161 KiB
0.6.15 (Unreleased)
FEATURES:
- New command:
terraform fmt
to automatically normalize config file style [GH-4955] - New interpolation function:
jsonencode
[GH-5890] - New provider:
fastly
[GH-5814] - New resource:
aws_iam_user_ssh_key
[GH-5774] - New resource:
aws_s3_bucket_notification
[GH-5473] - New resource:
cloudstack_static_nat
[GH-6004] - New resource:
consul_key_prefix
[GH-5988] - New resource:
triton_fabric
[GH-5920] - New resource:
triton_vlan
[GH-5920]
IMPROVEMENTS:
- provider/aws: Add support for Step Scaling in
aws_autoscaling_policy
[GH-4277] - provider/aws: Add support for
cname_prefix
toaws_elastic_beanstalk_environment
resource [GH-5966] - provider/aws: Adding outputs for elastic_beanstalk_environment resource [GH-5915]
- provider/aws: Adds
wait_for_ready_timeout
option toaws_elastic_beanstalk_environment
[GH-5967] - provider/aws: Allow
aws_db_subnet_group
description to be updated [GH-5921] - provider/aws: Change
aws_elb
access_logs to list type [GH-5065] - provider/aws: Making the Cloudwatch Event Rule Target
target_id
optional [GH-5787] - provider/aws: Timeouts for
elasticsearch_domain
are increased [GH-5910] - provider/aws:
aws_codecommit_repository
setdefault_branch
only if defined [GH-5904] - provider/aws:
aws_redshift_cluster
allows usernames with underscore in it [GH-5935] - provider/aws: normalize json for
aws_cloudwatch_event_rule
[GH-6025] - provider/aws: normalise json for
aws_sns_topic
[GH-6089] - provider/aws: Allow multiple EIPs to associate to single ENI [GH-6070]
- provider/aws: Add support for trigger_configuration to
aws_codedeploy_deployment_group
[GH-5599] - provider/clc: Override default
account
alias in provider config [GH-5785] - provider/datadog: Add heredoc support to message, escalation_message, and query [GH-5788]
- provider/docker: Add support for docker run --user option [GH-5300]
- provider/google: Accept GOOGLE_CLOUD_KEYFILE_JSON env var for credentials [GH-6007]
- provider/google: Make "project" attribute on provider configuration optional [GH-6112]
- provider/google: Add "project" argument and attribute to all GCP compute resources which inherit from the provider's value [GH-6112]
*provider/google: Deprecate unused "region" attribute in
global_forwarding_rule
; this attribute was never used anywhere in the computation of the resource [GH-6112] - provider/google: Read more common configuration values from the environment and clarify precedence ordering [GH-6114]
- provider/github: Add support for privacy to
github_team
[GH-6116] - provider/cloudstack: Deprecate
ipaddress
in favour ofip_address
in all resources [GH-6010] - provider/openstack: Allow subnets with no gateway [GH-6060]
- provider/openstack Add value_specs argument and attribute for routers [GH-4898]
BUG FIXES:
- provider/aws: Convert protocols to standard format for Security Groups [GH-5881]
- provider/aws: Fix
aws_route panic
when destination CIDR block is nil [GH-5781] - provider/aws: Fix issue re-creating deleted VPC peering connections [GH-5959]
- provider/aws: Fix issue with changing iops when also changing storage type to io1 on RDS [GH-5676]
- provider/aws: Fix issue with retrying deletion of Network ACLs [GH-5954]
- provider/aws: Fix potential crash when receiving malformed
aws_route
API responses [GH-5867] - provider/aws: Guard against empty responses from Lambda Permissions [GH-5838]
- provider/aws: Normalize and compact SQS Redrive, Policy JSON [GH-5888]
- provider/aws: Remove CloudTrail Trail from state if not found [GH-6024]
- provider/aws: Report better error message in
aws_route53_record
whenset_identifier
is required [GH-5777] - provider/aws: set ASG
health_check_grace_period
default to 300 [GH-5830] - provider/aws: Show human-readable error message when failing to read an EBS volume [GH-6038]
- provider/azurerm: Fix detection of
azurerm_storage_account
resources removed manually [GH-5878] - provider/docker: Docker Image will be deleted on destroy [GH-5801]
- provider/openstack: Fix resizing when Flavor Name changes [GH-6020]
- provider/openstack: Fix Disabling DHCP on Subnets [GH-6052]
- provider/vsphere: Add error handling to
vsphere_folder
[GH-6095]
0.6.14 (March 21, 2016)
FEATURES:
- New provider:
triton
- Manage Joyent Triton public cloud or on-premise installations (#5738) - New provider:
clc
- Manage CenturyLink Cloud resources (#4893) - New provider:
github
- Manage GitHub Organization permissions with Terraform config (#5194) - New provider:
influxdb
- Manage InfluxDB databases (#3478) - New provider:
ultradns
- Manage UltraDNS records (#5716) - New resource:
aws_cloudwatch_log_metric_filter
(#5444) - New resource:
azurerm_virtual_machine
(#5514) - New resource:
azurerm_template_deployment
(#5758) - New interpolation function:
uuid
(#5575)
IMPROVEMENTS:
- core: provisioners connecting via WinRM now respect HTTPS settings (#5761)
- provider/aws:
aws_db_instance
now makesidentifier
optional and generates a unique ID when it is omitted (#5723) - provider/aws:
aws_redshift_cluster
now allowspublicly_accessible
to be modified (#5721) - provider/aws:
aws_kms_alias
now allows name to be auto-generated with aname_prefix
(#5594)
BUG FIXES:
- core: Color output is now shown correctly when running Terraform on Windows (#5718)
- core: HEREDOCs can now be indented in line with configuration using
<<-
and hanging indent is removed (#5740) - core: Invalid HCL syntax of nested object blocks no longer causes a crash (#5740)
- core: Local directory-based modules now use junctions instead of symbolic links on Windows (#5739)
- core: Modules sourced from a Mercurial repository now work correctly on Windows (#5739)
- core: Address some issues with ignore_changes (#5635)
- core: Add a lock to fix an interpolation issue caught by the Go 1.6 concurrent map access detector (#5772)
- provider/aws: Fix crash when an
aws_rds_cluster_instance
is removed outside of Terraform (#5717) - provider/aws:
aws_cloudformation_stack
usetimeout_in_minutes
for retry timeout to prevent unecessary timeouts (#5712) - provider/aws:
aws_lambda_function
resources no longer error on refresh if deleted externally to Terraform (#5668) - provider/aws:
aws_vpn_connection
resources deleted via the console on longer cause a crash (#5747) - provider/aws: Fix crasher in Elastic Beanstalk Configuration when using options (#5756)
- provider/aws: Fix issue preventing
aws_opsworks_stck
from working with Windows set as the OS (#5724) - provider/digitalocean:
digitalocean_ssh_key
resources no longer cause a panic if there is no network connectivity (#5748) - provider/google: Default description
google_dns_managed_zone
resources to "Managed By Terraform" (#5428) - provider/google: Fix error message on invalid instance URL for
google_compute_instance_group
(#5715) - provider/vsphere: provide
host
to provisioner connections (#5558) - provisioner/remote-exec: Address race condition introduced with script cleanup step introduced in 0.6.13 (#5751)
0.6.13 (March 16, 2016)
BACKWARDS INCOMPATIBILITIES / NOTES:
- provider/aws:
aws_s3_bucket_object
fieldetag
is now trimming off quotes (returns raw MD5 hash) (#5305) - provider/aws:
aws_autoscaling_group
now supports metrics collection, so a diff installing the default value of1Minute
for themetrics_granularity
field is expected. This diff should resolve in the nextterraform apply
w/ no AWS API calls (#4688) - provider/consul:
consul_keys
key
blocks now respectdelete
flag for removing individual blocks. Previously keys would be deleted only when the entire resource was removed. - provider/google:
next_hop_network
ongoogle_compute_route
is now read-only, to mirror the behavior in the official docs (#5564) - state/remote/http: PUT requests for this backend will now have
Content-Type: application/json
instead ofapplication/octet-stream
(#5499)
FEATURES:
- New command:
terraform untaint
(#5527) - New resource:
aws_api_gateway_api_key
(#4295) - New resource:
aws_api_gateway_deployment
(#4295) - New resource:
aws_api_gateway_integration_response
(#4295) - New resource:
aws_api_gateway_integration
(#4295) - New resource:
aws_api_gateway_method_response
(#4295) - New resource:
aws_api_gateway_method
(#4295) - New resource:
aws_api_gateway_model
(#4295) - New resource:
aws_api_gateway_resource
(#4295) - New resource:
aws_api_gateway_rest_api
(#4295) - New resource:
aws_elastic_beanstalk_application
(#3157) - New resource:
aws_elastic_beanstalk_configuration_template
(#3157) - New resource:
aws_elastic_beanstalk_environment
(#3157) - New resource:
aws_iam_account_password_policy
(#5029) - New resource:
aws_kms_alias
(#3928) - New resource:
aws_kms_key
(#3928) - New resource:
google_compute_instance_group
(#4087)
IMPROVEMENTS:
- provider/aws: Add
repository_link
as a computed field foraws_ecr_repository
(#5524) - provider/aws: Add ability to update Route53 zone comments (#5318)
- provider/aws: Add support for Metrics Collection to
aws_autoscaling_group
(#4688) - provider/aws: Add support for
description
toaws_network_interface
(#5523) - provider/aws: Add support for
storage_encrypted
toaws_rds_cluster
(#5520) - provider/aws: Add support for routing rules on
aws_s3_bucket
resources (#5327) - provider/aws: Enable updates & versioning for
aws_s3_bucket_object
(#5305) - provider/aws: Guard against Nil Reference in Redshift Endpoints (#5593)
- provider/aws: Lambda S3 object version defaults to
$LATEST
if unspecified (#5370) - provider/aws: Retry DB Creation on IAM propigation error (#5515)
- provider/aws: Support KMS encryption of S3 objects (#5453)
- provider/aws:
aws_autoscaling_lifecycle_hook
now havenotification_target_arn
androle_arn
as optional (#5616) - provider/aws:
aws_ecs_service
validates number ofload_balancer
s before creation/updates (#5605) - provider/aws:
aws_instance
now allows changes to security groups without force new resource (#5193) - provider/aws: send Terraform version in User-Agent (#5621)
- provider/cloudflare: Change
cloudflare_record
type to ForceNew (#5353) - provider/consul:
consul_keys
now detects drift and supports deletion of individualkey
blocks (#5210) - provider/digitalocean: Guard against Nil reference in
digitalocean_droplet
(#5588) - provider/docker: Add support for
unless-stopped
to docker containerrestart_policy
(#5337) - provider/google: Mark
next_hop_network
as read-only ongoogle_compute_route
(#5564) - provider/google: Validate VPN tunnel peer_ip at plan time (#5501)
- provider/openstack: Add Support for Domain ID and Domain Name environment variables (#5355)
