chef-bcpc

Minor version 7.1.9

#1409 - fix ebtables / nova-network connection issues

Minor version 7.1.8

#1400 - add nova option to disable/enable new services by default
#1401 - add nameservers to storage interface
#1403 - add config_drive by default

Minor version 7.1.7

#1365 - add a preference for the max supported version of erlang
#1390 - preseed fixes
#1391 - ensure correct /etc/resolv.conf settings without requiring reboot
#1393 - update dhcp lease time to one hour; check this #PR for manual restart instructions
#1394 - don't auto-upgrade packages

Minor version 7.1.6

#1363 - reduce unneeded vip jumps
#1364 - reduce mysql lockups on commit

Minor version 7.1.5

#1362 - rework ksm script to capture all states

Minor version 7.1.4

Minor version 7.1.3

#1358 - mitigate performance issues and add ipmi serial console
#1359 - install kexec to remove 5 minute wait per node installation
#1360 - explicitly set rpc response timeouts
#1361 - add sleep to allow time for interface to finish coming up

Minor version 7.1.2

#1352 - add mysql tuning
#1353 - restart cinder-api correctly after monkey-patching
#1354 - add /etc/default/ebtables to not unload modules on restart
#1355 - remove nova policy duplication to remove warnings
#1356 - set sensible defaults for nova limits
#1357 - remove i386 arch from preseed and disable deb source repos

Minor version 7.1.0

#1267 - add screen recipe
#1268 - add kexec recipe
#1270 - I see the metadata is there from a different PR I just merged, so I believe this is now suitable to merge
#1271 - add bcpc-extra cookbook
#1272 - Enforce timeout on pdns_control execution
#1277 - Make example ansible inventory file compliant with ansible >= 2.0
#1278 - remove proposed as the default repo
#1279 - Rework Neutron/Keystone integration
#1283 - First-pass cleanup of current shell provisioning
#1284 - Print build time in hours, minutes and seconds as well as total
#1286 - add postfix recipe
#1288 - Fix proxy tests on linux
#1290 - use latest calico repo and calicoctl
#1291 - fix chef deprecated feature warnings
#1292 - remove ephemerals
#1295 - Declarative hosts config for Vagrant build
#1296 - Revert "Fix proxy tests on linux "
#1299 - remove i386 foreign arch from chef client
#1300 - Modified scrub settings to limit affect on performance during scrubbing
#1301 - Feature/cephtunable
#1302 - Add pre-requisite jq
#1304 - install apache2-utils (required for mitaka)
#1306 - Force ifup in case of missing dhclient during vm provision. Closes #1305.
#1307 - Monkeypatch URI HTTP/S classes to not strip default port during strin…
#1309 - Fix incorrect memcache configuration
#1310 - Add unique internal network names.
#1311 - pin ceph packages to hammer
#1313 - disable warnings on trusty
#1315 - remove ceph radosgw
#1317 - Facilitate domain scoped tokens
#1318 - Use SSH tunneling to access BCPC sites
#1325 - Use right values for admin_project*
#1328 - migration from liberty to mitaka
#1329 - update horizon to use memcached for session storage
#1330 - Deprecate unnecessary/redundant vagrant script
#1335 - better YAML conversion support
#1337 - Correct dpkg command for printing foreign architectures
#1338 - fix: test timeout must exceed OSD heartbeat grace
#1344 - check for ip address rather than link up status
#1345 - added modprobe.d/nf_conntrack.conf
#1346 - added support for additional fixed networks
#1347 - update rally pip install to use system certs
#1349 - Bump chef-client to 12.19.36

Major version 7.0.0

Mitaka release series: The default openstack release is now Mitaka. Additionally, a technical preview of Neutron+Calico networking is included.

Upgrading

Upgrading should be seamless, though a re-provision will be necessary for the rally changes to be incorporated (See https://github.com/bloomberg/chef-bcpc/blob/7.0.0/docs/vagrant_build_guide.md#getting-started). Additionally, it is expected that some Openstack service daemons will restart as per #1231.

