Releases: vmware-archive/pcf-pipelines
v0.15.3
Pipeline changes:
Install PCF on vSphere
- don't trigger deploy on new opsman; this is meant to only run once
- bump default param regexes from 1.10 to 1.11
- add support for NSX configuration
Install PCF on AWS
- bump default param regexes from 1.10 to 1.11
- Replace out of date README with better documented params
- add Known Issues section to README
Install PCF on GCP
- bump default param regexes from 1.10 to 1.11
- add Known Issue for #106 (google_compute_ssl_certificate recreated)
- Fix out of date job name in README
- Add support for adding trusted certs to Ops Manager
Upgrade Ops Manager on vSphere
- bump default param regexes from 1.10 to 1.11
- default opsman_timeout_seconds (3600)
- remove iaas_type param
- add Known Issue relating to failure to swap IP from old to new instance
Upgrade Ops Manager on AWS
- bump default param regexes from 1.10 to 1.11
Upgrade Ops Manager on GCP
- bump default param regexes from 1.10 to 1.11
Upgrade Tile
- bump default param regexes from 1.10 to 1.11
Upgrade ERT
- bump default param regexes from 1.10 to 1.11
Bugfixes:
- allow only patch upgrades works with multiple tiles having "cf" in their name
- import ops manager task specifies a 24-hour timeout to avoid import failures with large files
- upload-product-and-stemcell explicitly uploads the newest dependent stemcell
- upload-product-and-stemcell no longer deletes the downloaded stemcell in the container
Rootfs changes (czero/cflinuxfs2):
- add steamroll (https://github.com/krishicks/concourse-pipeline-steamroller)
New opfiles:
- create-install-pcf-vsphere-offline (not yet ready, requires Concourse 3.3.3)
New pipelines:
- install-pcf/openstack
This pipeline presently is a work in progress; it only supports opsman/ert 1.10
v0.15.2
- last tested with Ops Manager 1.11.0 and Elastic Runtime 1.11.0
Bug Fixes
- Fixed an issue where the upgrade pipelines were not running because of a bug with a recently introduced constraint on these pipelines which limited their use-cases to patch upgrades of PCF. These pipelines now work as expected, for patch upgrades only.
GCP
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | db_silk_username |
param was added earlier but not used, now it is used to configure ERT |
install-pcf | db_silk_password |
param was added earlier but not used, now it is used to configure ERT |
install-pcf | db_locket_username |
param was added earlier but not used, now it is used to configure ERT |
install-pcf | db_locket_password |
param was added earlier but not used, now it is used to configure ERT |
v0.15.1-rc.29 (since v0.15.1-rc.26)
- last tested with Ops Manager 1.11.0 and Elastic Runtime 1.11.0
vSphere
Install Pipeline Changes
- Renamed the Ops Mgr product PivNet resource from
pivnet-opsman-product
topcf-ops-manager
Task Changes
config-ert
added the two new params mentioned below.
Bug Fixes
- (vSphere install-pcf) Addressed issue from PR #129, which affected disabling errands
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | syslog_adapter_instances |
new param |
install-pcf | syslog_scheduler_instances |
new param |
GCP
Install Pipeline Changes:
- The terraform state file in the install-pipeline for GCP is now stored on Google Cloud Storage instead of AWS S3, since GCS is native to the GCP platform. The pipeline thus now uses the GCS resource type.