- provider/openstack: Add support for instances to have multiple ephemeral disks. (#5131)
- provider/openstack: Re-Add server.AccessIPv4 and server.AccessIPv6 (#5366)
- provider/vsphere: Add support for disk init types (#4284)
- provisioner/remote-exec: Clear out scripts after uploading (#5577)
- state/remote/http: Change content type of PUT requests to the more appropriate
application/json
(#5499)
BUG FIXES:
- core: Disallow negative indices in the element() interpolation function, preventing crash (#5263)
- core: Fix issue that caused tainted resource destroys to be improperly filtered out when using -target and a plan file (#5516)
- core: Fix several issues with retry logic causing spurious "timeout while waiting for state to become ..." errors and unnecessary retry loops (#5460), (#5538), (#5543), (#5553)
- core: Includes upstream HCL fix to properly detect unbalanced braces and throw an error (#5400)
- provider/aws: Allow recovering from failed CloudWatch Event Target creation (#5395)
- provider/aws: Fix EC2 Classic SG Rule issue when referencing rules by name (#5533)
- provider/aws: Fix
aws_cloudformation_stack
update forparameters
&capabilities
if unmodified (#5603) - provider/aws: Fix a bug where AWS Kinesis Stream includes closed shards in the shard_count (#5401)
- provider/aws: Fix a bug where ElasticSearch Domain tags were not being set correctly (#5361)
- provider/aws: Fix a bug where
aws_route
would show continual changes in the plan when not computed (#5321) - provider/aws: Fix a bug where
publicly_assessible
wasn't being set to state inaws_db_instance
(#5535) - provider/aws: Fix a bug where listener protocol on
aws_elb
resources was case insensitive (#5376) - provider/aws: Fix a bug which caused panics creating rules on security groups in EC2 Classic (#5329)
- provider/aws: Fix crash when
aws_lambda_function
VpcId is nil (#5182) - provider/aws: Fix error with parsing JSON in
aws_s3_bucket
policy attribute (#5474) - provider/aws:
aws_lambda_function
can be properly updated, either vias3_object_version
or viafilename
&source_code_hash
as described in docs (#5239) - provider/google: Fix managed instance group preemptible instance creation (#4834)
- provider/openstack: Account for a 403 reply when os-tenant-networks is disabled (#5432)
- provider/openstack: Fix crashing during certain network updates in instances (#5365)
- provider/openstack: Fix create/delete statuses in load balancing resources (#5557)
- provider/openstack: Fix race condition between instance deletion and volume detachment (#5359)
- provider/template: Warn when
template
attribute specified as path (#5563)
INTERNAL IMPROVEMENTS:
- helper/schema:
MaxItems
attribute on schema lists and sets (#5218)
0.6.12 (February 24, 2016)
BACKWARDS INCOMPATIBILITIES / NOTES:
- The
publicly_accessible
attribute onaws_redshift_cluster
resources now defaults to true
FEATURES:
- New command:
validate
to perform syntax validation (#3783) - New provider:
datadog
(#5251) - New interpolation function:
md5
(#5267) - New interpolation function:
signum
(#4854) - New resource:
aws_cloudwatch_event_rule
(#4986) - New resource:
aws_cloudwatch_event_target
(#4986) - New resource:
aws_lambda_permission
(#4826) - New resource:
azurerm_dns_a_record
(#5013) - New resource:
azurerm_dns_aaaa_record
(#5013) - New resource:
azurerm_dns_cname_record
(#5013) - New resource:
azurerm_dns_mx_record
(#5041) - New resource:
azurerm_dns_ns_record
(#5041) - New resource:
azurerm_dns_srv_record
(#5041) - New resource:
azurerm_dns_txt_record
(#5041) - New resource:
azurerm_dns_zone
(#4979) - New resource:
azurerm_search_service
(#5203) - New resource:
azurerm_sql_database
(#5003) - New resource:
azurerm_sql_firewall_rule
(#5057) - New resource:
azurerm_sql_server
(#4991) - New resource:
google_compute_subnetwork
(#5130)
IMPROVEMENTS:
- core: Backend names are now down cased during
init
in the same manner asremote config
(#5012) - core: Upgrade resource name validation warning to an error as planned (#5272)
- core: output "diffs didn't match" error details (#5276)
- provider/aws: Add
is_multi_region_trail
option to CloudTrail (#4939) - provider/aws: Add support for HTTP(S) endpoints that auto confirm SNS subscription (#4711)
- provider/aws: Add support for Tags to CloudTrail (#5135)
- provider/aws: Add support for Tags to ElasticSearch (#4973)
- provider/aws: Add support for deployment configuration to
aws_ecs_service
(#5220) - provider/aws: Add support for log validation + KMS encryption to
aws_cloudtrail
(#5051) - provider/aws: Allow name-prefix and auto-generated names for IAM Server Cert (#5178)
- provider/aws: Expose additional VPN Connection attributes (#5032)
- provider/aws: Return an error if no matching route is found for an AWS Route (#5155)
- provider/aws: Support custom endpoints for AWS EC2 ELB and IAM (#5114)
- provider/aws: The
cluster_type
onaws_redshift_cluster
resources is now computed (#5238) - provider/aws:
aws_lambda_function
resources now support VPC configuration (#5149) - provider/aws: Add support for Enhanced Monitoring to RDS Instances (#4945)
- provider/aws: Improve vpc cidr_block err message (#5255)
- provider/aws: Implement Retention Period for
aws_kinesis_stream
(#5223) - provider/aws: Enable
stream_arm
output for DynamoDB Table when streams are enabled (#5271) - provider/digitalocean:
digitalocean_record
resources now export a computedfqdn
attribute (#5071) - provider/google: Add assigned IP Address to CloudSQL Instance
google_sql_database_instance
(#5245) - provider/openstack: Add support for Distributed Routers (#4878)
- provider/openstack: Add support for optional cacert_file parameter (#5106)
BUG FIXES:
- core: Fix bug detecting deeply nested module orphans (#5022)
- core: Fix bug where
ignore_changes
could produce "diffs didn't match during apply" errors (#4965) - core: Fix race condition when handling tainted resource destroys (#5026)
- core: Improve handling of Provisioners in the graph, fixing "Provisioner already initialized" errors (#4877)
- core: Skip
create_before_destroy
processing during aterraform destroy
, solving several issues preventingdestroy
from working properly with CBD resources (#5096) - core: Error instead of panic on self var in wrong scope (#5273)
- provider/aws: Fix Copy of Tags to DB Instance when created from Snapshot (#5197)
- provider/aws: Fix DynamoDB Table Refresh to ensure deleted tables are removed from state (#4943)
- provider/aws: Fix ElasticSearch
domain_name
validation (#4973) - provider/aws: Fix issue applying security group changes in EC2 Classic RDS for aws_db_instance (#4969)
- provider/aws: Fix reading auto scaling group availability zones (#5044)
- provider/aws: Fix reading auto scaling group load balancers (#5045)
- provider/aws: Fix
aws_redshift_cluster
to allowpublicly_accessible
to be false (#5262) - provider/aws: Wait longer for internet gateways to detach (#5120)
- provider/aws: Fix issue reading auto scaling group termination policies (#5101)
- provider/cloudflare:
ttl
no longer shows a change on each plan oncloudflare_record
resources (#5042) - provider/docker: Fix the default docker_host value (#5088)
- provider/google: Fix backend service max_utilization attribute (#5075)
- provider/google: Fix reading of
google_compute_vpn_gateway
without an explicit (#5125) - provider/google: Fix crash when setting
ack_deadline_seconds
ongoogle_pubsub_subscription
(#5110) - provider/openstack: Fix crash when
access_network
was not defined in instances (#4966) - provider/powerdns: Fix refresh of
powerdns_record
no longer fails if the record name contains a-
(#5228) - provider/vcd: Wait for DHCP assignment when creating
vcd_vapp
resources with no static IP assignment (#5195)
0.6.11 (February 1, 2016)
BACKWARDS INCOMPATIBILITIES / NOTES:
- The
max_size
,min_size
anddesired_capacity
attributes onaws_autoscaling_schedule
resources now default to 0
FEATURES:
- New provider:
powerdns
- PowerDNS REST API (#4885) - New builtin function:
trimspace
for trimming whitespaces (#4910) - New builtin function:
base64sha256
for base64 encoding raw sha256 sum of a given string (#4899) - New resource:
openstack_lb_member_v1
(#4359)
IMPROVEMENTS:
- provider/template: Remove unnecessary mime-type validation from
template_cloudinit_config
resources (#4873) - provider/template: Correct spelling of "Boundary" in the part separator of rendered
template_cloudinit_config
resources (#4873) - provider/aws: Provide a better message if no AWS creds are found (#4869)
- provider/openstack: Ability to specify per-network Floating IPs (#4812)
BUG FIXES:
- provider/aws:
aws_autoscale_schedule
0 values (#4693) - provider/aws: Fix regression with VPCs and ClassicLink for regions that do not support it (#4879)
- provider/aws: Change VPC ClassicLink to be computed (#4933)
- provider/aws: Fix SNS Topic Refresh to ensure deleted topics are removed from state (#4891)
- provider/aws: Refactor Route53 record to fix regression in deleting records created in previous versions of Terraform (#4892)
- provider/azurerm: Fix panic if no creds supplied (#4902)
- provider/openstack: Changing the port resource to mark the ip_address as optional (#4850)
- provider/docker: Catch potential custom network errors in docker (#4918)
0.6.10 (January 27, 2016)
BACKWARDS INCOMPATIBILITIES / NOTES:
- The
-module-depth
flag available onplan
,apply
,show
, andgraph
now defaults to-1
, causing resources within modules to be expanded in command output. This is only a cosmetic change; it does not affect any behavior. - This release includes a bugfix for
$${}
interpolation escaping. These strings are now properly converted to${}
during interpolation. This may cause diffs on existing configurations in certain cases. - Users of
consul_keys
should note that thevalue
sub-attribute ofkey
will no longer be updated with the remote value of the key. It should be only used to set a key in Consul K/V. To reference key values, use thevar
attribute. - The 0.6.9 release contained a regression in
aws_autoscaling_group
capacity waiting behavior for configs wheremin_elb_capacity != desired_capacity
ormin_size != desired_capacity
. This release remedies that regression by un-deprecatingmin_elb_capacity
and restoring the prior behavior. - Users of
aws_security_group
may notice new diffs in initial plans with 0.6.10 due to a bugfix that fixes drift detection on nested security group rules. These new diffs should reflect the actual state of the resources, which Terraform previously was unable to see.
