Changelog

Unreleased

Improvements

A set of new microservice_node... base images added, including:
- microservice_node16 (based on microservice).
- microservice_node14_focal (based on microservice_focal).
- microservice_node16_focal (based on microservice_focal).

2.11.3 (2021-03-10)

We do best effort to support docker versions 1.12.0 - 20.10.4 with this release.

Bug fixes

Fixed armada ssh command for docker >= 19.03.14.

2.11.2 (2021-03-02)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Bug fixes

Fixed armada diagnose command for microservice_focal images on older armada versions by adding alias python --> python3 in those images.

2.11.1 (2021-02-11)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Bug fixes

Fix armada installation for docker >= 18.09.0.
Fix pip locale problems during armada installation.

2.11.0 (2021-02-02)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Improvements

A new microservice_focal (ubuntu20.04) base image added. It's version bumped to 2.11.0.
A new microservice_python3_focal base image added (based on microservice_focal).
Armada built on top of the microservice_python3_focal base image. It's version bumped to 2.11.0.
Armada commands migrated to python3. Python2 dependency dropped.
Haproxy version upgraded from 1.8 to 2.2.
Dropping support for Amazon Linux 1.
RPM package of Armada works fine with Amazon Linux 2.

Bug fixes

Fix broken dependency (Falcon Json Middleware) for backend.
Allow binding armada to interfaces with more than one IP address.

2.10.0 (2019-01-31)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Improvements

Python3 version in base image microservice_python3 has been upgraded from python3.6 to python3.7.
Don't build deprecated microservice_python3.5.
Armada internal API is now running on python3.7.
By default don't show logs from garbage collector in microservice_node*

2.9.1 (2018-11-02)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Bug fixes

Fix microservice image not starting.

2.9.0 (2018-10-31)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Features

Add nodejs v10.x, remove unsupported v4.x version.

Bug fixes

Fix generating CONFIG_PATH on renamed microservices.

2.8.1 (2018-05-09)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Bug fixes

-Fix stopping/recovering services saved to KV-store using ship IP instead of name.

2.8.0 (2018-04-24)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Features

Added new microservice_flask base image
Added flask template

2.7.1 (2018-03-29)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Bug fixes

Fix pushing images to remote dockyards with passwords.

2.7.0 (2018-03-21)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Features

armada build supports multi-staged Dockerfiles.
Base images in armada build may, and should now use explicit dockyard/docker registry addresses, e.g.: FROM docker.io/golang:1.7.3, or FROM dockyard.armada.sh/microservice.

Improvements

Set TimeoutStartSec=30min in systemd unit to prevent timeouts
Decreased number of Consul queries executed during saving containers in KV store
Save missing containers in KV store
If initial recovery is not completed after 10 minutes since armada start, armada has status warning

Bug fixes

Fix netcat version on ArchLinux
Fix filtering crashed services by name

2.6.0 (2018-03-09)

We do best effort to support docker versions 1.12.0 - 17.12.1 with this release.

Features

Build tar.xz (pacman) linux package.

Bug fixes

Fix stopping armada service, caused ship lefts cluster
Fix getting docker version in docker > 17.10

2.5.1 (2018-02-12)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Improvements

Increase uwsgi timeout (aka harakiri) from 11s to 31s.
Set uwsgi socket-timeout to 31s.
Use 3 threads per uwsgi process instead of 2.
Don't run http worker as separate process in uwsgi.
Microservice image is now building from deb package.
More verbose logs.

Bug fixes

Split unit tests for command (python 2) and backend (python 3).
Fix bug with updating microservice_version in armada list for microservices with subservices.

2.5.0 (2018-02-05)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Features

New API endpoint /v1/local/ports/{microservice_id} for microservices, available inside the container.
- It returns the mapping of local ports in container to external ports on host. E.g:
```
$ curl 172.17.0.1:8900/v1/local/ports/f30690a0f8af
{"80/tcp": "4261/tcp"}
```
New flag in armada list: --microservice-version. It shows the version of microservice package inside the container. Works only for microservices based on version >= 2.5.0.

Improvements

Armada internal API is now running on python3 + uwsgi + falcon.
New internal API for microservices (registering service, reporting health checks).
- It should not change behavior. This is only a step towards removing docker and consul dependencies in microservices.
Upgrade requests pip package in microservice base image from 2.9.1 to 2.18.4.

