testnode: Fuse for centos9

Fixes: https://tracker.ceph.com/issues/73630
Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: net-tools for centos9

Fixes: https://tracker.ceph.com/issues/74589
Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: Additional missing Rocky packages

Fixes: https://tracker.ceph.com/issues/73630
Fixes: https://tracker.ceph.com/issues/74572
Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #815 from ceph/wget-r10

testnode: Install wget on Rocky 10

testnode: Install wget on Rocky 10

Same as in commit ab478050c523 ("testnode: Install wget on CentOS 9").

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Merge pull request #814 from ceph/define-rocky-mirrors

Define rocky repo mirrors

testnode: Specify a few mirrors for Rocky 10

As was the case with EPEL mirrors, hitting the mirrorlist.rockylinux.org endpoint frequently fails and causes jobs to die.  We'll just specify a few mirrors manually.

DNF's behavior is to try them in order when multiple baseurls are defined.

Fixes: https://tracker.ceph.com/issues/74620
Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: Support for yum_repos_to_remove

A follow up commit will put specific repo files we want back in place.

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #813 from ceph/retry-mirrorlists

testnode: Try to work around flaky rocky mirrorlists

testnode: Try to work around flaky rocky mirrorlists

```
[root@trial052 ~]# cat /etc/yum.repos.d/rocky.repo

[baseos]
name=Rocky Linux $releasever - BaseOS
mirrorlist=https://mirrors.rockylinux.org/mirrorlist?arch=$basearch&repo=BaseOS-$releasever$rltype
gpgcheck=1
enabled=1
countme=1
metadata_expire=6h
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-Rocky-10
```

Fixes: https://tracker.ceph.com/issues/74563
Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #811 from ceph/wget-c9

testnode: Install wget on CentOS 9

testnode: Install wget on CentOS 9

We were doing this in kickstart via cobbler.

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #809 from ceph/fog-fixes

Fog fixes

Merge pull request #810 from ceph/https

testnode: More s/http/https

testnode: More s/http/https

Signed-off-by: David Galloway <david.galloway@ibm.com>

prep-fog-capture: systemd unit to generate host ssh keys

Ubuntu does not automatically regenerate SSH host keys after image-based
deployments. When keys are removed prior to FOG capture, sshd fails to
start on redeploy.

Add a one-shot systemd service to regenerate host keys on first boot and
disable itself after running.

Signed-off-by: David Galloway <david.galloway@ibm.com>

prep-fog-capture: install ntp if missing

Signed-off-by: David Galloway <david.galloway@ibm.com>

fog-server: Clone as fog user

Avoids 'dubious ownership' git error

Signed-off-by: David Galloway <david.galloway@ibm.com>

fog-server: primary branch is now 'stable'

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #808 from dmick/teuthology-conserver-client

teuthology: add conserver-client as package dependency

teuthology: add conserver-client as package dependency

left aside leap because I believe that code is dead and I don't
have systems to verify the package name

Signed-off-by: Dan Mick <dan.mick@redhat.com>

Merge pull request #806 from deepssin/update_lab_extras

testnode: Switch lab-extras repository URLs to HTTPS

testnode: Switch lab-extras repository URLs to HTTPS

Signed-off-by: deepssin <deepssin@redhat.com>

Merge pull request #805 from ceph/zap-again

Refactor Zap again

testnode: Refactor zap again...

We should be systematically tearing down any LVM we find.  LVs, VGs, then PVs, then device mapper.

Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: Support NVMe partitions in pvs output

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #804 from ceph/xattrs

testnode: Refactor xattrs tasks

testnode: Refactor xattrs tasks

This was failing on lvm-backed root filesystems and isn't required for xfs anyway.

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #798 from ceph/cm-oops

ansible-managed: Don't create user if he exists

Merge pull request #803 from ceph/lvm-by-disksize

testnode: Updates to support trial nodes in MaaS

testnode: Refactor lvm tag and support new dict

Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: Refactor zap to support existing LVM and NVMe root drives

Feeding "nvme0n1" to `root_disk: "{{ item.device|regex_replace('[0-9]+', '') }}" came out "nvmen"

MaaS also provisions root disks using lvm by default so we need some logic to wipe out device mapper devices from OSD disks but not the root disk.

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #802 from ceph/trial-noble

Adjustments to run ceph-cm-ansible on Noble on new "trial" testnodes

testnode: Strip partitions from root_disk

Signed-off-by: David Galloway <david.galloway@ibm.com>

testnode: Don't zap snap partitions

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #800 from ceph/nameserver-multiple-ipvars

nameserver: Support multiple ipvars

nameserver: Support multiple ipvars

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #799 from ceph/wip-xfs-v4-revert

roles/testnode: don't insist on creating XFS V4 filesystem

roles/testnode: don't insist on creating XFS V4 filesystem

The V4 format has been deprecated since 2020 and as of 6.18 kernel no
longer compiled in by default.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Merge pull request #786 from kshtsk/wip-rocky-9

roles/testnode: add rocky 9

roles/testnode: add sshd config template for rocky 9

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>

roles/testnode: add rocky 9 vars

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>

testnode/vars: add lvm2 to centos 9

The nvme-loop tests rely on lvm.

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>

roles/testnodes: add centos 9 stream specific vars

Looks like noone added the subj. yet.

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@clyso.com>

ansible-managed: Don't create user if he exists

When manually installing hosts, we use the cm user sometimes.  Don't clobber it or try to create when it exists already.

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #797 from dmick/fix-make-dnsmasq-conf

tools/make-dnsmasq-conf: fix up some outstanding issues

tools/make-dnsmasq-conf: fix up some outstanding issues

1) hack domain_tag calculation for pok, suggest a better approach
2) fully-qualify the ip manipulation module names
3) host-record instead of host-address
4) create host-record in canonical order to avoid anyone worrying

Signed-off-by: Dan Mick <dmick@ibm.com>

Merge pull request #796 from ceph/toko-disk

tools: Add bath and toko to jenkins-builder-disk

tools: Add bath and toko to jenkins-builder-disk

Signed-off-by: David Galloway <david.galloway@ibm.com>

Merge pull request #792 from dmick/fix-set-bmc-static

tools/set-bmc-static.yml: update for Tucson lab/Supermicro

tools/set-bmc-static.yml: update for Tucson lab/Supermicro

1) set to 'all' hosts, enabling --limit on playbook invocation
2) set power_uid to 3; newer Supermicros use 2 for their factory user
3) enable dhcp mode by default
4) introduce "only_lan_access" to avoid trying to ssh or even test it
5) add tasks to test initial_* and power_* creds and display the results
6) halt the playbook if neither set of creds works
7) add a "test_user" tag to allow just testing the creds and stopping
8) bugfix: use 'initial_*' for first 'user set password' for power_user
9) also add 'user enable' because it seems like Supermicro needs it

Signed-off-by: Dan Mick <dmick@ibm.com>

Merge pull request #793 from dmick/generate-dnsmasq-snippets

Add a role to generate dnsmasq.conf sections for tucson