Full set of additional features and fixes includes:
#1222 - Mitaka.
#1230 - Spaces between # and section titles. See https://github.github.com/gfm/#atx-headings.
#1231 - Add max_overflow/max_pool_size tuning for OpenStack services
#1232 - Fix parsing of mysql ping item to detect crashes
#1234 - Neutron+Calico technical preview
#1238 - Update sample mirror list to drop hwraid and Kilo UCA, add Mitaka
#1239 - Include xinetd dependency from headnodes for haproxy
#1240 - Move HAProxy maxconns for MySQL in lockstep with max_connections changes
#1241 - backup /etc/chef folder as well
#1242 - Fix incorrect comment regarding proxy URL schema
#1244 - Add logging and use unique filenames for S3 check
#1245 - Bash cleanup - [desrod::phase_1]
#1247 - Re-order prerequisite installations
#1249 - Exclude /var/lib/ceph from updatedb search paths
#1256 - Fixes to local cache update process
#1257 - rom-o-matic.net moved to .eu
#1259 - Toggle for conditional DNS forwarding

Minor version 6.4.0

This represents a bugfix release.

Upgrading

Some config options relating to proxy specification have changed in a
backwards-incompatible manner. Users will need to modify their configurations
accordingly (See #1227).

Full set of additional features and fixes includes:
#1153 - Refactor storage into separate roles
#1204 - Run PowerDNS in master mode
#1208 - Add empty recipes that can be overridden in prop
#1210 - Only filter on existing instances, not actively powered-on ones
#1211 - Add instance information to DNS TXT records
#1213 - Explicitly define the order of AZ joins/departs if in maintenance
#1214 - General DB cleanup script
#1215 - Add custom weigher that favors nodes with <1:1 CPU commit
#1216 - Update Nova policy with typo bugfix
#1217 - Add generic2/nondurable2 flavor classes
#1218 - Reformat for clarity && dir structure diagram.
#1223 - Add ability to set tc_malloc_thread_cache size for ceph
#1224 - Reduce S3 check timeout
#1225 - Tune default network send/receive buffers to be more aggressive
#1227 - Fix proxy specification to require full url.
#1228 - Revise TCP buffer tune to not allocate 32MB per socket descriptor. Closes #1229.

Minor version 6.3.0

This is the last scheduled release in the 6.x series.

Major features include support for Mitaka, which will serve as default
in subsequent release series.

Full set of additional features and fixes includes:
#1151 - Add Mitaka support.
#1195 - Add concept of diskless worknodes
#1197 - Detect headnode VIP jump
#1198 - Hardcode certain service ports into the firewall configuration
#1201 - Add rally support for mitaka
#1202 - Support classless float, management subnets for dns reverse zones.
#1203 - Move "option nolinger" to only affect MySQL
#1206 - Automatically archive stale nova-network virtual interfaces
#1207 - Small change to ignore subdirectories of cacerts dir.

Minor version 6.2.0

This represents a combined bugfix/feature release. Most of the features/hacks
specific to Openstack distributions before Liberty have been removed.

Monitoring

A number of checks have been re-implemented. Additionally, s3 endpoints are
now monitored.

Stability

Policy modifications no longer require restarts of openstack
services. Additionally, to address resource exhaustion, custom, tunable
resource limits are added for the nova-compute service (See #1166, #1186).

Upgrading

Since the last stable release, a new signing key for the td-agent repositories
has been deployed. Be aware and update accordingly (See #1192).