FEATURES:
- New resource:
aws_lambda_alias
(#4664) - New resource:
aws_redshift_cluster
(#3862) - New resource:
aws_redshift_parameter_group
(#3862) - New resource:
aws_redshift_security_group
(#3862) - New resource:
aws_redshift_subnet_group
(#3862) - New resource:
azurerm_cdn_endpoint
(#4759) - New resource:
azurerm_cdn_profile
(#4740) - New resource:
azurerm_network_interface
(#4598) - New resource:
azurerm_network_security_rule
(#4586) - New resource:
azurerm_route_table
(#4602) - New resource:
azurerm_route
(#4604) - New resource:
azurerm_storage_account
(#4698) - New resource:
azurerm_storage_blob
(#4862) - New resource:
azurerm_storage_container
(#4862) - New resource:
azurerm_storage_queue
(#4862) - New resource:
azurerm_subnet
(#4595) - New resource:
docker_network
(#4483) - New resource:
docker_volume
(#4483) - New resource:
google_sql_user
(#4669)
IMPROVEMENTS:
- core: Add
sha256()
interpolation function (#4704) - core: Validate lifecycle keys to show helpful error messages whe they are mistypes (#4745)
- core: Default
module-depth
parameter to-1
, which expands resources within modules in command output (#4763) - core: Variable types may now be specified explicitly using the
type
argument (#4795) - provider/aws: Add new parameters
az_mode
andavailability_zone(s)
in ElastiCache (#4631) - provider/aws: Allow ap-northeast-2 (Seoul) as valid region (#4637)
- provider/aws: Limit SNS Topic Subscription protocols (#4639)
- provider/aws: Add support for configuring logging on
aws_s3_bucket
resources (#4482) - provider/aws: Add AWS Classiclink for AWS VPC resource (#3994)
- provider/aws: Supporting New AWS Route53 HealthCheck additions (#4564)
- provider/aws: Store instance state (#3261)
- provider/aws: Add support for updating ELB availability zones and subnets (#4597)
- provider/aws: Enable specifying aws s3 redirect protocol (#4098)
- provider/aws: Added support for
encrypted
onebs_block_devices
in Launch Configurations (#4481) - provider/aws: Retry Listener Creation for ELBs (#4825)
- provider/aws: Add support for creating Managed Microsoft Active Directory and Directory Connectors (#4388)
- provider/aws: Mark some
aws_db_instance
fields as optional (#3138) - provider/digitalocean: Add support for reassigning
digitalocean_floating_ip
resources (#4476) - provider/dme: Add support for Global Traffic Director locations on
dme_record
resources (#4305) - provider/docker: Add support for adding host entries on
docker_container
resources (#3463) - provider/docker: Add support for mounting named volumes on
docker_container
resources (#4480) - provider/google: Add content field to bucket object (#3893)
- provider/google: Add support for
named_port
blocks ongoogle_compute_instance_group_manager
resources (#4605) - provider/openstack: Add "personality" support to instance resource (#4623)
- provider/packet: Handle external state changes for Packet resources gracefully (#4676)
- provider/tls:
tls_private_key
now exports attributes with public key in both PEM and OpenSSH format (#4606) - provider/vdc: Add
allow_unverified_ssl
for connections to vCloud API (#4811) - state/remote: Allow KMS Key Encryption to be used with S3 backend (#2903)
BUG FIXES:
- core: Fix handling of literals with escaped interpolations
$${var}
(#4747) - core: Fix diff mismatch when RequiresNew field and list both change (#4749)
- core: Respect module target path argument on
terraform init
(#4753) - core: Write planfile even on empty plans (#4766)
- core: Add validation error when output is missing value field (#4762)
- core: Fix improper handling of orphan resources when targeting (#4574)
- core: Properly handle references to computed set attributes (#4840)
- config: Detect a specific JSON edge case and show a helpful workaround (#4746)
- provider/openstack: Ensure valid Security Group Rule attribute combination (#4466)
- provider/openstack: Don't put fixed_ip in port creation request if not defined (#4617)
- provider/google: Clarify SQL Database Instance recent name restriction (#4577)
- provider/google: Split Instance network interface into two fields (#4265)
- provider/aws: Error with empty list item on security group (#4140)
- provider/aws: Fix issue with detecting drift in AWS Security Groups rules (#4779)
- provider/aws: Trap Instance error from mismatched SG IDs and Names (#4240)
- provider/aws: EBS optimised to force new resource in AWS Instance (#4627)
- provider/aws: Wait for NACL rule to be visible (#4734)
- provider/aws:
default_result
onaws_autoscaling_lifecycle_hook
resources is now computed (#4695) - provider/aws: fix ASG capacity waiting regression by un-deprecating
min_elb_capacity
(#4864) - provider/consul: fix several bugs surrounding update behavior (#4787)
- provider/mailgun: Handle the fact that the domain destroy API is eventually consistent (#4777)
- provider/template: Fix race causing sporadic crashes in template_file with count > 1 (#4694)
- provider/template: Add support for updating
template_cloudinit_config
resources (#4757) - provisioner/chef: Add ENV['no_proxy'] to chef provisioner if no_proxy is detected (#4661)
0.6.9 (January 8, 2016)
FEATURES:
- New provider:
vcd
- VMware vCloud Director (#3785) - New provider:
postgresql
- Create PostgreSQL databases and roles (#3653) - New provider:
chef
- Create chef environments, roles, etc (#3084) - New provider:
azurerm
- Preliminary support for Azure Resource Manager (#4226) - New provider:
mysql
- Create MySQL databases (#3122) - New resource:
aws_autoscaling_schedule
(#4256) - New resource:
aws_nat_gateway
(#4381) - New resource:
aws_network_acl_rule
(#4286) - New resources:
aws_ecr_repository
andaws_ecr_repository_policy
(#4415) - New resource:
google_pubsub_topic
(#3671) - New resource:
google_pubsub_subscription
(#3671) - New resource:
template_cloudinit_config
(#4095) - New resource:
tls_locally_signed_cert
(#3930) - New remote state backend:
artifactory
(#3684)
IMPROVEMENTS:
- core: Change set internals for performance improvements (#3992)
- core: Support HTTP basic auth in consul remote state (#4166)
- core: Improve error message on resource arity mismatch (#4244)
- core: Add support for unary operators + and - to the interpolation syntax (#3621)
- core: Add SSH agent support for Windows (#4323)
- core: Add
sha1()
interpolation function (#4450) - provider/aws: Add
placement_group
as an option foraws_autoscaling_group
(#3704) - provider/aws: Add support for DynamoDB Table StreamSpecifications (#4208)
- provider/aws: Add
name_prefix
to Security Groups (#4167) - provider/aws: Add support for removing nodes to
aws_elasticache_cluster
(#3809) - provider/aws: Add support for
skip_final_snapshot
toaws_db_instance
(#3853) - provider/aws: Adding support for Tags to DB SecurityGroup (#4260)
- provider/aws: Adding Tag support for DB Param Groups (#4259)
- provider/aws: Fix issue with updated route ids for VPC Endpoints (#4264)
- provider/aws: Added measure_latency option to Route 53 Health Check resource (#3688)
- provider/aws: Validate IOPs for EBS Volumes (#4146)
- provider/aws: DB Subnet group arn output (#4261)
- provider/aws: Get full Kinesis streams view with pagination (#4368)
- provider/aws: Allow changing private IPs for ENIs (#4307)
- provider/aws: Retry MalformedPolicy errors due to newly created principals in S3 Buckets (#4315)
- provider/aws: Validate
name
ondb_subnet_group
against AWS requirements (#4340) - provider/aws: wait for ASG capacity on update (#3947)
- provider/aws: Add validation for ECR repository name (#4431)
- provider/cloudstack: performance improvements (#4150)
- provider/docker: Add support for setting the entry point on
docker_container
resources (#3761) - provider/docker: Add support for setting the restart policy on
docker_container
resources (#3761) - provider/docker: Add support for setting memory, swap and CPU shares on
docker_container
resources (#3761) - provider/docker: Add support for setting labels on
docker_container
resources (#3761) - provider/docker: Add support for setting log driver and options on
docker_container
resources (#3761) - provider/docker: Add support for settings network mode on
docker_container
resources (#4475) - provider/heroku: Improve handling of Applications within an Organization (#4495)
- provider/vsphere: Add support for custom vm params on
vsphere_virtual_machine
(#3867) - provider/vsphere: Rename vcenter_server config parameter to something clearer (#3718)
- provider/vsphere: Make allow_unverified_ssl a configuable on the provider (#3933)
- provider/vsphere: Add folder handling for folder-qualified vm names (#3939)
- provider/vsphere: Change ip_address parameter for ipv6 support (#4035)
- provider/openstack: Increase instance timeout from 10 to 30 minutes (#4223)
- provider/google: Add
restart_policy
attribute togoogle_managed_instance_group
(#3892)
BUG FIXES:
- core: skip provider input for deprecated fields (#4193)
- core: Fix issue which could cause fields that become empty to retain old values in the state (#3257)
- provider/docker: Fix an issue running with Docker Swarm by looking up containers by ID instead of name (#4148)
- provider/openstack: Better handling of load balancing resource state changes (#3926)
- provider/aws: Treat
INACTIVE
ECS cluster as deleted (#4364) - provider/aws: Skip
source_security_group_id
determination logic for Classic ELBs (#4075) - provider/aws: Fix issue destroy Route 53 zone/record if it no longer exists (#4198)
- provider/aws: Fix issue force destroying a versioned S3 bucket (#4168)
- provider/aws: Update DB Replica to honor storage type (#4155)
- provider/aws: Fix issue creating AWS RDS replicas across regions (#4215)
- provider/aws: Fix issue with Route53 and zero weighted records (#4427)
- provider/aws: Fix issue with iam_profile in aws_instance when a path is specified (#3663)
- provider/aws: Refactor AWS Authentication chain to fix issue with authentication and IAM (#4254)
- provider/aws: Fix issue with finding S3 Hosted Zone ID for eu-central-1 region (#4236)
- provider/aws: Fix missing AMI issue with Launch Configurations (#4242)
- provider/aws: Opsworks stack SSH key is write-only (#4241)
- provider/aws: Update VPC Endpoint to correctly set route table ids (#4392)
- provider/aws: Fix issue with ElasticSearch Domain
access_policies
always appear changed (#4245) - provider/aws: Fix issue with nil parameter group value causing panic in
aws_db_parameter_group
(#4318) - provider/aws: Fix issue with Elastic IPs not recognizing when they have been unassigned manually (#4387)
- provider/aws: Use body or URL for all CloudFormation stack updates (#4370)
- provider/aws: Fix template_url/template_body conflict (#4540)
- provider/aws: Fix bug w/ changing ECS svc/ELB association (#4366)
- provider/aws: Fix RDS unexpected state config (#4490)
- provider/digitalocean: Fix issue where a floating IP attached to a missing droplet causes a panic (#4214)
- provider/google: Fix project metadata sshKeys from showing up and causing unnecessary diffs (#4512)
- provider/heroku: Retry drain create until log channel is assigned (#4823)
- provider/openstack: Handle volumes in "deleting" state (#4204)
- provider/rundeck: Tolerate Rundeck server not returning project name when reading a job (#4301)
- provider/vsphere: Create and attach additional disks before bootup (#4196)
- provider/openstack: Convert block_device from a Set to a List (#4288)
- provider/google: Terraform identifies deleted resources and handles them appropriately on Read (#3913)
0.6.8 (December 2, 2015)
FEATURES:
- New provider:
statuscake
(#3340) - New resource:
digitalocean_floating_ip
(#3748) - New resource:
aws_lambda_event_source_mapping
(#4093)
IMPROVEMENTS:
- provider/cloudstack: Reduce the number of network calls required for common operations (#4051)
- provider/aws: Make
publically_accessible
on anaws_db_instance
update existing instances instead of forcing new ones (#3895) - provider/aws: Allow
block_duration_minutes
to be set for spot instance requests (#4071) - provider/aws: Make setting
acl
on S3 buckets update existing buckets instead of forcing new ones (#4080) - provider/aws: Make updates to
assume_role_policy
modify existing IAM roles instead of forcing new ones (#4107)
BUG FIXES:
- core: Fix a bug which prevented HEREDOC syntax being used in lists (#4078)
- core: Fix a bug which prevented HEREDOC syntax where the anchor ends in a number (#4128)
- core: Fix a bug which prevented HEREDOC syntax being used with Windows line endings (#4069)
- provider/aws: Fix a bug which could result in a panic when reading EC2 metadata (#4024)
- provider/aws: Fix issue recreating security group rule if it has been destroyed (#4050)
- provider/aws: Fix issue with some attributes in Spot Instance Requests returning as nil (#4132)
- provider/aws: Fix issue where SPF records in Route 53 could show differences with no modification to the configuration (#4108)
- provisioner/chef: Fix issue with path separators breaking the Chef provisioner on Windows (#4041)
0.6.7 (November 23, 2015)
FEATURES:
- New provider:
tls
- A utility provider for generating TLS keys/self-signed certificates for development and testing (#2778) - New provider:
dyn
- Manage DNS records on Dyn - New resource:
aws_cloudformation_stack
(#2636) - New resource:
aws_cloudtrail
(#3094), (#4010) - New resource:
aws_route
(#3548) - New resource:
aws_codecommit_repository
(#3274) - New resource:
aws_kinesis_firehose_delivery_stream
(#3833) - New resource:
google_sql_database
andgoogle_sql_database_instance
(#3617) - New resource:
google_compute_global_address
(#3701) - New resource:
google_compute_https_health_check
(#3883) - New resource:
google_compute_ssl_certificate
(#3723) - New resource:
google_compute_url_map
(#3722) - New resource:
google_compute_target_http_proxy
(#3727) - New resource:
google_compute_target_https_proxy
(#3728) - New resource:
google_compute_global_forwarding_rule
(#3702) - New resource:
openstack_networking_port_v2
(#3731) - New interpolation function:
coalesce
(#3814)
IMPROVEMENTS:
- core: Improve message to list only resources which will be destroyed when using
--target
(#3859) - connection/ssh: Accept
private_key
contents instead of paths (#3846) - provider/google:
preemptible
option for instance_template (#3667) - provider/google: Accurate Terraform Version (#3554)
- provider/google: Simplified auth (DefaultClient support) (#3553)
- provider/google:
automatic_restart
,preemptible
,on_host_maintenance
options (#3643) - provider/google: Read credentials as contents instead of path (#3901)
- null_resource: Enhance and document [GH-3244, GH-3659]
- provider/aws: Add CORS settings to S3 bucket (#3387)
- provider/aws: Add notification topic ARN for ElastiCache clusters (#3674)
- provider/aws: Add
kinesis_endpoint
