Hide Forgot
Description of problem: Blindly updating from RHEL 7.3 to RHEL 7.4 would pull in several 32-bit packages for no reason at all due to rdma.noarch being replaced by rdma-core.{i686,x86_64}. # yum update ... rdma-core i686 13-7.el7 repo 43 k replacing rdma.noarch 7.3_4.7_rc2-6.el7_3 rdma-core x86_64 13-7.el7 repo 43 k replacing rdma.noarch 7.3_4.7_rc2-6.el7_3 This can be avoided, for example, by doing "yum install rdma-core" first or with something like "yum update --exclude='*.i686'. However, the default is to install unnecessary 32-bit packages on 64-bit hardware which in 2017 is unhelpful. Please see if rdma packaging could be adjusted to prevent the above. If that is not possible, please reassign to yum/rpm to see if the package manager could do the right thing by default. Thanks. Version-Release number of selected component (if applicable): RHEL 7.3 / rdma-7.3_4.7_rc2-6.el7_3.noarch RHEL 7.4 / rdma-core.x86_64 0:13-7.el7.x86_64
This was already looked into in the 7.4 devel cycle, there's a bug that was used to generate a knowledgebase entry. It's a known issue, and we have no good way around it from the rdma-core side. I'll reassign it to yum to get input there, I guess.
Hi Michal, what behaviour is expected for update and multilib_policy=best when currently installed package is secondary arch (or noarch) and all variants of updates are available (primary, secondary and noarch)? Is it expected to preserve the current arch (e.g. secondary -> secondary) or update to primary arch (e.g. secondary -> primary)? In yum.conf man page it is described expected behaviour only for installation (not for updating).
(In reply to Eva Mrakova from comment #8) > Is it expected to preserve the current arch (e.g. secondary -> secondary) Assuming that you're talking about regular (non-obsoleting) updates, then yes, this is what will (and should) happen. Please note that this patch is supposed to impact obsoletion scenarios _only_ (more specifically, noarch -> arch obsoletions).
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:3279