testnode: Install redhat-lsb-core

This usually gets installed via Cobbler during FOG image capturing.  That doesn't do us any good when FOG isn't being used.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #636 from ceph/wip-fromhosts

public_facing: Support defining a src port for UFW rules

public_facing: Support defining a src port for UFW rules

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #632 from ceph/wip-git

container-host: Ensure git is installed

container-host: Ensure git is installed

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #631 from ceph/wip-notin

cobbler: Fix ansible conditional

Merge pull request #630 from ceph/wip-centos84

Add CentOS 8.4

cobbler: Fix ansible conditional

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #629 from ceph/wip-rhel84

RHEL 8.4

Add CentOS 8.4

Signed-off-by: David Galloway <dgallowa@redhat.com>

RHEL 8.4

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #628 from ceph/wip-anybody

We can use this for any system

We can use this for any system

But the ipmi group makes more sense

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #627 from ceph/wip-comma

testnode: Always ifup the secondary NIC

testnode: Always ifup the secondary NIC

If the NIC was already down, `ifdown` causes the playbook to fail.  We want to always (at least attempt to) take the NIC down, then bring it back up.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #626 from ceph/wip-move-sentinel

testnode: Write sentinel file after container-host role

testnode: Write sentinel file after container-host role

Now that we're running the entire cephlab.yml playbook during Cobbler kickstart, the sentinel file is getting touched while the container-host playbook is still running.

This change will still write the /ceph-qa-ready sentinel file during the testnodes role if the testnodes role got called directly

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #625 from ceph/wip-cephlabnottestnodes

cobbler: Run all available playbooks during postinstall

cobbler: Run all available playbooks during postinstall

Not just testnodes playbook

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #624 from ceph/wip-klbahsr

Allow defining an alternate container_packages var

container-host: Try to import an alternate {{ container_packages }}

Signed-off-by: David Galloway <dgallowa@redhat.com>

testnode: Do not install podman-docker here

We will install it in the container-host role if we need it.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #609 from ceph/wip-ignore-snap

common: Ignore /snap mountpoints when checking disk usage

Merge pull request #622 from ceph/wip-c9stream

Add CentOS 9 Stream

Add CentOS 9 Stream

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #621 from ceph/wip-cstream

cobbler: Support CentOS Stream

cobbler: Support CentOS Stream

"But David, why aren't we just using the CentOS 8 Stream ISO?"

Thank you for asking.
1) The vmlinuz and initrd that are baked into the CentOS 8 Stream ISO as of this writing are not compatible.  Filed https://bugs.centos.org/view.php?id=18188
2) Even if I take the vmlinuz and initrd from CentOS 8.3, after firstboot, the system still bizarrely boots as CentOS 8.3.  That's right, when using the CentOS 8 Stream ISO and we kickstart using Cobbler, you end up with CentOS 8.3.  I'm sure it has something to do with our kickstarts and snippets but nothing obvious stood out to me.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #620 from ceph/wip-gw-routes

gateway: Support writing addl routes

gateway: Support writing addl routes

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #618 from ceph/wip-releasever-sooner

common: Set releasever on testnodes sooner

common: Set releasever on testnodes sooner

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #611 from ceph/wip-centos83

Add CentOS 8.3

Merge pull request #617 from ceph/wip-releasevar-yum

Adding the releasever variable to rhel7.9 testnodes

Adding the releasever varible to rhel7.9 testnodes

Fixes: https://tracker.ceph.com/issues/49771
Signed-off-by: Adam Kraitman <akraitma@redhat.com>

Merge pull request #616 from ceph/wip-centos79

Add CentOS 7.9

Add CentOS 7.9

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #615 from ceph/wip-cobbler-imported

cobbler_profile: s/import/imported

cobbler_profile: s/import/imported

I guess something changed in ansible so that it doesn't like 'import' as the variable name.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #613 from ceph/wip-norecommends

testnode: Install collectl without recommended packages

testnode: Install collectl without recommended packages

collectl has a bunch of Gnome Desktop Environment packages that are recommended.  So on Ubuntu 20.04 at least, a desktop environment was getting installed and would eventually go to sleep.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #612 from ceph/wip-fix-centos8-qemu

testnode: libgcrypt needs to be updated to match QEMU build

testnode: libgcrypt needs to be updated to match QEMU build

See https://bugzilla.redhat.com/show_bug.cgi?id=1828681

Signed-off-by: Jason Dillaman <dillaman@redhat.com>

Add CentOS 8.3

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #610 from ceph/wip-lz4

paddles: Install lz4 for backups

paddles: Install lz4 for backups

Signed-off-by: David Galloway <dgallowa@redhat.com>

common: Ignore /snap mountpoints when checking disk usage

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #608 from kshtsk/wip-nsupdate_web-opensuse

nsupdate web opensuse

nsupdate_web: use .conf extention for opensuse nginx

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@gmail.com>
(cherry picked from commit cbb0a82a0380038d33d1b59eb65eed9622ec61d6)

nsupdate_web: copy keys from secrets dir if present

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@gmail.com>
(cherry picked from commit bac5f08eca5fa018adb16be9db4457cb42c73039)

Merge pull request #607 from ceph/wip-rhel8-package-list-change

Adding redhat-lsb, firewalld and wget to rhel8 testnodes

Adding redhat-lsb, firewalld and wget to rhel8 testnodes

Fixes: https://tracker.ceph.com/issues/48767
Signed-off-by: Adam Kraitman <akraitma@redhat.com>

Merge pull request #606 from ceph/wip-focal-nosleep

common: Disable sleeping on Ubuntu 20.04 and later

common: Disable sleeping on Ubuntu 20.04 and later

Fixes: https://tracker.ceph.com/issues/49152
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #605 from ceph/wip-keys-branch

users: Support defining different branch

users: Support defining different branch

This will allow one to run the users role prior to a PR being merged.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #604 from ceph/wip-rook-master-update

Creating a new tag for updating the master os - master-update

Creating a new tag for updating the master os - master-update

And also doing changes to the rook-recovery role

Signed-off-by: Adam Kraitman <akraitma@redhat.com>

Merge pull request #603 from kshtsk/pulpito-py3

teuthology: py3 for pulpito and paddles

teuthology: ensure scheduler user bin exists

Before copying crontab script we need to ensure the bin directory
exists for scheduler user.

