Bug 1537322
Summary: | Yum update behavior is inconsistent with yum install behavior | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 7 | Reporter: | nathan |
Component: | yum | Assignee: | Packaging Maintenance Team <packaging-team-maint> |
Status: | CLOSED WONTFIX | QA Contact: | BaseOS QE Security Team <qe-baseos-security> |
Severity: | low | Docs Contact: | |
Priority: | unspecified | ||
Version: | 7.7 | CC: | emrakova, james.antill, mdomonko |
Target Milestone: | rc | ||
Target Release: | --- | ||
Hardware: | x86_64 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2018-11-06 13:56:41 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
nathan
2018-01-23 00:04:47 UTC
The reason "yum install notepadqq" picks notepadqq.x86_64 0:0.46.1-0.el7.centos is not the dist tag but the arch matching the underlaying system. It is controlled by the multilib_policy setting (see yum.conf(5) for details). In your case, you most likely have it set to "best". On the other hand, when updating, yum always picks the latest available package, even if it's a different arch. However, since RHEL-7.5, it's possible to say that specific packages shouldn't change their architecture in updates using the exactarchlist setting (again, see the man page). To make all packages stick to their current arch, you can set exactarchlist=* in yum.conf. However, the default is an empty list. Why multilib_policy is not honored on updates as well, I'm not entirely sure, but that's how things have always been. My guess is that this is to cover for packages changing the architecture they're built for (e.g. noarch -> x86_64). Either way, I'm afraid we cannot change the default behavior at this point without risking breakage for existing (enterprise) users who may rely on it for one reason or another. There's a related BZ with some more details: https://bugzilla.redhat.com/show_bug.cgi?id=1361609#c5 Therefore, I'm closing this as WONTFIX. |