Task Changes
- Added
create-initial-terraform-state
task
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | terraform_statefile_bucket |
new param |
install-pcf | s3_endpoint |
removed |
install-pcf | s3_bucket_terraform |
removed |
install-pcf | s3_access_key_id |
removed |
install-pcf | s3_secret_access_key |
removed |
install-pcf | s3_region |
removed |
v0.15.1-rc.26
- last tested GCP, AWS, and vSphere install-PCF pipelines with Ops Manager 1.11.0 and Elastic Runtime 1.11.0
Enhancements
- Tooling - added
yaml-patch
andom-linux-beta
to czero/cflinuxfs2 image - Improved testing for availability of Ops Manager API
vSphere
Task Changes
- Added
upload-product
request timeout of 3600 seconds - Added
wipe-env
job set -eu
added to alltask.sh
files- Removed
https://
scheme fromOPS_MGR_HOST
ininstall-ert/configure-json/task.sh
gen_ssl_certs.sh
moved out of tasks and intopcf-pipelines/scripts
- Removed
STEMCELL_GLOB
in favour ofIAAS
forupload-product-and-stemcell
task - Moved terraform folders out of tasks and into the
install-pcf
IAAS specific directories
Bug Fixes
- (vSphere install-pcf) Trusted certificates can now handle multi-line YAML strings
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | bosh_vm_folder |
new param |
install-pcf | bosh_template_folder |
new param |
install-pcf | bosh_disk_path |
new param |
install-pcf | om_generate_ssl_endpoint |
removed param |
upgrade-tile | product_version_regex |
previous name: product_version |
upgrade-tile | resource_name |
removed; name is now hardcoded as tile |
AWS
Task Changes
- Added
disable-errands
task - Added
upload-product
request timeout of 3600 seconds set -eu
added to alltask.sh
files- Removed
https://
scheme fromOPS_MGR_HOST
ininstall-ert/configure-json/task.sh
gen_ssl_certs.sh
moved out of tasks and intopcf-pipelines/scripts
wipe-env
now checks for existence of Ops Manager before attempting to delete the installation- Moved terraform folders out of tasks and into the
install-pcf
IAAS specific directories
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | opsman_allow_ssh_cidr_ranges |
new param |
install-pcf | opsman_allow_https_cidr_ranges |
new param |
install-pcf | db_locket_username |
new param |
install-pcf | db_locket_password |
new param |
install-pcf | db_silk_username |
new param |
install-pcf | db_silk_password |
new param |
install-pcf | S3_REGION |
new param |
install-pcf | S3_ENDPOINT |
new param |
upgrade-tile | product_version_regex |
previous name: product_version |
upgrade-tile | resource_name |
removed; name is now hardcoded as tile |
Note: (install-pcf) - now when ERT_SSL_CERT
is left blank, a cert is automatically generated
GCP
Task Changes
- Added
disable-errands
task - Added
upload-product
request timeout of 3600 seconds - Removed
https://
scheme fromOPS_MGR_HOST
ininstall-ert/configure-json/task.sh
gen_ssl_certs.sh
moved out of tasks and intopcf-pipelines/scripts
- (install-pcf) now using om's ability to configure bosh
- Moved terraform folders out of tasks and into the
install-pcf
IAAS specific directories
Param Changes
Pipeline | Param | Notes |
---|---|---|
install-pcf | db_locket_username |
new param |
install-pcf | db_locket_password |
new param |
install-pcf | db_silk_username |
new param |
install-pcf | db_silk_password |
new param |
install-pcf | S3_REGION |
new param |
install-pcf | S3_ENDPOINT |
new param |
upgrade-tile | product_version_regex |
previous name: product_version |
upgrade-tile | resource_name |
removed; name is now hardcoded as tile |
v0.15.0-rc.7
Task Changes
- unpack-tarball is less verbose
Bug Fixes
- install-pcf/vsphere uses same ERT in upload-ert in deploy-ert
Pipeline Changes
Install PCF on vSphere
Job Changes
- Rename configure-ops-director to configure-director
v0.15.0-rc.6
This released focused on making changes to the Install PCF on vSphere pipeline more consistent with the AWS/GCP pipelines as far as job/task layout. Additional attention was put into making the jobs more idempotent or repeatable.
A large change was made to the way we configure ERT; we're now using a more modern technique for manipulating JSON than before.