Full set of additional features and fixes includes:
#1132 - Provision private flavor access to specific projects.
#1161 - Add ipmi task tags to allow skipping in absence of ipmi connectivity.
#1162 - doc fix (hw-raid)
#1163 - Remove pre-Liberty clutter
#1164 - Add ability to specify readahead value for all disks
#1165 - Remove unused Ceph PG trigger
#1166 - Avoid restarting Openstack daemons on policy changes
#1167 - Rework ephemeral disk check
#1168 - Remove pre-installed Puppet/Chef from box
#1169 - Remove php from hypervisors
#1171 - Break TPM and hardware RNG into separate recipes
#1173 - Simplify cpupower control
#1174 - Allow users to update metadata on their instances
#1175 - Remove Apache 2 from work nodes
#1176 - Slightly decrease enthusiasm for scourging PHP from every node
#1177 - Enable Zabbix server parameters and add monitoring template for S3
#1180 - Add sync resource on host aggregates recipe.
#1182 - Parameterize Graphite replication factor
#1184 - Limit Zabbix template import run to VIP holder
#1185 - Move zabbix-agent after host-aggregates
#1186 - Allow setting custom nofile soft/hard limits for nova-compute
#1187 - Add lock/unlock to hypervisor restart
#1189 - Restart Apache as reloads raise errors
#1190 - Add playbooks to do independent stops and starts of hypervisors
#1191 - Fixes various minor issues in restart playboooks
#1192 - Update td-agent GPG key
#1193 - Ignore errors on nova lock/unlock
#1194 - Parameterize memcached connection count

Minor version 6.1.0

This represents the first minor release since the introduction of Openstack
Liberty as the default. It includes a number of features and fixes added
after a period of extensive testing with a large user base as opposed
to a battery of contrived test scenarios. A large focus has been placed
upon stability, including improvements to monitoring and some performance
optimizations. Even a fair amount of attention has been afforded to the
documentation of long-existing architectural and design decisions.

Build Improvements

The network architecture has now been documented (just prior to a re-working
in the forthcoming releases) and some 4.x series documentation is now
deprecated. Some cosmetic changes relating to Horizon are now visible and
underlying query optimizations have been implemented to improve convergence
times during build.

Monitoring

• Some RGW checks have been deprecated.
• Ansible playbooks have been extended to handle deployments of monitoring nodes.
• IPMI metrics are collected. See #1101, #1105.
• MySQL slow queries are now logged for monitoring.

Stability

Much of the improvements involve migrating to, or adding support for more
recent versions of Openstack APIs in various places:

• Support for Glance APIv1 in recipe glanceclient commands is now removed.
• Support for Identity APIv3 to dns population scripts is added.
• Nova APIv2.1 is preferred over Nova APIv2.0.

Additionally, a more robust strategy for handling Fernet keys has been
implemented (See #1136), and a more complete set of Nova Scheduler Filters
is specified (See #1126, #1159). Ceph has been re-configured to mitigate
monitor node starvation and properly support some tunables (See #1098, #1105,
#1120). Finally, incorrect quota calculations, effectively resulting in a
denial of service to tenants preferring interaction through the Horizon UI
have been addressed (See #1102).

Upgrading

Though there are no known issues at the time of this release with upgrading
from 6.0.0 to 6.1.0, there does exists an important series of commits from
6.1.0 that would be required in order to build 6.0.0 successfully. These are:

• 9cd7513dcfb3a7983a8e613c96c1d75e5125d993
• 9d17bd50a40eeabb8d61ad5e6f53e32f864a5647
• c957c6b0a7ee3881485ebbefdc384006f1c53efb (full monitoring build only)

