cobbler: Fix RPM snippet conditional

We need to check for RHEL or CentOS

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

Merge pull request #430 from ceph/wip-ddzero

testnode: Blow away lingering OSD data and partition data

Merge pull request #431 from ceph/wip-37614-2

testnode: Only bounce interface on Xenial

testnode: Only bounce interface on Xenial

resolvconf isn't installed on Bionic by default so we don't need to
bounce the interface to work around the bug.

Fixes: https://tracker.ceph.com/issues/37614
Signed-off-by: David Galloway <dgallowa@redhat.com>

testnode: Blow away lingering OSD and partition data

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

Merge pull request #427 from ceph/wip-int

Fix distro version comparison conditionals

Merge pull request #428 from ceph/wip-callypsodrive

cobbler: Support installing to /dev/sdi for callypso machines

cobbler: Support installing to /dev/sdi for callypso machines

This will get used when the Octo FOG Images job in Jenkins runs

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

Fix distro version comparison conditionals

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

Merge pull request #426 from ceph/wip-rhel74-jwt

cobbler: Only install python-jwt on RHEL7.5 or later

cobbler: Only install python-jwt on RHEL7.5 or later

Fixes: https://tracker.ceph.com/issues/37395
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #425 from ceph/wip-enforcing

cobbler: Set SELinux = enforcing by default

Merge pull request #424 from ceph/wip-fix-prep

prep-fog-capture: Fix failed_when

prep-fog-capture: Fix failed_when

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

Merge pull request #422 from ceph/wip-rhel8

Add RHEL8!

cobbler: Set SELinux = enforcing by default

Requested by downstream QE.  This shouldn't be an issue since only
downstream QE needs SELinux in Enforcing mode.  They don't use
ceph-cm-ansible which sets it to Permissive in the testnodes role.

Fixes: http://tracker.ceph.com/issues/37338
Signed-off-by: David Galloway <dgallowa@redhat.com>

Run ansible-managed playbook as cm user

The ubuntu (teuthology_user) doesn't get created until the testnode
playbook gets run

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

Set ansible_python_interpreter to python3 if necessary

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

testnode: Support RHEL8

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

common: Support RHEL8

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

Merge pull request #423 from ceph/wip-python3

prep-fog-capture: Support RHEL8/python3

prep-fog-capture: Support RHEL8/python3

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

cobbler: Configure snippets & triggers for RHEL8

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

Merge pull request #421 from ceph/wip-f2b-multiline

public_facing: Add bantime override and multiline filter support

public_facing: Support overriding fail2ban bantime per service

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

public_facing: Support multiline regex fail2ban filters

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

Merge pull request #420 from ceph/wip-lockfile

tools: Clean up /ceph-qa-ready lockfile

tools: Clean up /ceph-qa-ready lockfile

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

Merge pull request #419 from ceph/wip-jwt

cobbler: Install python-jwt on RPM-based testnodes

cobbler: Install python-jwt on RPM-based testnodes

This package is needed for ceph-mgr but is not available in the
base/default repos.  Therefore, we need to install it during kickstart
and bake it into our RHEL FOG images.

Fixes: https://tracker.ceph.com/issues/36653
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #415 from ceph/wip-suse

Add support for OpenSUSE Leap 15.0

Merge pull request #418 from kshtsk/wip-zypper-for-teuthology

Add zypper support for teuthology role

Merge pull request #417 from kshtsk/wip-zypper-for-paddles

Add zypper support for paddles role

Merge pull request #416 from kshtsk/wip-zypper-for-pulpito

Add zypper support for pulpito role

Add zypper support for teuthology role

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

Add zypper support for paddles role

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

Add zypper support for pulpito role

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

tools: Support OpenSUSE Leap 15.0 in prep-fog-capture playbook

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

testnode: Add support to configure chronyd instead of ntpd

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

testnode: Add support for OpenSUSE Leap 15.0

Some notes:

  - Setting hostname with ansible fails.  It's a known issue.  Bug is in
the yaml.  teuthology sets the hostname after FOG provisioning anyway so
NBD.
  - Leap mounts all of the root subdirs as separate mountpoints so the `zap`
tasks had to be modified to not wipe those out
  - The sshd config is named `sshd_config_opensuse_leap_NA` because the
task that installs the SSHD config uses {{
ansible_distribution_major_version }} which isn't set to 15.0 in Leap
apparently.

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

common: Add default repos on OpenSUSE Leap

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

cobbler: Modify snippets to work with OpenSUSE Leap 15.0

The usermod command is a little different in OpenSUSE.

Also, in Leap, the /etc/rc.local equivalent is /etc/init.d/boot.local.

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

cobbler: Add OpenSUSE Leap 15.0

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

Merge pull request #413 from ceph/wip-35989

testnode: install gcc-c++ on RPM-based systems

testnode: install gcc-c++ on RPM-based systems

Fixes: https://tracker.ceph.com/issues/35989
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #412 from ceph/wip-curl

cobbler: Install curl during Ubuntu kickstart

cobbler: Install curl during Ubuntu kickstart

I replaced wget with curl in rc.local because curl is installed by
default in RPM-based distros while wget isn't even available for RHEL8
(yet?).  It apparently is not installed by default in Ubuntu though so
we'll install it during kickstart.