for configuring Kinesis (#3255) - provider/aws: Add a computed ARN for S3 Buckets (#3685)
- provider/aws: Add S3 support for Lambda Function resource (#3794)
- provider/aws: Add
name_prefix
option to launch configurations (#3802) - provider/aws: Add support for group name and path changes with IAM group update function (#3237)
- provider/aws: Provide
source_security_group_id
for ELBs inside a VPC (#3780) - provider/aws: Add snapshot window and retention limits for ElastiCache (Redis) (#3707)
- provider/aws: Add username updates for
aws_iam_user
(#3227) - provider/aws: Add AutoMinorVersionUpgrade to RDS Instances (#3677)
- provider/aws: Add
access_logs
to ELB resource (#3756) - provider/aws: Add a retry function to rescue an error in creating Autoscaling Lifecycle Hooks (#3694)
- provider/aws:
engine_version
is now optional for DB Instance (#3744) - provider/aws: Add configuration to enable copying RDS tags to final snapshot (#3529)
- provider/aws: RDS Cluster additions (
backup_retention_period
,preferred_backup_window
,preferred_maintenance_window
) (#3757) - provider/aws: Document and validate ELB
ssl_certificate_id
and protocol requirements (#3887) - provider/azure: Read
publish_settings
as contents instead of path (#3899) - provider/openstack: Use IPv4 as the default IP version for subnets (#3091)
- provider/aws: Apply security group after restoring
db_instance
from snapshot (#3513) - provider/aws: Make the AutoScalingGroup
name
optional (#3710) - provider/openstack: Add "delete on termination" boot-from-volume option (#3232)
- provider/digitalocean: Make
user_data
force a new droplet (#3740) - provider/vsphere: Do not add network interfaces by default (#3652)
- provider/openstack: Configure Fixed IPs through ports (#3772)
- provider/openstack: Specify a port ID on a Router Interface (#3903)
- provider/openstack: Make LBaaS Virtual IP computed (#3927)
BUG FIXES:
terraform remote config
: update--help
output (#3632)- core: Modules on Git branches now update properly (#1568)
- core: Fix issue preventing input prompts for unset variables during plan (#3843)
- core: Fix issue preventing input prompts for unset variables during refresh (#4017)
- core: Orphan resources can now be targets (#3912)
- helper/schema: Skip StateFunc when value is nil (#4002)
- provider/google: Timeout when deleting large
instance_group_manager
(#3591) - provider/aws: Fix issue with order of Termination Policies in AutoScaling Groups. This will introduce plans on upgrade to this version, in order to correct the ordering (#2890)
- provider/aws: Allow cluster name, not only ARN for
aws_ecs_service
(#3668) - provider/aws: Fix a bug where a non-lower-cased
maintenance_window
can cause unnecessary planned changes (#4020) - provider/aws: Only set
weight
on anaws_route53_record
if it has been set in configuration (#3900) - provider/aws: Ignore association not existing on route table destroy (#3615)
- provider/aws: Fix policy encoding issue with SNS Topics (#3700)
- provider/aws: Correctly export ARN in
aws_iam_saml_provider
(#3827) - provider/aws: Fix issue deleting users who are attached to a group (#4005)
- provider/aws: Fix crash in Route53 Record if Zone not found (#3945)
- provider/aws: Retry deleting IAM Server Cert on dependency violation (#3898)
- provider/aws: Update Spot Instance request to provide connection information (#3940)
- provider/aws: Fix typo in error checking for IAM Policy Attachments (#3970)
- provider/aws: Fix issue with LB Cookie Stickiness and empty expiration period (#3908)
- provider/aws: Tolerate ElastiCache clusters being deleted outside Terraform (#3767)
- provider/aws: Downcase Route 53 record names in state file to match API output (#3574)
- provider/aws: Fix issue that could occur if no ECS Cluster was found for a given name (#3829)
- provider/aws: Fix issue with SNS topic policy if omitted (#3777)
- provider/aws: Support scratch volumes in
aws_ecs_task_definition
(#3810) - provider/aws: Treat
aws_ecs_service
w/ Status==INACTIVE as deleted (#3828) - provider/aws: Expand ~ to homedir in
aws_s3_bucket_object.source
(#3910) - provider/aws: Fix issue with updating the
aws_ecs_task_definition
whereaws_ecs_service
didn't wait for a new computed ARN (#3924) - provider/aws: Prevent crashing when deleting
aws_ecs_service
that is already gone (#3914) - provider/aws: Allow spaces in
aws_db_subnet_group.name
(undocumented in the API) (#3955) - provider/aws: Make VPC ID required on subnets (#4021)
- provider/azure: Various bug fixes (#3695)
- provider/digitalocean: Fix issue preventing SSH fingerprints from working (#3633)
- provider/digitalocean: Fix the DigitalOcean Droplet 404 potential on refresh of state (#3768)
- provider/openstack: Fix several issues causing unresolvable diffs (#3440)
- provider/openstack: Safely delete security groups (#3696)
- provider/openstack: Ignore order of
security_groups
in instance (#3651) - provider/vsphere: Fix d.SetConnInfo error in case of a missing IP address (#3636)
- provider/openstack: Fix boot from volume (#3206)
- provider/openstack: Fix crashing when image is no longer accessible (#2189)
- provider/openstack: Better handling of network resource state changes (#3712)
- provider/openstack: Fix crashing when no security group is specified (#3801)
- provider/packet: Fix issue that could cause errors when provisioning many devices at once (#3847)
- provider/packet: Fix connection information for devices, allowing provisioners to run (#3948)
- provider/openstack: Fix issue preventing security group rules from being removed (#3796)
- provider/template:
template_file
: source contents instead of path (#3909)
0.6.6 (October 23, 2015)
FEATURES:
- New interpolation functions:
cidrhost
,cidrnetmask
andcidrsubnet
(#3127)
IMPROVEMENTS:
- "forces new resource" now highlighted in plan output (#3136)
BUG FIXES:
- helper/schema: Better error message for assigning list/map to string (#3009)
- remote/state/atlas: Additional remote state conflict handling for semantically neutral state changes (#3603)
0.6.5 (October 21, 2015)
FEATURES:
- New resources:
aws_codeploy_app
andaws_codeploy_deployment_group
(#2783) - New remote state backend:
etcd
(#3487) - New interpolation functions:
upper
andlower
(#3558)
BUG FIXES:
- core: Fix remote state conflicts caused by ambiguity in ordering of deeply nested modules (#3573)
- core: Fix remote state conflicts caused by state metadata differences (#3569)
- core: Avoid using http.DefaultClient (#3532)
INTERNAL IMPROVEMENTS:
- provider/digitalocean: use official Go client (#3333)
- core: extract module fetching to external library (#3516)
0.6.4 (October 15, 2015)
FEATURES:
- New provider:
rundeck
(#2412) - New provider:
packet
(#2260), (#3472) - New provider:
vsphere
: Initial support for a VM resource (#3419) - New resource:
cloudstack_loadbalancer_rule
(#2934) - New resource:
google_compute_project_metadata
(#3065) - New resources:
aws_ami
,aws_ami_copy
,aws_ami_from_instance
(#2784) - New resources:
aws_cloudwatch_log_group
(#2415) - New resource:
google_storage_bucket_object
(#3192) - New resources:
google_compute_vpn_gateway
,google_compute_vpn_tunnel
(#3213) - New resources:
google_storage_bucket_acl
,google_storage_object_acl
(#3272) - New resource:
aws_iam_saml_provider
(#3156) - New resources:
aws_efs_file_system
andaws_efs_mount_target
(#2196) - New resources:
aws_opsworks_*
(#2162) - New resource:
aws_elasticsearch_domain
(#3443) - New resource:
aws_directory_service_directory
(#3228) - New resource:
aws_autoscaling_lifecycle_hook
(#3351) - New resource:
aws_placement_group
(#3457) - New resource:
aws_glacier_vault
(#3491) - New lifecycle flag:
ignore_changes
(#2525)
IMPROVEMENTS:
- core: Add a function to find the index of an element in a list. (#2704)
- core: Print all outputs when
terraform output
is called with no arguments (#2920) - core: In plan output summary, count resource replacement as Add/Remove instead of Change (#3173)
- core: Add interpolation functions for base64 encoding and decoding. (#3325)
- core: Expose parallelism as a CLI option instead of a hard-coding the default of 10 (#3365)
- core: Add interpolation function
compact
, to remove empty elements from a list. (#3239), (#3479) - core: Allow filtering of log output by level, using e.g.
TF_LOG=INFO
(#3380) - provider/aws: Add
instance_initiated_shutdown_behavior
to AWS Instance (#2887) - provider/aws: Support IAM role names (previously just ARNs) in
aws_ecs_service.iam_role
(#3061) - provider/aws: Add update method to RDS Subnet groups, can modify subnets without recreating (#3053)
- provider/aws: Paginate notifications returned for ASG Notifications (#3043)
- provider/aws: Adds additional S3 Bucket Object inputs (#3265)
- provider/aws: add
ses_smtp_password
toaws_iam_access_key
(#3165) - provider/aws: read
iam_instance_profile
foraws_instance
and save to state (#3167) - provider/aws: allow
instance
to be computed inaws_eip
(#3036) - provider/aws: Add
versioning
option toaws_s3_bucket
(#2942) - provider/aws: Add
configuation_endpoint
toaws_elasticache_cluster
(#3250) - provider/aws: Add validation for
app_cookie_stickiness_policy.name
(#3277) - provider/aws: Add validation for
db_parameter_group.name
(#3279) - provider/aws: Set DynamoDB Table ARN after creation (#3500)
- provider/aws:
aws_s3_bucket_object
allows interpolated content to be set with newcontent
attribute. (#3200) - provider/aws: Allow tags for
aws_kinesis_stream
resource. (#3397) - provider/aws: Configurable capacity waiting duration for ASGs (#3191)
- provider/aws: Allow non-persistent Spot Requests (#3311)
- provider/aws: Support tags for AWS DB subnet group (#3138)
- provider/cloudstack: Add
project
parameter tocloudstack_vpc
,cloudstack_network
,cloudstack_ipaddress
andcloudstack_disk
(#3035) - provider/openstack: add functionality to attach FloatingIP to Port (#1788)
- provider/google: Can now do multi-region deployments without using multiple providers (#3258)
- remote/s3: Allow canned ACLs to be set on state objects. (#3233)
- remote/s3: Remote state is stored in S3 with
Content-Type: application/json
(#3385)
BUG FIXES:
- core: Fix problems referencing list attributes in interpolations (#2157)
- core: don't error on computed value during input walk (#2988)
- core: Ignore missing variables during destroy phase (#3393)
- provider/google: Crashes with interface conversion in GCE Instance Template (#3027)
- provider/google: Convert int to int64 when building the GKE cluster.NodeConfig struct (#2978)
- provider/google: google_compute_instance_template.network_interface.network should be a URL (#3226)
- provider/aws: Retry creation of
aws_ecs_service
if IAM policy isn't ready yet (#3061) - provider/aws: Fix issue with mixed capitalization for RDS Instances (#3053)
- provider/aws: Fix issue with RDS to allow major version upgrades (#3053)
- provider/aws: Fix shard_count in
aws_kinesis_stream
(#2986) - provider/aws: Fix issue with
key_name
and using VPCs with spot instance requests (#2954) - provider/aws: Fix unresolvable diffs coming from
aws_elasticache_cluster
names being downcased by AWS (#3120) - provider/aws: Read instance source_dest_check and save to state (#3152)
- provider/aws: Allow
weight = 0
in Route53 records (#3196) - provider/aws: Normalize aws_elasticache_cluster id to lowercase, allowing convergence. (#3235)
- provider/aws: Fix ValidateAccountId for IAM Instance Profiles (#3313)
- provider/aws: Update Security Group Rules to Version 2 (#3019)
- provider/aws: Migrate KeyPair to version 1, fixing issue with using
file()
(#3470) - provider/aws: Fix force_delete on autoscaling groups (#3485)
- provider/aws: Fix crash with VPC Peering connections (#3490)
- provider/aws: fix bug with reading GSIs from dynamodb (#3300)
- provider/docker: Fix issue preventing private images from being referenced (#2619)
- provider/digitalocean: Fix issue causing unnecessary diffs based on droplet slugsize case (#3284)
- provider/openstack: add state 'downloading' to list of expected states in
blockstorage_volume_v1
creation (#2866) - provider/openstack: remove security groups (by name) before adding security groups (by id) (#2008)
INTERNAL IMPROVEMENTS:
- core: Makefile target "plugin-dev" for building just one plugin. (#3229)
- helper/schema: Don't allow
Update
func if no attributes can actually be updated, per schema. (#3288) - helper/schema: Default hashing function for sets (#3018)
- helper/multierror: Remove in favor of github.com/hashicorp/go-multierror. (#3336)
0.6.3 (August 11, 2015)
BUG FIXES:
- core: Skip all descendents after error, not just children; helps prevent confusing additional errors/crashes after initial failure (#2963)
- core: fix deadlock possibility when both a module and a dependent resource are removed in the same run (#2968)
- provider/aws: Fix issue with authenticating when using IAM profiles (#2959)
0.6.2 (August 6, 2015)
FEATURES:
- New resource:
google_compute_instance_group_manager
(#2868) - New resource:
google_compute_autoscaler
(#2868) - New resource:
aws_s3_bucket_object
(#2898)
IMPROVEMENTS:
- core: Add resource IDs to errors coming from
apply
/refresh
(#2815) - provider/aws: Validate credentials before walking the graph (#2730)
- provider/aws: Added website_domain for S3 buckets (#2210)
- provider/aws: ELB names are now optional, and generated by Terraform if omitted (#2571)
- provider/aws: Downcase RDS engine names to prevent continuous diffs (#2745)
- provider/aws: Added
source_dest_check
attribute to the aws_network_interface (#2741) - provider/aws: Clean up externally removed Launch Configurations (#2806)
- provider/aws: Allow configuration of the DynamoDB Endpoint (#2825)
- provider/aws: Compute private ip addresses of ENIs if they are not specified (#2743)
- provider/aws: Add
arn
attribute for DynamoDB tables (#2924) - provider/aws: Fail silently when account validation fails while from instance profile (#3001)
- provider/azure: Allow
settings_file
to accept XML string (#2922) - provider/azure: Provide a simpler error when using a Platform Image without a Storage Service (#2861)
- provider/google:
account_file
is now expected to be JSON. Paths are still supported for backwards compatibility. (#2839)
BUG FIXES:
- core: Prevent error duplication in
apply
(#2815) - core: Fix crash when a provider validation adds a warning (#2878)
- provider/aws: Fix issue with toggling monitoring in AWS Instances (#2794)
- provider/aws: Fix issue with Spot Instance Requests and cancellation (#2805)
- provider/aws: Fix issue with checking for ElastiCache cluster cache node status (#2842)
- provider/aws: Fix issue when unable to find a Root Block Device name of an Instance Backed AMI (#2646)
- provider/dnsimple: Domain and type should force new records (#2777)