Compatibility
The Install PCF on vSphere pipeline has been tested with this release to install OpsMgr v1.10.9.0 and ERT v1.10.12-build.2
Bug Fixes
- install-pcf/vsphere doesn't redownload opsmgr for tasks that don't need it
Pipeline Changes
Install PCF on vSphere
- Remove comments from pipeline
- Replace disable-ert-errands with configurable disable-errands task
- Use new mechanism for configuring ERT tile
Task/Job Changes
- Move config-opsman task into configure-ops-director job
- Move stage-tile task into deploy-ert job
- Rename install-opsmgr to create-opsman-vm
- Rename create-opsman-vm to deploy-opsman
- Rename config-opsdir to configure-director
- Rename opsman-apply-changes to deploy-director
- Rename upload-elastic-runtime to upload-ert
- Rename upload-er-tile to upload-tile
- Rename config-elastic-runtime-tile to deploy-ert
- Rename config-er-tile to configure
- Collapse product-apply-changes into deploy-ert
Params Changes
- Add
ert_major_minor_version
v0.15.0-rc.3
Params Changes
- All parameterized _major_minor_version values by default use a fully anchored version, e.g.
^1\.10\..*$
Bug Fixes
- gcp/wipe-env checks for existence of ops manager before trying to nuke installation
- upload-product-and-stemcell works when no stemcells are uploaded
- upgrade-ops-manager/gcp uses correct IAAS_TYPE
- upgrade-ert/params.yml uses correct iaas_type
- install-pcf/vsphere tracks product versions sorting by semver
Pipeline Changes
Upgrade ERT
- now takes a parameter for ERT minor version to track
Install PCF on AWS
- aws-pcf-prepare is now create-infrastructure
- aws-pcf-destroy is now wipe-env
- create-infrastructure is idempotent
- config-opsman/config-director are rolled into one job
- Latest tracked version of OpsMgr AMI is automatically chosen for installation
- AWS RDS instance is changed from mysql 5.6.27 to mariadb 10.1.19
- Individual credentials are now being used for database access
- Director certificates can now be specified
- Un-pin pivnet resource from v0.26.1
Params Changes
TILE_UPLOAD_TIMEOUT
is removedTF_VAR_opsman_ami
is removedTF_VAR_environment
is renamed toterraform_prefix
director_certificates
is added
v0.15.0-rc.2
Bug fixes
- create-initial-terraform-state creates a correct bare .tfstate file
v0.15.0-rc.1
A big focus this release was on fixing lingering issues with the Install PCF on GCP pipeline. Future releases will place a strong focus on each of the Install PCF pipelines for the other IaaSes.
Params changes:
-
Install PCF on AWS
- Remove TF_VAR_rds_db_username_ertsql
- Remove TF_VAR_rds_db_password_ertsql
-
Install PCF on GCP
- Remove ert_sql_db_username
- Remove ert_sql_db_password
- Remove tile_upload_timeout
- Change pcf_ert_ssl_cert to support 'generate' as the value (which is also the default)
- Add ospman_uri
-
Install PCF on vSphere
- Add opsman_ssh_pasword
Pipeline changes:
-
opsman_uri is no longer expected to be prepended with https protocol
-
Install PCF on GCP
- Add bootstrap-terraform-state job
- Change where terraform state file is stored (root of bucket rather than in bucket/terraform)
- Remove dependency on tfstate-version file
create-infrastructure
now uses terraform state file properlycreate-infrastructure
no longer rolls back in the face of errorsconfigure-director
no longer automatically runs aftercreate-infrastructure
upload-ert
is split out ofdeploy-ert
so they can be re-ran independently- Reduce parallelism of
create-infrastructure
to ward off API errors
-
Install PCF on vSphere
- elastic-runtime is now checked less frequently for updates
- elastic-runtime versions are now sorted by semver
- elastic-runtime is now hardcoded to track 1.10.x
Task changes:
-
upload-product-and-stemcell
- Only try to upload stemcell when the product needs it
- Skip downloading stemcell when it already exists in OpsMgr
-
install-pcf-gcp/wipe-env
- Add wipe of OpsMgr installation + BOSH director in addition to Terraform-managed resources
Bugfixes:
-
Install PCF on GCP
- Fix problem with incorrect OPSMAN_URI value passed in
- Fix problem with incorrect value used for OpsMgr SSH password
- Fix problem with SQL authorized users not being set correctly
-
Install PCF on vSphere
- Fix broken reference to pivnet cli
-
czero/cflinuxfs2
- Fix problem with the cleanup task that broke the aws CLI
-
Remove reference to czero/busybox, which isn't a thing