2.4.3 (2018-01-11)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Improvements

Bump python3 to 3.6.4 in microservice_python3
Python scripts in microservice base image are now a proper python package, and provide command microservice.
- This is the new suggested way of calling microservice scripts:
  - microservice require instead of python require_service.py
  - microservice register instead of python register_in_service_discovery.py
Upgrade haproxy in microservice base image from 1.7 to 1.8.
Remove unused address_adapter in microservice base image.

Bug fixes

Fix setting empty flags in require_service.py/microservice require, i.e. --app_id="" and --env="".

2.4.2 (2018-01-02)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Improvements

local-magellan, in microservice base image, besides of configuring haproxy dumps mapping of local port -> IP:port to /var/opt/service_to_addresses.json. E.g.

root@c1823ea048c5:/# cat /var/opt/service_to_addresses.json
{
    "2000": [
        "192.168.3.19:32822", 
        "192.168.3.19:32821"
    ]
}

2.4.1 (2017-12-01)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Improvements

Improve sorting in armada list for services with subservices

2.4.0 (2017-11-17)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Features

Add useful bash functions atail, acd, actl - use ahelp for more informations.

2.3.0 (2017-11-03)

We do best effort to support docker versions 1.12.0 - 17.10.0 with this release.

Features

Add IMAGE_NAME environmental variable.
Compatibility with new docker binary location for versions 17.03 - 17.10. It makes remote armada ssh/diagnose work again in those docker versions.
Set default value of CONFIG_DIR to 'config', in microservice base image. Hermes will automatically search for configs in 'config' directory in your service.
New command alogs available in microservice base image. It opens up midnight commander with supervisor logs on the left pane and service source on the right pane.
Add bash completion for supervisorctl
Add possibility to configure service requirements via hermes config. This allows us to require different services in different environments. In base microservice image required_service.py will try to find service_discovery.json in config path. More info in docs

Improvements

Get rid of ugly WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! warnings showing up in remote armada ssh/diagnose invocations, after restarting/upgrading armada.
Get rid of ugly bash: warning: setlocale: LC_ALL: cannot change locale (en_US.UTF-8) warning in remote armada ssh/diagnose invocations.
Clean up code responsible for support of services built on microservice image older than v0.11.0 (2015-12-07) in armada diagnose.

Bug fixes

Fix build-armada and restart-armada development scripts to work outside of armada development vagrant as well.
Fix armada's health-check for validating if armada runtime settings have been restored, to work during development of armada.

2.2.0 (2017-10-19)

We do best effort to support docker versions 1.12.0 - 17.05.0 with this release.

Features

Add customization of max_old_space_size in base nodejs microservices by setting MAX_OLD_SPACE_SIZE_MB environmental variable. By default MAX_OLD_SPACE_SIZE_MB=256
Add vim to microservice
Set build-time variables using --build-arg flag.
Build base nodejs microservices with currently supported nodejs version - v4, v6, v8 and v0.12.

Improvements

armada develop is now compatible with official armada-vagrant boxes.
armada stop now supports stopping multiple services
armada list --quiet returns only distinct container IDs
apache2 is running in the foreground in microservice_php, adding logrotate for log retention management.
Use template for building base nodejs images.

Bug fixes

armada stop can remove remaining services from removed ship

2.1.2 (2017-09-20)

We do best effort to support docker versions 1.12.0 - 17.05.0 with this release.

Bug fixes

HashiCorp Consul downgraded to 0.7.5 version due to performance problems.

2.1.1 (2017-09-13)

We do best effort to support docker versions 1.12.0 - 17.05.0 with this release.

Improvements

armada-runner script now supports ip command instead ifconfig and route.
armada-runner can easliy bind local network interface from now.
armada-runner network interface autodetection when $default_interface in /etc/default/armada is empty or commented.
armada-runner network interface can be set statically in /etc/default/armada and would not be changed even if not exists.
Removed net-tools from dependecies of installation packages.
HashiCorp Consul upgraded to 0.9.3 version.

2.1.0 (2017-07-25)

We do best effort to support docker versions 1.12.0 - 17.05.0 with this release.