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

teuthology: add opensuse 15.2 specific deps

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

teuthology: use python3 dependencies

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

teuthology: py3 for pulpito and paddles

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

Merge pull request #602 from ceph/wip-bothplaces

container-host: Install mirror cert in both podman and docker paths

container-host: Install mirror cert in both podman and docker paths

Fixes: https://tracker.ceph.com/issues/48715
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #601 from ceph/wip-make

testnode: Install make on el8

testnode: Install make on el8

This package isn't included in the cloud image that the Octo PSI nodes use.

Fixes: https://tracker.ceph.com/issues/48751
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #600 from ceph/wip-container-js

container-host: let me skip registries-conf-ctl tasks

container-host: Create tag for registries-conf-ctl

This will let us skip those tasks when we don't want to run them (like on a Jenkins builder)

Signed-off-by: David Galloway <dgallowa@redhat.com>

container-host: Configure Jenkins builders too

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #599 from ceph/wip-ignore-errorsrue

container-host: Ignore service restart error on new kernel

container-host: Ignore service restart error on new kernel

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #598 from ceph/wip-tab

container-host: Fix service restart syntax

container-host: Fix service restart syntax

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #597 from guits/guits-restart-docker

container-host: restart docker service

container-host: restart docker service

Once the mirror has been added in `/etc/docker/daemon.json`
we need to restart the daemon.

This is not very elegant but it's a workaround for now.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #596 from guits/guits-quick_enhancement

container-host: quick improvement in container_mirror code

container-host: quick improvement in container_mirror code

Let's get rid of these consecutives `set_fact`.
We can inject `PATH` variable on the task calling `registries-conf-ctl`
instead.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #595 from ceph/wip-containers

Move container tasks to separate role

Move container tasks to separate role

I would've liked to keep all this in common but there's a chicken and egg situation.

docker and/or podman get installed during the testnode role.  The testnode role can only be run after the common role.  The testnode role is also where some repos are added.

So we need to install docker/podman and configure it after the testnodes role runs.  Since we also want to be able to configure docker/podman on other systems, I couldn't put these tasks in the testnode role.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #594 from ceph/wip-reload

testnode: Restart cachefilesd in new kernel

testnode: Restart cachefilesd in new kernel

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #593 from ceph/wip-fscache

testnode: Install and configure cachefilesd

Merge pull request #592 from ceph/wip-fs

testnode: Basic filesystem and mountpoint support

testnode: Install and configure cachefilesd

Fixes: https://tracker.ceph.com/issues/6373
Signed-off-by: David Galloway <dgallowa@redhat.com>

testnode: Basic filesystem and mountpoint support

Fixes: https://tracker.ceph.com/issues/6373
This will really only be useful is `drives_to_partition` or `logical_volumes` gets overridden in ansible.cephlab overrides in teuthology yaml.  e.g.,

```
overrides:
  ansible.cephlab:
    vars:
      drives_to_partition:
        nvme0n1:
          device: "/dev/nvme0n1"
          unit: "GB"
          sizes:
            - "0 80"
            - "80 160"
            - "160 240"
            - "240 320"
            - "320 340"
            - "340 400"
          scratch_devs:
            - p1
            - p2
            - p3
            - p4
      filesystems:
        nvme0n1p6:
          device: "/dev/nvme0n1p6"
          fstype: xfs
          mountpoint: "/var/cache/fscache"
```

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #590 from ceph/wip-smart-bruuni

smart.sh enhancements for totally failed drives

common: smart.sh - print serial numbers of bad drives

Signed-off-by: David Galloway <dgallowa@redhat.com>

common: smart.sh - sort failure messages

Signed-off-by: David Galloway <dgallowa@redhat.com>

common: Support smart.sh report failed drive

The SSDs in the bruuni, specifically, have mostly catastrophically failed.  There are no details about sectors in the SMART output.  Just obvious failure.  `smart.sh` wasn't capable of detecting that.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #589 from ceph/wip-trust-dockermirror

common: Support for container mirror CA certs

common: Support for container mirror CA certs

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #588 from ceph/wip-chrony-step

Modify playbook to function with chrony

Modify playbook to function with chrony

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #587 from ceph/wip-rhel83

Add RHEL8.3

Add RHEL8.3

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #586 from ceph/wip-rhel79-centos82-cobbler

Adding RHEL7.9 and Centos8.2 to Sepia & Octo

Adding RHEL7.9 and Centos8.2 to Sepia & Octo

Signed-off-by: Adam Kraitman <akraitma@redhat.com>

Merge pull request #585 from ceph/wip-lvm2

testnode: Install lvm2 on RHEL8 machines

testnode: Install lvm2 on RHEL8 machines

The PSI nodes downstream don't have lvm2 pre-installed like the baremetal ISO image.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #584 from ceph/wip-centos7-aarch64

Adding Centos-7.8-aarch64 to cobbler