- provider/aws: Fix issue with IAM Server Certificates and Chains (#2871)
- provider/aws: Fix issue with IAM Server Certificates when using
path
(#2871) - provider/aws: Fix issue in Security Group Rules when the Security Group is not found (#2897)
- provider/aws: allow external ENI attachments (#2943)
- provider/aws: Fix issue with S3 Buckets, and throwing an error when not found (#2925)
0.6.1 (July 20, 2015)
FEATURES:
IMPROVEMENTS:
- connection/ssh: Print SSH bastion host details to output (#2684)
- provider/aws: Create RDS databases from snapshots (#2062)
- provider/aws: Add support for restoring from Redis backup stored in S3 (#2634)
- provider/aws: Add
maintenance_window
to ElastiCache cluster (#2642) - provider/aws: Availability Zones are optional when specifying VPC Zone Identifiers in Auto Scaling Groups updates (#2724)
- provider/google: Add metadata_startup_script to google_compute_instance (#2375)
BUG FIXES:
- core: Don't prompt for variables with defaults (#2613)
- core: Return correct number of planned updates (#2620)
- core: Fix "provider not found" error that can occur while running a destroy plan with grandchildren modules (#2755)
- core: Fix UUID showing up in diff for computed splat (
foo.*.bar
) variables. (#2788) - core: Orphan modules that contain no resources (only other modules) are properly destroyed up to arbitrary depth (#2786)
- core: Fix "attribute not available" during destroy plans in cases where the parameter is passed between modules (#2775)
- core: Record schema version when destroy fails (#2923)
- connection/ssh: fix issue on machines with an SSH Agent available
preventing
key_file
from being read without explicitly settingagent = false
(#2615) - provider/aws: Allow uppercase characters in
aws_elb.name
(#2580) - provider/aws: Allow underscores in
aws_db_subnet_group.name
(undocumented by AWS) (#2604) - provider/aws: Allow dots in
aws_db_subnet_group.name
(undocumented by AWS) (#2665) - provider/aws: Fix issue with pending Spot Instance requests (#2640)
- provider/aws: Fix issue in AWS Classic environment with referencing external Security Groups (#2644)
- provider/aws: Bump internet gateway detach timeout (#2669)
- provider/aws: Fix issue with detecting differences in DB Parameters (#2728)
- provider/aws:
ecs_cluster
rename (recreation) and deletion is handled correctly (#2698) - provider/aws:
aws_route_table
ignores routes generated for VPC endpoints (#2695) - provider/aws: Fix issue with Launch Configurations and enable_monitoring (#2735)
- provider/openstack: allow empty api_key and endpoint_type (#2626)
- provisioner/chef: Fix permission denied error with ohai hints (#2781)
0.6.0 (June 30, 2015)
BACKWARDS INCOMPATIBILITIES:
- command/push: If a variable is already set within Atlas, it won't be
updated unless the
-overwrite
flag is present (#2373) - connection/ssh: The
agent
field now defaults totrue
if theSSH_AGENT_SOCK
environment variable is present. In other words,ssh-agent
support is now opt-out instead of opt-in functionality. (#2408) - provider/aws: If you were setting access and secret key to blank ("") to force Terraform to load credentials from another source such as the EC2 role, this will now error. Remove the blank lines and Terraform will load from other sources.
concat()
has been repurposed to combine lists instead of strings (old behavior of joining strings is maintained in this version but is deprecated, strings should be combined using interpolation syntax, like "${var.foo}{var.bar}") (#1790)
FEATURES:
- New provider:
azure
[GH-2052, GH-2053, GH-2372, GH-2380, GH-2394, GH-2515, GH-2530, GH-2562] - New resource:
aws_autoscaling_notification
(#2197) - New resource:
aws_autoscaling_policy
(#2201) - New resource:
aws_cloudwatch_metric_alarm
(#2201) - New resource:
aws_dynamodb_table
(#2121) - New resource:
aws_ecs_cluster
(#1803) - New resource:
aws_ecs_service
(#1803) - New resource:
aws_ecs_task_definition
[GH-1803, GH-2402] - New resource:
aws_elasticache_parameter_group
(#2276) - New resource:
aws_flow_log
(#2384) - New resource:
aws_iam_group_association
(#2273) - New resource:
aws_iam_policy_attachment
(#2395) - New resource:
aws_lambda_function
(#2170) - New resource:
aws_route53_delegation_set
(#1999) - New resource:
aws_route53_health_check
(#2226) - New resource:
aws_spot_instance_request
(#2263) - New resource:
cloudstack_ssh_keypair
(#2004) - New remote state backend:
swift
: You can now store remote state in a OpenStack Swift. (#2254) - command/output: support display of module outputs (#2102)
- core:
keys()
andvalues()
funcs for map variables (#2198) - connection/ssh: SSH bastion host support and ssh-agent forwarding (#2425)
IMPROVEMENTS:
- core: HTTP remote state now accepts
skip_cert_verification
option to ignore TLS cert verification. (#2214) - core: S3 remote state now accepts the 'encrypt' option for SSE (#2405)
- core:
plan
now reports sum of resources to be changed/created/destroyed (#2458) - core: Change string list representation so we can distinguish empty, single element lists (#2504)
- core: Properly close provider and provisioner plugin connections [GH-2406, GH-2527]
- provider/aws: AutoScaling groups now support updating Load Balancers without recreation (#2472)
- provider/aws: Allow more in-place updates for ElastiCache cluster without recreating (#2469)
- provider/aws: ElastiCache Subnet Groups can be updated without destroying first (#2191)
- provider/aws: Normalize
certificate_chain
inaws_iam_server_certificate
to prevent unnecessary replacement. (#2411) - provider/aws:
aws_instance
supports `monitoring' (#2489) - provider/aws:
aws_launch_configuration
now supportsenable_monitoring
(#2410) - provider/aws: Show outputs after
terraform refresh
(#2347) - provider/aws: Add backoff/throttling during DynamoDB creation (#2462)
- provider/aws: Add validation for aws_vpc.cidr_block (#2514)
- provider/aws: Add validation for aws_db_subnet_group.name (#2513)
- provider/aws: Add validation for aws_db_instance.identifier (#2516)
- provider/aws: Add validation for aws_elb.name (#2517)
- provider/aws: Add validation for aws_security_group (name+description) (#2518)
- provider/aws: Add validation for aws_launch_configuration (#2519)
- provider/aws: Add validation for aws_autoscaling_group.name (#2520)
- provider/aws: Add validation for aws_iam_role.name (#2521)
- provider/aws: Add validation for aws_iam_role_policy.name (#2552)
- provider/aws: Add validation for aws_iam_instance_profile.name (#2553)
- provider/aws: aws_auto_scaling_group.default_cooldown no longer requires resource replacement (#2510)
- provider/aws: add AH and ESP protocol integers (#2321)
- provider/docker:
docker_container
has theprivileged
option. (#2227) - provider/openstack: allow
OS_AUTH_TOKEN
environment variable to set the openstackapi_key
field (#2234) - provider/openstack: Can now configure endpoint type (public, admin, internal) (#2262)
- provider/cloudstack:
cloudstack_instance
now supports projects (#2115) - provisioner/chef: Added a
os_type
to specifically specify the target OS (#2483) - provisioner/chef: Added a
ohai_hints
option to upload hint files (#2487)
BUG FIXES:
- core: lifecycle
prevent_destroy
can be any value that can be coerced into a bool (#2268) - core: matching provider types in sibling modules won't override each other's config. (#2464)
- core: computed provider configurations now properly validate (#2457)
- core: orphan (commented out) resource dependencies are destroyed in the correct order (#2453)
- core: validate object types in plugins are actually objects (#2450)
- core: fix
-no-color
flag in subcommands (#2414) - core: Fix error of 'attribute not found for variable' when a computed resource attribute is used as a parameter to a module (#2477)
- core: moduled orphans will properly inherit provider configs (#2476)
- core: modules with provider aliases work properly if the parent doesn't implement those aliases (#2475)
- core: unknown resource attributes passed in as parameters to modules now error (#2478)
- core: better error messages for missing variables (#2479)
- core: removed set items now properly appear in diffs and applies (#2507)
- core: '*' will not be added as part of the variable name when you attempt multiplication without a space (#2505)
- core: fix target dependency calculation across module boundaries (#2555)
- command/*: fixed bug where variable input was not asked for unset vars if terraform.tfvars existed (#2502)
- command/apply: prevent output duplication when reporting errors (#2267)
- command/apply: destroyed orphan resources are properly counted (#2506)
- provider/aws: loading credentials from the environment (vars, EC2 role, etc.) is more robust and will not ask for credentials from stdin (#1841)
- provider/aws: fix panic when route has no
cidr_block
(#2215) - provider/aws: fix issue preventing destruction of IAM Roles (#2177)
- provider/aws: fix issue where Security Group Rules could collide and fail to save to the state file correctly (#2376)
- provider/aws: fix issue preventing destruction self referencing Securtity Group Rules (#2305)
- provider/aws: fix issue causing perpetual diff on ELB listeners when non-lowercase protocol strings were used (#2246)
- provider/aws: corrected frankfurt S3 website region (#2259)
- provider/aws:
aws_elasticache_cluster
port is required (#2160) - provider/aws: Handle AMIs where RootBlockDevice does not appear in the BlockDeviceMapping, preventing root_block_device from working (#2271)
- provider/aws: fix
terraform show
with remote state (#2371) - provider/aws: detect
instance_type
drift onaws_instance
(#2374) - provider/aws: fix crash when
security_group_rule
referenced non-existent security group (#2434) - provider/aws:
aws_launch_configuration
retries if IAM instance profile is not ready yet. (#2452) - provider/aws:
fqdn
is populated during creation foraws_route53_record
(#2528) - provider/aws: retry VPC delete on DependencyViolation due to eventual consistency (#2532)
- provider/aws: VPC peering connections in "failed" state are deleted (#2544)
- provider/aws: EIP deletion works if it was manually disassociated (#2543)
- provider/aws:
elasticache_subnet_group.subnet_ids
is now a required argument (#2534) - provider/aws: handle nil response from VPN connection describes (#2533)
- provider/cloudflare: manual record deletion doesn't cause error (#2545)
- provider/digitalocean: handle case where droplet is deleted outside of terraform (#2497)
- provider/dme: No longer an error if record deleted manually (#2546)
- provider/docker: Fix issues when using containers with links (#2327)
- provider/openstack: fix panic case if API returns nil network (#2448)
- provider/template: fix issue causing "unknown variable" rendering errors when an existing set of template variables is changed (#2386)
- provisioner/chef: improve the decoding logic to prevent parameter not found errors (#2206)
0.5.3 (June 1, 2015)
IMPROVEMENTS:
- New resource:
aws_kinesis_stream
(#2110) - New resource:
aws_iam_server_certificate
(#2086) - New resource:
aws_sqs_queue
(#1939) - New resource:
aws_sns_topic
(#1974) - New resource:
aws_sns_topic_subscription
(#1974) - New resource:
aws_volume_attachment
(#2050) - New resource:
google_storage_bucket
(#2060) - provider/aws: support ec2 termination protection (#1988)
- provider/aws: support for RDS Read Replicas (#1946)
- provider/aws:
aws_s3_bucket
add support forpolicy
(#1992) - provider/aws:
aws_ebs_volume
add support fortags
(#2135) - provider/aws:
aws_elasticache_cluster
Confirm node status before reporting available - provider/aws:
aws_network_acl
Add support for ICMP Protocol (#2148) - provider/aws: New
force_destroy
parameter for S3 buckets, to destroy Buckets that contain objects (#2007) - provider/aws: switching
health_check_type
on ASGs no longer requires resource refresh (#2147) - provider/aws: ignore empty
vpc_security_group_ids
onaws_instance
(#2311)
BUG FIXES:
- provider/aws: Correctly handle AWS keypairs which no longer exist (#2032)
- provider/aws: Fix issue with restoring an Instance from snapshot ID (#2120)
- provider/template: store relative path in the state (#2038)
- provisioner/chef: fix interpolation in the Chef provisioner (#2168)
- provisioner/remote-exec: Don't prepend shebang on scripts that already have one (#2041)
0.5.2 (May 15, 2015)
FEATURES:
- Chef provisioning: You can now provision new hosts (both Linux and Windows) with Chef using a native provisioner (#1868)
IMPROVEMENTS:
- New config function:
formatlist
- Format lists in a similar way toformat
. Useful for creating URLs from a list of IPs. (#1829) - New resource:
aws_route53_zone_association
- provider/aws:
aws_autoscaling_group
can wait for capacity in ELB viamin_elb_capacity
(#1970) - provider/aws:
aws_db_instances
supportslicense_model
(#1966) - provider/aws:
aws_elasticache_cluster
add support for Tags (#1965) - provider/aws:
aws_network_acl
Network ACLs can be applied to multiple subnets (#1931) - provider/aws:
aws_s3_bucket
exportshosted_zone_id
andregion
(#1865) - provider/aws:
aws_s3_bucket
add support for websiteredirect_all_requests_to
(#1909) - provider/aws:
aws_route53_record
exportsfqdn
(#1847) - provider/aws:
aws_route53_zone
can create private hosted zones (#1526) - provider/google:
google_compute_instance
scratch
attribute added (#1920)
BUG FIXES:
- core: fix "resource not found" for interpolation issues with modules
- core: fix unflattenable error for orphans (#1922)
- core: fix deadlock with create-before-destroy + modules (#1949)
- core: fix "no roots found" error with create-before-destroy (#1953)
- core: variables set with environment variables won't validate as not set without a default (#1930)
- core: resources with a blank ID in the state are now assumed to not exist (#1905)
- command/push: local vars override remote ones (#1881)
- provider/aws: Mark
aws_security_group
description asForceNew
(#1871) - provider/aws:
aws_db_instance
ARN value is correct (#1910) - provider/aws:
aws_db_instance
only submit modify request if there is a change. (#1906) - provider/aws:
aws_elasticache_cluster
export missing information on cluster nodes (#1965) - provider/aws: bad AMI on a launch configuration won't block refresh (#1901)
- provider/aws:
aws_security_group
+aws_subnet
- destroy timeout increased to prevent DependencyViolation errors. (#1886) - provider/google:
google_compute_instance
Local SSDs no-longer cause crash (#1088) - provider/google:
google_http_health_check
Defaults now driven from Terraform, avoids errors on update (#1894) - provider/google:
google_compute_template
Update Instance Template network definition to match changes to Instance (#980) - provider/template: Fix infinite diff (#1898)
0.5.1 (never released)
This version was never released since we accidentally skipped it!
