Hi all,
This bug[1] was failing (simplex in the baremetal case) because the /etc/init.d/compute_config was unable to configure the compute causing a "degraded" state and nova-compute to be disabled.
While running the compute_config script a lot of kernel messages like this appeared:
DMAR: intel_iommu_map: iommu width (39) is not sufficient for the mapped address (7fbdc0000000)
It turns out that this is a known issue in DPDK[2] and a patch is available[3]. I applied this patch into the openvswitch package and then we were able to solve the problem.
My question is, what would be the best way to handle this patch?
- Should we patch our openvswitch package? : we get openvswitch from centos and that version doesn't include this patch. We can apply this patch here and wait for an upgrade in upstream to remove it. - Should be look for a newer openvswitch version? We could look for a newer version but this one won't be a centos standard version.
What do you think?
Thanks in advance
-Erich
[1] https://bugs.launchpad.net/starlingx/+bug/1796420 [2] http://mails.dpdk.org/archives/stable/2018-January/004109.html [3] https://github.com/DPDK/dpdk/commit/54a328f552ff2e0098c3f96f9e32302 675f2bcf4