Full set of additional features and fixes includes:
#978 - Update vagrant_create.sh
#1087 - Remove zabbix RGW checks
#1088 - Always specify Glance API v2 since glanceclient commands are in v2 format.
#1089 - Delete some legacy documentation
#1093 - Add custom flavor sorting in horizon
#1095 - Extend cobbler and chef bootstrap functionality to monitoring nodes
#1098 - Do not re-nice osds on headnodes. Set primary-affinity to 0 on headnodes
#1100 - Works as expected
#1101 - Ensure ipmi_devintf kernel module is inserted
#1102 - Change compute:get_all_tenants to avoid quota summation across tenancies
#1104 - Update path to validation key on bootstrap node
#1105 - Fixes #1091: do not reapply optimal tunables if anything is changed
#1106 - Fixes #1096: override td-agent restart command
#1107 - Fix a few minor Ansible 2.0 deprecation warnings
#1108 - Ensure rabbitmq is started up
#1110 - Add support for rotating Fernet keys after a max age is reached
#1111 - Fix knife format parameter
#1113 - Move imports back to top-level in dns_fill.py
#1115 - Add option to deploy Berks packaged cookbooks via Ansible
#1116 - Check /vnc_auto.html for novnc status
#1117 - Correctly detect mysql ping failure
#1118 - Fixes #1097: boot in text mode for LOM serial consoles
#1119 - Install python-ldappool to work around Canonical Liberty packaging bug
#1120 - Create separate ceph health warn trigger for blocked ops
#1122 - Fix upstream bug #1608934: broken ephemeral disk creation in Liberty
#1123 - Fixes upstream #1593571: Horizon project workflow fails with LDAP bac…
#1124 - Use Nova API v2.1 instead of v2
#1125 - Ensure br_netfilter is loaded if present (kernel 3.18+)
#1126 - Extend ability to manage CPU and NUMA configurations
#1127 - Parameterize Glance debug logging
#1128 - Parameterize innodb_buffer_pool_size and innodb_buffer_pool_instances
#1129 - Remove API endpoint information and direct people to Horizon
#1130 - Add biosgrub partition so preseed is happy regardless of partition type
#1134 - Put the booleans in need_to_add_keys so #any? returns true
#1136 - Works as expected
#1138 - Make tftpd-hpa explicitly listen on IPv4/6 addresses by default
#1139 - Make dns_fill.py compliant with Keystone v3
#1140 - Parameterize MySQL slow query logging
#1141 - Add monitoring for MySQL slow queries
#1142 - Use partial search and only return attributes we need
#1143 - Additional monitoring band and parameterized Diamond handlers
#1146 - draft network architecture, for V1 architecture (layer 2 spanned)
#1147 - Switch slow query logging off during database backups
#1148 - Add new Percona signing key
#1155 - Terminate swap partition entry with a period
#1157 - Install/upgrade xtrabackup before MySQL
#1158 - Fix various minor issues in the restart playbook
#1159 - Add core and disk filters

Major version 6.0.0

First release in Liberty series. Moving forward, Kilo is considered deprecated, though support remains in the codebase to allow the migration from 5.x to 6.0.0.

Build Improvements

• Apt/dpkg behavior has changed with respect to upgraded packages. Older versions of config files will be retained where possible, which should ease conflicts with files under Chef management and simplify recipes.

Upgrading

See Installing Liberty, though installing BCPC 6.0.0 will include Liberty by default. Otherwise, updating sources to version 6.0.0 and running chef-client across all nodes will also achieve the same effect. For a list of additional features/fixes added by liberty, visit the following:

http://docs.openstack.org/releasenotes/cinder/liberty.html
http://docs.openstack.org/releasenotes/glance/liberty.html
http://docs.openstack.org/releasenotes/horizon/liberty.html
http://docs.openstack.org/releasenotes/keystone/liberty.html
http://docs.openstack.org/releasenotes/nova/liberty.html

In addition, some BCPC-specific features have been added:

• Horizon will now provide API version environment variables in generated OpenRC files
• Keystone is now configured to use Fernet Tokens
• Identity API is now V3

Known issues

• Rally deployment checks will complain about service names in the catalog. Something like:

NOTE: '__unknown__' service name means that Keystone service catalog doesn't return name for this service and Rally can not identify service by its type. BUT you still can use such services with api_versions context, specifying type of service (execute `rally plugin show api_versions` for more details).

Minor revision 5.13.1

This is a bugfix release. Additionally, some logs have been configured to
rotate.

Full set of additional features and fixes includes:

#1076 - Don't delete old flavors automatically in order to allow for transition
#1080 - Chef Server upgrade fixes and cleanup
#1084 - Add logrotate_app resource for memcached and parameterize verbose logs

Minor version 5.13.0

This represents the last stable release of the openstack kilo-based 5.x
series. A number of bug fixes and new features have been integrated into
the release. It is compatible with 5.12.0, but requires a few preparatory
steps to upgraded cleanly; See below in Upgrading for further details.