0.5.0 (May 7, 2015)
BACKWARDS INCOMPATIBILITIES:
- provider/aws: Terraform now remove the default egress rule created by AWS in a new security group.
FEATURES:
- Multi-provider (a.k.a multi-region): Multiple instances of a single provider can be configured so resources can apply to different settings. As an example, this allows Terraform to manage multiple regions with AWS.
- Environmental variables to set variables: Environment variables can be
used to set variables. The environment variables must be in the format
TF_VAR_name
and this will be checked last for a value. - New remote state backend:
s3
: You can now store remote state in an S3 bucket. (#1723) - Automatic AWS retries: This release includes a lot of improvement around automatic retries of transient errors in AWS. The number of retry attempts is also configurable.
- Templates: A new
template_file
resource allows long strings needing variable interpolation to be moved into files. (#1778) - Provision with WinRM: Provisioners can now run remote commands on Windows hosts. (#1483)
IMPROVEMENTS:
- New config function:
length
- Get the length of a string or a list. Useful in conjunction withsplit
. (#1495) - New resource:
aws_app_cookie_stickiness_policy
- New resource:
aws_customer_gateway
- New resource:
aws_ebs_volume
- New resource:
aws_elasticache_cluster
- New resource:
aws_elasticache_security_group
- New resource:
aws_elasticache_subnet_group
- New resource:
aws_iam_access_key
- New resource:
aws_iam_group_policy
- New resource:
aws_iam_group
- New resource:
aws_iam_instance_profile
- New resource:
aws_iam_policy
- New resource:
aws_iam_role_policy
- New resource:
aws_iam_role
- New resource:
aws_iam_user_policy
- New resource:
aws_iam_user
- New resource:
aws_lb_cookie_stickiness_policy
- New resource:
aws_proxy_protocol_policy
- New resource:
aws_security_group_rule
- New resource:
aws_vpc_dhcp_options_association
- New resource:
aws_vpc_dhcp_options
- New resource:
aws_vpn_connection_route
- New resource:
google_dns_managed_zone
- New resource:
google_dns_record_set
- Migrate to upstream AWS SDK: Migrate the AWS provider to
awslabs/aws-sdk-go,
the official
awslabs
library. Previously we had forked the library for stability whileawslabs
refactored. Now that work has completed, and we've migrated back to the upstream version. - core: Improve error message on diff mismatch (#1501)
- provisioner/file: expand
~
in source path (#1569) - provider/aws: Better retry logic, now retries up to 11 times by default with exponentional backoff. This number is configurable. (#1787)
- provider/aws: Improved credential detection (#1470)
- provider/aws: Can specify a
token
via the config file (#1601) - provider/aws: Added new
vpc_security_group_ids
attribute for AWS Instances. If using a VPC, you can now modify the security groups for that Instance without destroying it (#1539) - provider/aws: White or blacklist account IDs that can be used to protect against accidents. (#1595)
- provider/aws: Add a subset of IAM resources (#939)
- provider/aws:
aws_autoscaling_group
retries deletes through "in progress" errors (#1840) - provider/aws:
aws_autoscaling_group
waits for healthy capacity during ASG creation (#1839) - provider/aws:
aws_instance
supports placement groups (#1358) - provider/aws:
aws_eip
supports network interface attachment (#1681) - provider/aws:
aws_elb
supports in-place changing of listeners (#1619) - provider/aws:
aws_elb
supports connection draining settings (#1502) - provider/aws:
aws_elb
increase default idle timeout to 60s (#1646) - provider/aws:
aws_key_pair
name can be omitted and generated (#1751) - provider/aws:
aws_network_acl
improved validation for network ACL ports and protocols (#1798) (#1808) - provider/aws:
aws_route_table
can target network interfaces (#968) - provider/aws:
aws_route_table
can specify propagating VGWs (#1516) - provider/aws:
aws_route53_record
supports weighted sets (#1578) - provider/aws:
aws_route53_zone
exports nameservers (#1525) - provider/aws:
aws_s3_bucket
website support (#1738) - provider/aws:
aws_security_group
name becomes optional and can be automatically set to a unique identifier; this helps withcreate_before_destroy
scenarios (#1632) - provider/aws:
aws_security_group
description becomes optional with a static default value (#1632) - provider/aws: automatically set the private IP as the SSH address if not specified and no public IP is available (#1623)
- provider/aws:
aws_elb
exportssource_security_group
field (#1708) - provider/aws:
aws_route53_record
supports alias targeting (#1775) - provider/aws: Remove default AWS egress rule for newly created Security Groups (#1765)
- provider/consul: add
scheme
configuration argument (#1838) - provider/docker:
docker_container
can specify links (#1564) - provider/google:
resource_compute_disk
supports snapshots (#1426) - provider/google:
resource_compute_instance
supports specifying the device name (#1426) - provider/openstack: Floating IP support for LBaaS (#1550)
- provider/openstack: Add AZ to
openstack_blockstorage_volume_v1
(#1726)
BUG FIXES:
- core: Fix graph cycle issues surrounding modules (#1582) (#1637)
- core: math on arbitrary variables works if first operand isn't a numeric primitive. (#1381)
- core: avoid unnecessary cycles by pruning tainted destroys from graph if there are no tainted resources (#1475)
- core: fix issue where destroy nodes weren't pruned in specific edge cases around matching prefixes, which could cause cycles (#1527)
- core: fix issue causing diff mismatch errors in certain scenarios during resource replacement (#1515)
- core: dependencies on resources with a different index work when count > 1 (#1540)
- core: don't panic if variable default type is invalid (#1344)
- core: fix perpetual diff issue for computed maps that are empty (#1607)
- core: validation added to check for
self
variables in modules (#1609) - core: fix edge case where validation didn't pick up unknown fields if the value was computed (#1507)
- core: Fix issue where values in sets on resources couldn't contain hyphens. (#1641)
- core: Outputs removed from the config are removed from the state (#1714)
- core: Validate against the worst-case graph during plan phase to catch cycles that would previously only show up during apply (#1655)
- core: Referencing invalid module output in module validates (#1448)
- command: remote states with uppercase types work (#1356)
- provider/aws: Support
AWS_SECURITY_TOKEN
env var again (#1785) - provider/aws: Don't save "instance" for EIP if association fails (#1776)
- provider/aws: launch configuration ID set after create success (#1518)
- provider/aws: Fixed an issue with creating ELBs without any tags (#1580)
- provider/aws: Fix issue in Security Groups with empty IPRanges (#1612)
- provider/aws: manually deleted S3 buckets are refreshed properly (#1574)
- provider/aws: only check for EIP allocation ID in VPC (#1555)
- provider/aws: raw protocol numbers work in
aws_network_acl
(#1435) - provider/aws: Block devices can be encrypted (#1718)
- provider/aws: ASG health check grace period can be updated in-place (#1682)
- provider/aws: ELB security groups can be updated in-place (#1662)
- provider/aws:
aws_main_route_table_association
can be deleted manually (#1806) - provider/docker: image can reference more complex image addresses, such as with private repos with ports (#1818)
- provider/openstack: region config is not required (#1441)
- provider/openstack:
enable_dhcp
for networking subnet should be bool (#1741) - provisioner/remote-exec: add random number to uploaded script path so that parallel provisions work (#1588)
- provisioner/remote-exec: chmod the script to 0755 properly (#1796)
0.4.2 (April 10, 2015)
BUG FIXES:
- core: refresh won't remove outputs from state file (#1369)
- core: clarify "unknown variable" error (#1480)
- core: properly merge parent provider configs when asking for input
- provider/aws: fix panic possibility if RDS DB name is empty (#1460)
- provider/aws: fix issue detecting credentials for some resources (#1470)
- provider/google: fix issue causing unresolvable diffs when using legacy
network
field ongoogle_compute_instance
(#1458)
0.4.1 (April 9, 2015)
IMPROVEMENTS:
- provider/aws: Route 53 records can now update
ttl
andrecords
attributes without destroying/creating the record (#1396) - provider/aws: Support changing additional attributes of RDS databases without forcing a new resource (#1382)
BUG FIXES:
- core: module paths in ".terraform" are consistent across different systems so copying your ".terraform" folder works. (#1418)
- core: don't validate providers too early when nested in a module (#1380)
- core: fix race condition in
count.index
interpolation (#1454) - core: properly initialize provisioners, fixing resource targeting during destroy (#1544)
- command/push: don't ask for input if terraform.tfvars is present
- command/remote-config: remove spurrious error "nil" when initializing remote state on a new configuration. (#1392)
- provider/aws: Fix issue with Route 53 and pre-existing Hosted Zones (#1415)
- provider/aws: Fix refresh issue in Route 53 hosted zone (#1384)
- provider/aws: Fix issue when changing map-public-ip in Subnets #1234
- provider/aws: Fix issue finding db subnets (#1377)
- provider/aws: Fix issues with
*_block_device
attributes on instances and launch configs creating unresolvable diffs when certain optional parameters were omitted from the config (#1445) - provider/aws: Fix issue with
aws_launch_configuration
causing an unnecessary diff for pre-0.4 environments (#1371) - provider/aws: Fix several related issues with
aws_launch_configuration
causing unresolvable diffs (#1444) - provider/aws: Fix issue preventing launch configurations from being valid in EC2 Classic (#1412)
- provider/aws: Fix issue in updating Route 53 records on refresh/read. (#1430)
- provider/docker: Don't ask for
cert_path
input on every run (#1432) - provider/google: Fix issue causing unresolvable diff on instances with
network_interface
(#1427)
0.4.0 (April 2, 2015)
BACKWARDS INCOMPATIBILITIES:
- Commands
terraform push
andterraform pull
are now nested under theremote
command:terraform remote push
andterraform remote pull
. The oldremote
functionality is now atterraform remote config
. This consolidates all remote state management under one command. - Period-prefixed configuration files are now ignored. This might break existing Terraform configurations if you had period-prefixed files.