Fixes: http://tracker.ceph.com/issues/35952
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #407 from ceph/wip-f2b-ufw

public_facing: Support blocking multiple ports in ufw fail2ban action

Merge pull request #411 from ceph/wip-prep-fog

prep-fog-capture: Modify to work with RHEL8

prep-fog-capture: Modify to work with RHEL8

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

Merge pull request #408 from ceph/wip-rclocal2

cobbler: break at the appropriate location in rc.local

Merge pull request #410 from ceph/wip-rhsm-disable

cobbler: Disable all repos before subscribing during kickstart

cobbler: Disable all repos before subscribing during kickstart

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

Merge pull request #409 from ceph/wip-rhel8

Add RHEL8

Add RHEL8

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

cobbler: break at the appropriate location in rc.local

Before this change, if the first NIC on a machine had a link but wasn't
the uplinked NIC, the break would still get called and no NICs would be
left up.

Now we'll break only after we can ping the Cobbler host.

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

public_facing: Support blocking multiple ports in ufw fail2ban action

UFW requires the protocol be specified when multiple ports are given for
a deny rule.

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

Merge pull request #406 from ceph/wip-mero-disk

cobbler: Add kickstarts that will install to /dev/sdm

cobbler: Add kickstarts that will install to /dev/sdm

The root drive on mero nodes is /dev/sdm.  Since Cobbler is only used to
provision testnodes prior to a FOG image capture, I didn't think it'd be
worth the time/effort to automate this further.

The kickstart will have to be manually selected if the Cobbler Web UI is
used.  I'll automate setting the kickstart when the octo-fog-images
Jenkins job runs.

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

Merge pull request #405 from ceph/wip-rhel76

Add RHEL 7.6

libev-devel isn't available in RHEL7.6

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

Add RHEL 7.6

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

Merge pull request #404 from ceph/wip-ovh-rhel

common/testnode: Support skipping entitlements & packages

Merge pull request #403 from ceph/wip-centos75

Add CentOS 7.5

common/testnode: Support skipping entitlements & packages

`skip_entitlements` and `skip_packaging` will be set to true in the
Ansible inventory for OVH nodes

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

Add CentOS 7.5

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

Merge pull request #402 from ceph/wip-dhcp-ns

dhcp-server: Support defining nameservers by host or group var

dhcp-server: Support defining nameservers by host or group var

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

Merge pull request #401 from tchaikov/libev

testnodes: s/libevent-dev/libev-dev/

Reviewed-by: Sage Weil <sage@redhat.com>

testnodes: s/libevent-dev/libev-dev/

see https://github.com/ceph/swift/pull/7 and
    https://github.com/ceph/s3-tests/pull/104

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #399 from ceph/wip-retry-rhsm

common: Retry rhsm command due to CDN or network flakiness

common: Retry rhsm command due to CDN or network flakiness

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

Merge pull request #398 from ceph/wip-force

common: Force register with RHSM

common: Force register with RHSM

Fixes: https://tracker.ceph.com/issues/23954
See also: https://bugzilla.redhat.com/show_bug.cgi?id=1510658

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

Merge pull request #397 from ceph/wip-bionic

Add support for Ubuntu 18.04

testnode: Redo LVM removal

For whatever reason, `dmsetup remove_all` fails on Bionic.  As long as
there are leftover lvs or vgs, a pv must be linked to them.  We can just
force remove the physical volume which wipes out the rest of the LVM
data.

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

testnode: Create /etc/default/rcS if missing

File does not exist in Bionic

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

testnode: Create sshd config and set package vars for Bionic

Packages missing in Bionic that were in Xenial:
  - libapache2-mod-fastcgi
  - libboost-thread1.58.0
  - libcrypto++9v5

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

cobbler: Set packages to install for Bionic

  - udev-discover is no more
  - net-tools provides ifconfig which is used in rc.local
  - ifupdown provides `ifdown` and `ifup` which are in rc.local
  - python is required for ansible
  - ntp isn't installed by default anymore apparently

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

Cobbler: Add Bionic Beaver

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

Merge pull request #396 from ceph/wip-23845

testnode: Exclude dm devices from list of physical volumes

testnode: Exclude dm devices from list of physical volumes

This was actually happening because when the playbook first runs, the
setup module is run and sees the device mapper devices.  We zap them
later in the playbook but ansible doesn't know that.  We could just
re-run the setup module but this method will instead guarantee we don't
use dm-* devices.

Fixes: https://tracker.ceph.com/issues/23845
Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #394 from ceph/wip-hwclock

Always set hardware clock

Merge pull request #395 from ceph/wip-fog-time

tools: Force time synchronization in FOG images

tools: Force time synchronization in FOG images

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

common: Always set the hwclock on all systems

Even if the OS time is correct, the hardware clock could be off.
There's no harm in making sure it is accurate with every job.

https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/system_administrators_guide/sect-Configuring_the_Date_and_Time-hwclock

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

Merge pull request #391 from ceph/wip-crontabs

teuthology: Script to keep teuthology user's crontab up to date

Merge pull request #386 from ceph/wip-lrc-role

long_running_cluster role

Merge pull request #383 from ceph/wip-status

public_facing: Updates to OpenVPN check for status portal

Merge pull request #379 from ceph/wip-fix-subscribe

cobbler: Fix typo in subscription manager snippet command

Merge pull request #390 from ceph/wip-console-fix

cobbler: Just output to ttyS1

Merge pull request #393 from ceph/wip-smart-ssd

common: Fixes to smart.sh to support multiple attributes

Merge pull request #392 from ceph/wip-dhcp-server

Add dhcp-server role

common: Comment bad nvme error-log check

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

common: Fixes to smart.sh to support multiple attributes

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

dhcp-server: Update set-next-server script to work with role

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

Add dhcp-server role

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

teuthology: Script to keep teuthology user's crontab up to date

Fixes: https://tracker.ceph.com/issues/23441
Signed-off-by: David Galloway <dgallowa@redhat.com>

long_running_cluster: Custom logrotate rules

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

long_running_cluster: Add check for LRC data pool capacity

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