Build Improvements

A fair amount of work has gone into the ansible playbooks, including further
improvements to the no-clobber features added in 5.11.0 for parallel ansible
and standard chef-bcpc builds. Additionally, newer Chef versions are now
supported, build-generated binary files are now stored in a separate cookbook,
and all remaining configurations supporting Ubuntu Precise have been removed.

Monitoring

• Zabbix auto-registration is now handled by a Chef Provider.
• CPU stats collection is now normalized in Diamond.
• Graphite now supports remote database backends.
• Monitoring utilizes ipsets to in firewall rule management.

Stability

Unicast DHCP requests are now handled correctly by nova-network and nodes can
now be added and removed from a maintenance state by setting an attribute:
See #1054, #1028. Additionally, the manner by which PG counts are calculated
has changed to address issues of insufficient placement groups (See
#1043). Finally, only validated labels are generated for tenant VM DNS records.

Upgrading

A change to the mechanism for handling build-generated binary artifacts will
require any potential upgrade to re-generate all those previously generated
as they have been relocated.

Once the updated sources have been distributed to
vm-bootstrap:~vagrant/chef-bcpc, and before running chef-client
on all nodes, something like the following, executed in the
bootstrap/vagrant_scripts directory, should do the trick:

$ vagrant ssh -c "sudo bash -c '$(./dump_config.sh | sed 's/^/export /') ; cd chef-bcpc && source \$HOME/proxy_config.sh && bootstrap/shared/shared_build_bins.sh'" vm-bootstrap

If chef-client is executed prior to re-generating the artifacts in the
correct destination, the upgrade procedure will simply fail. Running the
above command first and rerunning chef-client on all nodes should rectify
the issue.

Dependencies

A number of dependencies have been updated:

• Minimum openstack version supported is 2015.1.2
• Chef
  • server: 12.0.8 -> 12.6.0
  • client: 12.3.0 -> 12.9.41
  • dependency cookbooks
    • apt: 3.0.0
    • concat: 0.3.3
    • cron: 1.7.6
    • hostsfile: 2.4.5
    • logrotate: 1.9.2
    • ntp: 2.0.0
    • ubuntu: 1.2.0
    • windows: 1.40.0
• Diamond: 3409b118be6af7b3ae8868297a2edf8cc7c5614c

Miscellanea

• All flavors have been renamed, others adjusted (See #1039)
• Configured signing key for rabbitmq packages has been updated.
  This affects previous releases as well.
• Ubuntu preseed template forces install on /dev/sda

Full set of additional features and fixes includes:

#1006 - Change from using cp -r to git clone
#1007 - Install dhcpdump
#1008 - Enable compatibility with newer Chef versions. Closes #953.
#1009 - Patch nova-network so unicast DHCP requests are responded to
#1011 - Add log output to ephemeral test script and name LV with UUID
#1013 - Always roll up Graphite metrics to next archive
#1014 - Parameterize max_concurrent_builds in nova.conf and reduce default to 4
#1016 - Raise minimum required OpenStack version to 2015.1.2
#1017 - Persist ipset setup across reboots
#1018 - Provider for creating Zabbix client auto-registration actions
#1019 - Update Zabbix templates
#1021 - Adds control_openstack command
#1024 - Wait to give nova-network time to set up tenant networks
#1027 - Improve ephemeral test behavior regarding autobackup behavior
#1028 - Add node maintenance flag
#1029 - Normalize Diamond CPU collection
#1030 - Parameterize Graphite metric whitelist/blacklist
#1031 - Make the EC2 compatibility APIs work
#1032 - Make installation of gems for fluentd idempotent
#1034 - Upgrade Diamond to 4.0.439
#1036 - Only allow LVM to scan whitelisted physical disks
#1037 - Fixes error introduced by #1036 that causes failure to converge
#1038 - Update Chef download URLs
#1039 - Rename flavors and purge legacy flavor names
#1042 - Re-work patches to handle python-nova 2015.1.2 - 2015.1.4
#1043 - TEST FIRST - Updated default PG and added nearest power_of_2
#1044 - Make bcpc_cephconfig provider idempotent
#1045 - Update README.MD
#1046 - Update to most current available versions of dependency cookbooks
#1047 - Separate monitoring roles
#1048 - Access the database directly to work around Nova API pagination
#1050 - Break binary files off into separate cookbook
#1051 - Fixes #1012: install whois package to provide mkpasswd
#1053 - Fixes #1049: generates validated hostnames in dns_fill.py
#1054 - Fixes #1033.
#1055 - Ensure all HAProxy stanzas have binds configured on their own line
#1057 - Replace RabbitMQ signing key
#1060 - Upgrade Graphite to 0.9.15
#1061 - Toggle for bootstrap data location
#1062 - Control I/O scheduler
#1065 - Update dependency cookbook versions (fixes #1064)
#1066 - Extend Cobbler for heterogeneous hardware
#1071 - Only install local gem file - fixes #1070
#1074 - 5.13.0 additional fixes
#1075 - Return hostname based on procfs
#879 - Fixes #513 - makes scrub load threshold configurable
#904 - Adding the ability to mount the apt mirror via VirtualBox shared folders
#966 - updates to address Ansible 2.0 issues/deprecation warnings
#968 - Moving Chef playbook contents to task lists
#989 - Miscellaneous fixes to address issues and comments on upgrade playbooks

Minor version 5.12.0

This represents primary a feature release. Major areas of change include
Monitoring, Stability, and Build Improvements.

Build Improvements

Additional ansible playbooks have been included for the routine maintenance
of software deployed to clusters. Also implemented is the facility to better
control the kernel versions deployed on hardware during build as well as
those configured for boot on cluster nodes. See #999, #1003 for full details.

Monitoring

This release introduces baseline bootstrap node monitoring as well as basic
metrics for connection tracking. Nova notifications can now be toggled on/off.

Stability

• AdminTenant is afforded more generous quotas (#1004)
• apport is included (#1002)
• a lock is placed around powerdns service restarts (#987)

Upgrading

This is incompatible with earlier versions due to change in the naming of build
VMs in #990. It is recommended that the existing VM bootstrap is deleted
and a new build from the current sources executed, rather than attempting to
integrate changes into an existing build. It is still possible to do so,
but it will require some manual steps. See Upgrading to 5.12.0 further in
this document.

Full set of additional features and fixes includes:

#985 - Upstreaming playbooks that orchestrate clusterwide software upgrades
#987 - Fixes #986 (maybe): try to lock PowerDNS restart to avoid multiple spawns
#988 - Parameterize and add configuration for Nova notification system
#990 - Rename Vagrant-built bootstrap node
#992 - Add baseline bootstrap node monitoring
#993 - Disable selection of compute host in functional check
#994 - Parameterize and change conntrack defaults
#995 - Rework Keystone endpoint/service creation
#997 - Reorder sysctl loading
#999 - Allow specifying a particular kernel for GRUB to boot
#1001 - Make monitoring node searches more specific
#1002 - Add apport so that debugging information from app crashes is captured
#1003 - Parameterize bcpc.preseed.kernel
#1004 - Openstack quota bump fixes

Upgrading to 5.12.0

If you attempt integrate the changes from 5.12.0 on top of a previous release,
you may find yourself in the following situation:

$ vagrant global-status
id       name         provider   state   directory
-----------------------------------------------------------------------------------------------
7e7b9dc  bootstrap    virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
ed87122  vm-bootstrap virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
7ab95b4  vm1          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
aa16704  vm2          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
f21751d  vm3          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
c0edf75  vm4          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
3063d81  vm5          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
e38f557  vm6          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts

In this case, the solution is to delete the bootstrap VM as recommended
above. Attempting this after the fact is not recommended but possible. First
one would need to revert Vagrantfile to last stable release. It is not
possible to simply call vagrant destroy <id> as illustrated by the following
status query commands:

$ echo {vm-,}bootstrap | xargs -tn1 vagrant status
vagrant status vm-bootstrap
Current machine states:

vm-bootstrap              running (virtualbox)

The VM is running. To stop this VM, you can run `vagrant halt` to
shut it down forcefully, or you can run `vagrant suspend` to simply
suspend the virtual machine. In either case, to restart it again,
simply run `vagrant up`.
vagrant status bootstrap
The machine with the name 'bootstrap' was not found configured for
this Vagrant environment.