- The
block_device
attribute ofaws_instance
has been removed in favor of three more specific attributes to specify block device mappings:root_block_device
,ebs_block_device
, andephemeral_block_device
. Configurations using the old attribute will generate a validation error indicating that they must be updated to use the new fields (#1045).
FEATURES:
- New provider:
dme
(DNSMadeEasy) (#855) - New provider:
docker
(Docker) - Manage container lifecycle using the standard Docker API. (#855) - New provider:
openstack
(OpenStack) - Interact with the many resources provided by OpenStack. (#924) - New feature:
terraform_remote_state
resource - Reference remote states from other Terraform runs to use Terraform outputs as inputs into another Terraform run. - New command:
taint
- Manually mark a resource as tainted, causing a destroy and recreate on the next plan/apply. - New resource:
aws_vpn_gateway
(#1137) - New resource:
aws_elastic_network_interfaces
(#1149) - Self-variables can be used to reference the current resource's
attributes within a provisioner. Ex.
${self.private_ip_address}
(#1033) - Continuous state saving during
terraform apply
. The state file is continuously updated as apply is running, meaning that the state is less likely to become corrupt in a catastrophic case: terraform panic or system killing Terraform. - Math operations in interpolations. You can now do things like
${count.index + 1}
. (#1068) - New AWS SDK: Move to
aws-sdk-go
(hashicorp/aws-sdk-go), a fork of the officialawslabs
repo. We forked for stability whileawslabs
refactored the library, and will move back to the officially supported version in the next release.
IMPROVEMENTS:
- New config function:
format
- Format a string usingsprintf
format. (#1096) - New config function:
replace
- Search and replace string values. Search can be a regular expression. See documentation for more info. (#1029) - New config function:
split
- Split a value based on a delimiter. This is useful for faking lists as parameters to modules. - New resource:
digitalocean_ssh_key
(#1074) - config: Expand
~
with homedir infile()
paths (#1338) - core: The serial of the state is only updated if there is an actual change. This will lower the amount of state changing on things like refresh.
- core: Autoload
terraform.tfvars.json
as well asterraform.tfvars
(#1030) - core:
.tf
files that start with a period are now ignored. (#1227) - command/remote-config: After enabling remote state, a
pull
is automatically done initially. - providers/google: Add
size
option to disk blocks for instances. (#1284) - providers/aws: Improve support for tagging resources.
- providers/aws: Add a short syntax for Route 53 Record names, e.g.
www
instead ofwww.example.com
. - providers/aws: Improve dependency violation error handling, when deleting Internet Gateways or Auto Scaling groups (#1325).
- provider/aws: Add non-destructive updates to AWS RDS. You can now upgrade
engine_version
,parameter_group_name
, andmulti_az
without forcing a new database to be created.(#1341) - providers/aws: Full support for block device mappings on instances and launch configurations [GH-1045, GH-1364]
- provisioners/remote-exec: SSH agent support. (#1208)
BUG FIXES:
- core: module outputs can be used as inputs to other modules (#822)
- core: Self-referencing splat variables are no longer allowed in provisioners. (#795)(#868)
- core: Validate that
depends_on
doesn't contain interpolations. (#1015) - core: Module inputs can be non-strings. (#819)
- core: Fix invalid plan that resulted in "diffs don't match" error when a computed attribute was used as part of a set parameter. (#1073)
- core: Fix edge case where state containing both "resource" and "resource.0" would ignore the latter completely. (#1086)
- core: Modules with a source of a relative file path moving up directories work properly, i.e. "../a" (#1232)
- providers/aws: manually deleted VPC removes it from the state
- providers/aws:
source_dest_check
regression fixed (now works). (#1020) - providers/aws: Longer wait times for DB instances.
- providers/aws: Longer wait times for route53 records (30 mins). (#1164)
- providers/aws: Fix support for TXT records in Route 53. (#1213)
- providers/aws: Fix support for wildcard records in Route 53. (#1222)
- providers/aws: Fix issue with ignoring the 'self' attribute of a Security Group rule. (#1223)
- providers/aws: Fix issue with
sql_mode
in RDS parameter group always causing an update. (#1225) - providers/aws: Fix dependency violation with subnets and security groups (#1252)
- providers/aws: Fix issue with refreshing
db_subnet_groups
causing an error instead of updating state (#1254) - providers/aws: Prevent empty string to be used as default
health_check_type
(#1052) - providers/aws: Add tags on AWS IG creation, not just on update (#1176)
- providers/digitalocean: Waits until droplet is ready to be destroyed (#1057)
- providers/digitalocean: More lenient about 404's while waiting (#1062)
- providers/digitalocean: FQDN for domain records in CNAME, MX, NS, etc. Also fixes invalid updates in plans. (#863)
- providers/google: Network data in state was not being stored. (#1095)
- providers/heroku: Fix panic when config vars block was empty. (#1211)
PLUGIN CHANGES:
- New
helper/schema
fields for resources:Deprecated
andRemoved
allow plugins to generate warning or error messages when a given attribute is used.
0.3.7 (February 19, 2015)
IMPROVEMENTS:
- New resources:
google_compute_forwarding_rule
,google_compute_http_health_check
, andgoogle_compute_target_pool
- Together these provide network-level load balancing. (#588) - New resource:
aws_main_route_table_association
- Manage the main routing table of a VPC. (#918) - New resource:
aws_vpc_peering_connection
(#963) - core: Formalized the syntax of interpolations and documented it very heavily.
- core: Strings in interpolations can now contain further interpolations,
e.g.:
foo ${bar("${baz}")}
. - provider/aws: Internet gateway supports tags (#720)
- provider/aws: Support the more standard environmental variable names for access key and secret keys. (#851)
- provider/aws: The
aws_db_instance
resource no longer requires bothfinal_snapshot_identifier
andskip_final_snapshot
; the presence or absence of the former now implies the latter. (#874) - provider/aws: Avoid unnecessary update of
aws_subnet
whenmap_public_ip_on_launch
is not specified in config. (#898) - provider/aws: Add
apply_method
toaws_db_parameter_group
(#897) - provider/aws: Add
storage_type
toaws_db_instance
(#896) - provider/aws: ELB can update listeners without requiring new. (#721)
- provider/aws: Security group support egress rules. (#856)
- provider/aws: Route table supports VPC peering connection on route. (#963)
- provider/aws: Add
root_block_device
toaws_db_instance
(#998) - provider/google: Remove "client secrets file", as it's no longer necessary for API authentication (#884).
- provider/google: Expose
self_link
ongoogle_compute_instance
(#906)
BUG FIXES:
- core: Fixing use of remote state with plan files. (#741)
- core: Fix a panic case when certain invalid types were used in the configuration. (#691)
- core: Escape characters
\"
,\n
, and\\
now work in interpolations. - core: Fix crash that could occur when there are exactly zero providers installed on a system. (#786)
- core: JSON TF configurations can configure provisioners. (#807)
- core: Sort
depends_on
in state to prevent unnecessary file changes. (#928) - core: State containing the zero value won't cause a diff with the lack of a value. (#952)
- core: If a set type becomes empty, the state will be properly updated to remove it. (#952)
- core: Bare "splat" variables are not allowed in provisioners. (#636)
- core: Invalid configuration keys to sub-resources are now errors. (#740)
- command/apply: Won't try to initialize modules in some cases when no arguments are given. (#780)
- command/apply: Fix regression where user variables weren't asked (#736)
- helper/hashcode: Update
hash.String()
to always return a positive index. Fixes issue where specific strings would convert to a negative index and be omitted when creating Route53 records. (#967) - provider/aws: Automatically suffix the Route53 zone name on record names. (#312)
- provider/aws: Instance should ignore root EBS devices. (#877)
- provider/aws: Fix
aws_db_instance
to not recreate each time. (#874) - provider/aws: ASG termination policies are synced with remote state. (#923)
- provider/aws: ASG launch configuration setting can now be updated in-place. (#904)
- provider/aws: No read error when subnet is manually deleted. (#889)
- provider/aws: Tags with empty values (empty string) are properly managed. (#968)
- provider/aws: Fix case where route table would delete its routes on an unrelated change. (#990)
- provider/google: Fix bug preventing instances with metadata from being created (#884).
PLUGIN CHANGES:
- New
helper/schema
type:TypeFloat
(#594) - New
helper/schema
field for resources:Exists
must point to a function to check for the existence of a resource. This is used to properly handle the case where the resource was manually deleted. (#766) - There is a semantic change in
GetOk
where it will returntrue
if there is any value in the diff that is non-zero. Before, it would return true only if there was a value in the diff.
0.3.6 (January 6, 2015)
FEATURES:
- New provider:
cloudstack
IMPROVEMENTS:
- New resource:
aws_key_pair
- Import a public key into AWS. (#695) - New resource:
heroku_cert
- Manage Heroku app certs. - provider/aws: Support
eu-central-1
,cn-north-1
, and GovCloud. (#525) - provider/aws:
route_table
can have tags. (#648) - provider/google: Support Ubuntu images. (#724)
- provider/google: Support for service accounts. (#725)
BUG FIXES:
- core: temporary/hidden files that look like Terraform configurations are no longer loaded. (#548)
- core: Set types in resources now result in deterministic states, resulting in cleaner plans. (#663)
- core: fix issue where "diff was not the same" would come up with diffing lists. (#661)
- core: fix crash where module inputs weren't strings, and add more validation around invalid types here. (#624)
- core: fix error when using a computed module output as an input to another module. (#659)
- core: map overrides in "terraform.tfvars" no longer result in a syntax error. (#647)
- core: Colon character works in interpolation (#700)
- provider/aws: Fix crash case when internet gateway is not attached to any VPC. (#664)
- provider/aws:
vpc_id
is no longer required. (#667) - provider/aws:
availability_zones
on ELB will contain more than one AZ if it is set as such. (#682) - provider/aws: More fields are marked as "computed" properly, resulting in more accurate diffs for AWS instances. (#712)
- provider/aws: Fix panic case by using the wrong type when setting volume size for AWS instances. (#712)
- provider/aws: route table ignores routes with 'EnableVgwRoutePropagation' origin since those come from gateways. (#722)
- provider/aws: Default network ACL ID and default security group ID
support for
aws_vpc
. (#704) - provider/aws: Tags are not marked as computed. This introduces another issue with not detecting external tags, but this will be fixed in the future. (#730)
0.3.5 (December 9, 2014)
FEATURES:
- Remote State: State files can now be stored remotely via HTTP, Consul, or HashiCorp's Atlas.
- New Provider:
atlas
: Retrieve artifacts for deployment from HashiCorp's Atlas service. - New
element()
function to index into arrays
IMPROVEMENTS:
- provider/aws: Support tenancy for aws_instance
- provider/aws: Support block devices for aws_instance
- provider/aws: Support virtual_name on block device
- provider/aws: Improve RDS reliability (more grace time)
- provider/aws: Added aws_db_parameter_group resource
- provider/aws: Added tag support to aws_subnet
- provider/aws: Routes in RouteTable are optional
- provider/aws: associate_public_ip_address on aws_launch_configuration
- provider/aws: Added aws_network_acl
- provider/aws: Ingress rules in security groups are optional
- provider/aws: Support termination policy for ASG
- provider/digitalocean: Improved droplet size compatibility
BUG FIXES:
- core: Fixed issue causing double delete. (#555)
- core: Fixed issue with create-before-destroy not being respected in some circumstances.
