To anyone on the cloud SIG team, specifically dealing with a TripleO
deployment:
As of this week, a stock install of the "Ussuri" and "Victoria"
releases of TripleO undercloud fail due to internal puppet module issues
as follows:
Note that at least Ussuri worked recently (in the last few weeks):
Ussuri:
"puppet-user: Error: no parameter named 'key_size' (file:
/etc/puppet/modules/tripleo/manifests/certmonger/haproxy.pp, line: 102)
on Certmonger_certificate[haproxy-external-cert] (file:
/etc/puppet/modules/tripleo/manifests/certmonger/haproxy.pp, line: 102)
on node centos8-testing.localdomain"
It seams that this may have been fixed in
https://github.com/openstack/puppet-tripleo/commit/963b47338008f16039a815b03f00ad6e387b6ca0
A new puppet-certmonger is needed for the puppet-tripleo in stable/ussuri. I proposed
https://review.rdoproject.org/r/#/c/31942/ to add it to -testing repo, once it's merged will push it to -release. Will update this list once it's available in
mirror.centos.org ussuri repos. It's possible that more such packages are missing.
Victoria:
... "Could not find class ::neutron::plugins::ml2::ovs_driver" ...
I'm not sure if this is related/fixed, but it may be:
https://review.opendev.org/c/openstack/puppet-neutron/+/757803
Steps to reproduce:
Base install of CentOS8.3 Minimal (VM and baremetal behave the same)
install centos-release-openstack-<version>
[Optional?] dnf module disable container-tools:rhel8 and dnf
module enable container-tools:2.0
dnf install python3-tripleoclient ceph-ansible openstack-selinux
dnf distro-sync
add the "stack" user and such, add sudo rules, etc.
[copy undercloud.conf.sample to undercloud.conf]
NOTE: minor customization of IPs in my test environment.
[Optional] generate the container parameters yaml
openstack undercloud install --verbose
Has anybody else experienced this? Any ideas on how to rectify the
issue or prevent this from happening in the future. This seems like a
pretty big breakage since it is effectively a no-go for any new install
at least. (I'd assume the overcloud install will break in the same
manner since it should be using the same repos and containers.)
We have limited coverage in those scenarios(3 packstack scenarios + 4 puppet-openstack scenarios + 1 TripleO standalone job which builds and deploy containers), We can add some more TripleO jobs to catch such issues early.
Another option is you can use DLRN trunk repos[1][2] as defined in [2] which get fixes/features early as they don't rely on tag releases but get builds from latest commits in branches and also have good CI coverage so you will not see the above issues using these.
Please pardon my ignorance if I'm missing something obvious since
I'm a new poster to this thread.
Thanks,
-Nathan
_______________________________________________
CentOS-devel mailing list
CentOS-devel@centos.org
https://lists.centos.org/mailman/listinfo/centos-devel