Reverting to a Vagrantfile that has bootstrap defined will allow us to
manipulate the bootstrap VM using the vagrant interface:

$ git checkout 5.11.1 -- Vagrantfile
$ echo {vm-,}bootstrap | xargs -tn1 vagrant status
vagrant status vm-bootstrap
The machine with the name 'vm-bootstrap' was not found configured for
this Vagrant environment.
vagrant status bootstrap
Current machine states:

bootstrap                 running (virtualbox)

The VM is running. To stop this VM, you can run `vagrant halt` to
shut it down forcefully, or you can run `vagrant suspend` to simply
suspend the virtual machine. In either case, to restart it again,
simply run `vagrant up`.


$ vagrant destroy bootstrap
    bootstrap: Are you sure you want to destroy the 'bootstrap' VM? [y/N] y
==> bootstrap: Forcing shutdown of VM...
==> bootstrap: Destroying VM and associated drives...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...
==> bootstrap: Running cleanup tasks for 'shell' provisioner...

Then the current Vagrantfile is restored and vagrant should return to
a sane state:

bash-3.2$ git checkout 5.12.0 -- Vagrantfile
bash-3.2$ vagrant global-status
id       name         provider   state   directory
-----------------------------------------------------------------------------------------------
ed87122  vm-bootstrap virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
7ab95b4  vm1          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
aa16704  vm2          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
f21751d  vm3          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
c0edf75  vm4          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
3063d81  vm5          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts
e38f557  vm6          virtualbox running /Users/bcpc/build/chef-bcpc/bootstrap/vagrant_scripts

Minor revision 5.11.1

This is a maintenance release. An issue with the way monitoring alerts are generated has been
addressed. Details can be found below:

#981 - Report ceph health details to Zabbix and pagerduty

Minor version 5.11.0

This represents a combined bugfix/feature release.

Notable features include increased configurability for both the build process
and the resulting openstack deployment. Additionally, parallel build
environments for the standard build and local hardware build simulations are
supported. Various minor bugs have been addressed. See the full listing below.

Upgrading

Package versions are no longer specified explicitly in the Chef attributes
files; rather package repositories are specified instead. This entails that the
latest available versions of certain package dependencies will be installed
during the upgrade, triggering other packages to also be upgraded in certain
situations. See #875 for more information. Furthermore, successfully
authenticating to radosgw as the Admin user will require a manual restart of
the radosgw-all service. See #885 for additional details.

Full set of additional features and fixes includes:

#875 - Removing explicit package versions, adding repository pins
#885 - Remove Apache frontend for RADOSgw
#961 - Allow specification of chef versions
#962 - Make swap size configurable for VMs
#963 - Route VNC traffic through Haproxy to prevent SSL certificate issues
#964 - Parameterize rbd_flatten_volume_from_snapshot/rbd_max_clone_depth
#965 - Update broken link
#967 - Spawn local vms without clobber
#970 - Add ability to suppress ceph 'pg max obj skew' warning
#971 - Add ability to use vip instead of domain name for VNC proxy
#972 - Lower default RAM quota to 8Gb vs 50Gb
#975 - Fixes #974: NS record content should be cluster domain, not IP
#976 - Parameterize Carbon's MAX_UPDATES_PER_SECOND
#977 - Split default attributes file into multiple files
#980 - Parameterize sync_power_state_interval and resume_guests_state_on_host_boot
#983 - Remove default instance quota
#984 - Parameterize and increase Zabbix check timeouts