- core: Fixing issue with count expansion with non-homogenous instance plans.
- core: Fix issue with referencing resource variables from resources that don't exist yet within resources that do exist, or modules.
- core: Fixing depedency handling for modules
- core: Fixing output handling (#474)
- core: Fixing count interpolation in modules
- core: Fixing multi-var without module state
- core: Fixing HCL variable declaration
- core: Fixing resource interpolation for without state
- core: Fixing handling of computed maps
- command/init: Fixing recursion issue (#518)
- command: Validate config before requesting input (#602)
- build: Fixing GOPATHs with spaces
MISC:
- provider/aws: Upgraded to helper.Schema
- provider/heroku: Upgraded to helper.Schema
- provider/mailgun: Upgraded to helper.Schema
- provider/dnsimple: Upgraded to helper.Schema
- provider/cloudflare: Upgraded to helper.Schema
- provider/digitalocean: Upgraded to helper.Schema
- provider/google: Upgraded to helper.Schema
0.3.1 (October 21, 2014)
IMPROVEMENTS:
- providers/aws: Support tags for security groups.
- providers/google: Add "external_address" to network attributes (#454)
- providers/google: External address is used as default connection host. (#454)
- providers/heroku: Support
locked
andpersonal
booleans on organization settings. (#406)
BUG FIXES:
- core: Remove panic case when applying with a plan that generates no new state. (#403)
- core: Fix a hang that can occur with enough resources. (#410)
- core: Config validation will not error if the field is being computed so the value is still unknown.
- core: If a resource fails to create and has provisioners, it is marked as tainted. (#434)
- core: Set types are validated to be sets. (#413)
- core: String types are validated properly. (#460)
- core: Fix crash case when destroying with tainted resources. (#412)
- core: Don't execute provisioners in some cases on destroy.
- core: Inherited provider configurations will be properly interpolated. (#418)
- core: Refresh works properly if there are outputs that depend on resources that aren't yet created. (#483)
- providers/aws: Refresh of launch configs and autoscale groups load the correct data and don't incorrectly recreate themselves. (#425)
- providers/aws: Fix case where ELB would incorrectly plan to modify listeners (with the same data) in some cases.
- providers/aws: Retry destroying internet gateway for some amount of time if there is a dependency violation since it is probably just eventual consistency (public facing resources being destroyed). (#447)
- providers/aws: Retry deleting security groups for some amount of time if there is a dependency violation since it is probably just eventual consistency. (#436)
- providers/aws: Retry deleting subnet for some amount of time if there is a dependency violation since probably asynchronous destroy events take place still. (#449)
- providers/aws: Drain autoscale groups before deleting. (#435)
- providers/aws: Fix crash case if launch config is manually deleted. (#421)
- providers/aws: Disassociate EIP before destroying.
- providers/aws: ELB treats subnets as a set.
- providers/aws: Fix case where in a destroy/create tags weren't reapplied. (#464)
- providers/aws: Fix incorrect/erroneous apply cases around security group rules. (#457)
- providers/consul: Fix regression where
key
param changed to `keys. (#475)
0.3.0 (October 14, 2014)
FEATURES:
- Modules: Configuration can now be modularized. Modules can live on GitHub, BitBucket, Git/Hg repos, HTTP URLs, and file paths. Terraform automatically downloads/updates modules for you on request.
- New Command:
init
. This command initializes a Terraform configuration from an existing Terraform module (also new in 0.3). - New Command:
destroy
. This command destroys infrastructure created withapply
. - Terraform will ask for user input to fill in required variables and provider configurations if they aren't set.
terraform apply MODULE
can be used as a shorthand to quickly build infrastructure from a module.- The state file format is now JSON rather than binary. This allows for easier machine and human read/write. Old binary state files will be automatically upgraded.
- You can now specify
create_before_destroy
as an option for replacement so that new resources are created before the old ones are destroyed. - The
count
metaparameter can now contain interpolations (such as variables). - The current index for a resource with a
count
set can be interpolated using${count.index}
. - Various paths can be interpolated with the
path.X
variables. For example, the path to the current module can be interpolated using${path.module}
.
IMPROVEMENTS:
- config: Trailing commas are now allowed for the final elements of lists.
- core: Plugins are loaded from
~/.terraform.d/plugins
(Unix) or%USERDATA%/terraform.d/plugins
(Windows). - command/show: With no arguments, it will show the default state. (#349)
- helper/schema: Can now have default values. (#245)
- providers/aws: Tag support for most resources.
- providers/aws: New resource
db_subnet_group
. (#295) - providers/aws: Add
map_public_ip_on_launch
for subnets. (#285) - providers/aws: Add
iam_instance_profile
for instances. (#319) - providers/aws: Add
internal
option for ELBs. (#303) - providers/aws: Add
ssl_certificate_id
for ELB listeners. (#350) - providers/aws: Add
self
option for security groups for ingress rules with self as source. (#303) - providers/aws: Add
iam_instance_profile
option toaws_launch_configuration
. (#371) - providers/aws: Non-destructive update of
desired_capacity
for autoscale groups. - providers/aws: Add
main_route_table_id
attribute to VPCs. (#193) - providers/consul: Support tokens. (#396)
- providers/google: Support
target_tags
for firewalls. (#324) - providers/google:
google_compute_instance
supportscan_ip_forward
(#375) - providers/google:
google_compute_disk
supportstype
to support disks such as SSDs. (#351) - provisioners/local-exec: Output from command is shown in CLI output. (#311)
- provisioners/remote-exec: Output from command is shown in CLI output. (#311)
BUG FIXES:
- core: Providers are validated even without a
provider
block. (#284) - core: In the case of error, walk all non-dependent trees.
- core: Plugin loading from CWD works properly.
- core: Fix many edge cases surrounding the
count
meta-parameter. - core: Strings in the configuration can escape double-quotes with the
standard
\"
syntax. - core: Error parsing CLI config will show properly. (#288)
- core: More than one Ctrl-C will exit immediately.
- providers/aws: autoscaling_group can be launched into a vpc (#259)
- providers/aws: not an error when RDS instance is deleted manually. (#307)
- providers/aws: Retry deleting subnet for some time while AWS eventually destroys dependencies. (#357)
- providers/aws: More robust destroy for route53 records. (#342)
- providers/aws: ELB generates much more correct plans without extranneous data.
- providers/aws: ELB works properly with dynamically changing count of instances.
- providers/aws: Terraform can handle ELBs deleted manually. (#304)
- providers/aws: Report errors properly if RDS fails to delete. (#310)
- providers/aws: Wait for launch configuration to exist after creation (AWS eventual consistency) (#302)
0.2.2 (September 9, 2014)
IMPROVEMENTS:
- providers/amazon: Add
ebs_optimized
flag. (#260) - providers/digitalocean: Handle 404 on delete
- providers/digitalocean: Add
user_data
argument for creating droplets - providers/google: Disks can be marked
auto_delete
. (#254)
BUG FIXES:
- core: Fix certain syntax of configuration that could cause hang. (#261)
- core:
-no-color
flag properly disables color. (#250) - core: "~" is expanded in
-var-file
flags. (#273) - core: Errors with tfvars are shown in console. (#269)
- core: Interpolation function calls with more than two args parse. (#282)
- providers/aws: Refreshing EIP from pre-0.2 state file won't error. (#258)
- providers/aws: Creating EIP without an instance/network won't fail.
- providers/aws: Refreshing EIP manually deleted works.
- providers/aws: Retry EIP delete to allow AWS eventual consistency to detect it isn't attached. (#276)
- providers/digitalocean: Handle situations when resource was destroyed manually. (#279)
- providers/digitalocean: Fix a couple scenarios where the diff was incorrect (and therefore the execution as well).
- providers/google: Attaching a disk source (not an image) works properly. (#254)
0.2.1 (August 31, 2014)
IMPROVEMENTS:
- core: Plugins are automatically discovered in the executable directory or pwd if named properly. (#190)
- providers/mailgun: domain records are now saved to state
BUG FIXES:
- core: Configuration parses when identifier and '=' have no space. (#243)
- core:
depends_on
withcount
generates the proper graph. (#244) - core: Depending on a computed variable of a list type generates a
plan without failure. i.e.
${type.name.foos.0.bar}
wherefoos
is computed. (#247) - providers/aws: Route53 destroys in parallel work properly. (#183)
0.2.0 (August 28, 2014)
BACKWARDS INCOMPATIBILITIES:
- We've replaced the configuration language in use from a C library to
a pure-Go reimplementation. In the process, we removed some features
of the language since it was too flexible:
- Semicolons are no longer valid at the end of lines
- Keys cannot be double-quoted strings:
"foo" = "bar"
is no longer valid. - JSON style maps
{ "foo": "bar" }
are no longer valid outside of JSON. Maps must be in the format of{ foo = "bar" }
(like other objects in the config)
- Heroku apps now require (will not validate without)
region
andname
due to an upstream API change. (#239)
FEATURES:
- New Provider:
google
: Manage Google Compute instances, disks, firewalls, and more. - New Provider:
mailgun
: Manage mailgun domains. - New Function:
concat
: Concatenate multiple strings together. Example:concat(var.region, "-", var.channel)
.
IMPROVEMENTS:
- core: "~/.terraformrc" (Unix) or "%APPDATA%/terraform.rc" (Windows) can be used to configure custom providers and provisioners. (#192)
- providers/aws: EIPs now expose
allocation_id
andpublic_ip
attributes. - providers/aws: Security group rules can be updated without a destroy/create.
- providers/aws: You can enable and disable dns settings for VPCs. (#172)
- providers/aws: Can specify a private IP address for
aws_instance
(#217)
BUG FIXES:
- core: Variables are validated to not contain interpolations. (#180)
- core: Key files for provisioning can now contain
~
and will be expanded to the user's home directory. (#179) - core: The
file()
function can load files in sub-directories. (#213) - core: Fix issue where some JSON structures didn't map properly into Terraform structures. (#177)
- core: Resources with only
file()
calls will interpolate. (#159) - core: Variables work in block names. (#234)
- core: Plugins are searched for in the same directory as the executable before the PATH. (#157)
- command/apply: "tfvars" file no longer interferes with plan apply. (#153)
- providers/aws: Fix issues around failing to read EIPs. (#122)
- providers/aws: Autoscaling groups now register and export load balancers. (#207)
- providers/aws: Ingress results are treated as a set, so order doesn't matter anymore. (#87)
- providers/aws: Instance security groups treated as a set (#194)
- providers/aws: Retry Route53 requests if operation failed because another operation is in progress (#183)
- providers/aws: Route53 records with multiple record values work. (#221)
- providers/aws: Changing AMI doesn't result in errors anymore. (#196)
- providers/heroku: If you delete the
config_vars
block, config vars are properly nuked. - providers/heroku: Domains and drains are deleted before the app.
- providers/heroku: Moved from the client library bgentry/heroku-go to cyberdelia/heroku-go (#239).
- providers/heroku: Plans without a specific plan name for heroku_addon work. (#198)
PLUGIN CHANGES:
- New Package:
helper/schema
. This introduces a high-level framework for easily writing new providers and resources. The Heroku provider has been converted to this as an example.
0.1.1 (August 5, 2014)
FEATURES:
- providers/heroku: Now supports creating Heroku Drains (#97)
IMPROVEMENTS:
- providers/aws: Launch configurations accept user data (#94)
- providers/aws: Regions are now validated (#96)
- providers/aws: ELB now supports health check configurations (#109)
BUG FIXES:
- core: Default variable file "terraform.tfvars" is auto-loaded. (#59)
- core: Multi-variables (
foo.*.bar
) work even whencount = 1
. (#115) - core:
file()
function can have string literal arg (#145) - providers/cloudflare: Include the proper bins so the cloudflare provider is compiled
- providers/aws: Engine version for RDS now properly set (#118)
- providers/aws: Security groups now depend on each other and
- providers/aws: DB instances now wait for destroys, have proper dependencies and allow passing skip_final_snapshot
- providers/aws: Add associate_public_ip_address as an attribute on the aws_instance resource (#85)
- providers/aws: Fix cidr blocks being updated [GH-65, GH-85]
- providers/aws: Description is now required for security groups
- providers/digitalocean: Private IP addresses are now a separate attribute
- provisioner/all: If an SSH key is given with a password, a better error message is shown. (#73)
0.1.0 (July 28, 2014)
- Initial release