 100%, because this seems to be a significant change from how Pulp has worked in the past. So bear with me.
Let's say I create an RPM repository, add an erratum to it (but no RPMs), and publish it. Will the errata be available via the published repository? My understanding is "no."
Here's a second scenario. Let's say I create an RPM repository, and add:
- an rpm "foo-package",
- an errata "foo-errata" referencing "foo-package"
- an rpm "bar-package", and
- an errata "bar-errata" referencing "bar-package".
When I publish this repository, both errata will be available, correct? And if I remove "bar-package" from the repository and publish the repository again, then both "bar-package" and "bar-errata" will be omitted, correct?
Your understanding is correct in both cases.
- Platform Release set to 2.12.2
- Status changed from MODIFIED to 5
We had a good team chat about the "Verification Required" flag on Monday, and decided that the release of 2.12.2 should not be blocked on the verification of this issue.
- Status changed from 5 to CLOSED - CURRENTRELEASE
- Sprint/Milestone deleted (
33)
Also available in: Atom
PDF
.
Filter out empty errata pkglists during publish
Stop publishing errata with empty pkglist. If an erratum has multiple pkglists, publish only non-empty ones.
closes #2559 https://pulp.plan.io/issues/2559