Features

EXPERIMENTAL: armada-heal script, for healing overloaded armada ships, is available inside of the armada container. Use with caution. Details: https://github.com/armadaplatform/armada-heal

Improvements

Exponentially increasing periods between recover retries.
Internal refactoring of storing services data in consul.
Use only docker to build armada packages

2.0.0 (2017-07-10)

We do best effort to support docker versions 1.12.0 - 17.05.0 with this release.

Breaking changes

Dropped support for Docker <1.12.0

Features

Add hooks. Currently only pre-stop hook is available. Place your custom scripts under ./hooks/pre-stop/ directory to have them executed, when certain action occurs.
New command armada develop. It sets up the environment for development of given service, i.e. MICROSERVICE_NAME, so that service's name is implied in other armada commands. By default it mounts current working directory to container, and assigns sticky port from range 4000..4999, based on hash of service name, that should be easy to memorize. Exporting $MICROSERVICE_NAME env variable is no longer recommended for development.

Improvements

Bump docker-py to 2.4.2

Bug fixes

armada list --local don't choose active instance among services registered with --single-active-instance flag. This is to prevent unnecessary changes of active one between instances running on different ships.

1.17.0 (2017-06-20)

We do best effort to support docker versions 1.6.0 - 17.05.0 with this release.

Features

Compatibility with docker versions 1.13.0 - 17.05.0.

Improvements

armada build --squash uses --squash flag built into Docker (since 1.13), which is much faster and merges only the layers from currently built Dockerfile. To use it you need to turn on experimental flag in your Docker daemon: https://sreeninet.wordpress.com/2017/01/27/docker-1-13-experimental-features/

1.16.2 (2017-05-29)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fix armada list when one of ships has version < 1.6.0

1.16.1 (2017-05-24)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fix armada list when one of ships was renamed.

1.16.0 (2017-05-18)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Add DISTRIB_CODENAME=xenial and DISTRIB_RELEASE=16.04 environmental variables.
Add less and sudo to microservice
You can now set retention for containers backups stored in /opt/armada/saved_containers_backup/. To do that, set SAVED_CONTAINERS_BACKUP_RETENTION=N variable (where N is integer with number of days stored) in /etc/default/armada on host.

Improvements

Don't install old gcc version in microservice_python3.
Regular cleaning duplicated containers backup in /opt/armada/saved_containers_backup/.
Optimize armada list.

Bug fixes

Fix autoreloading armada development version.
Fix clearing sessions files in PHP Docker container.

1.15.1 (2017-05-12)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fix node version in microservice_node image based on Ubuntu 16.04.
Add link to php config at /etc/php5 in microservice_php for compatibility with pre Ubuntu 16.04 images.

1.15.0 (2017-05-08)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Show "Env" and "AppID" as separate columns, instead of "Tags", in armada list.

Improvements

Cron clear sessions files in PHP Docker container.
Use Ubuntu 16.04 instead of 14.04 as base in microservice image.
Upgrade HAProxy from 1.6 to 1.7.

1.14.0 (2017-04-03)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Improvements

Update consul from v0.6.4 to v0.7.5.

1.13.0 (2017-03-22)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

[EXPERIMENTAL] Stub of new command armada deploy. Currently this command is an alias for armada restart -a, however it accepts all arguments accepted by armada run command, as well as desired number of instances for given microservice. Eventually this command will be responsible for deploying new version of microservice: if it's already running then it will perform restart, otherwise it'll launch as many instances as needed to satisfy requirements.

1.12.2 (2017-03-01)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fix running add-apt-repository in microservice_python3.
Fix building armada development version.

Improvements

Log exceptions in consul query.

1.12.1 (2017-02-21)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Improvements

Log info about removed kv entries in armada cleaner.

1.12.0 (2017-02-06)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Add flag --single-active-instance to register_in_service_discovery.py script. Services registered with this flag will have at most one available (status 'passing' or 'warning') instance in armada service discovery mechanisms. The other instances will receive status 'standby'.
Add -vv/--verbose flag to all armada commands.
armada shutdown now removes by default previously joined ships from /opt/armada/runtime_settings.json. Use --keep-joined flag for old behavior.

Bug fixes

Fix building pip packages in microservice_python3.
Fixed cleaning crashed services on not promoted ships.
Fix stop and restart of armada service on systemd.

1.11.0 (2017-01-13)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Add -vv/--verbose flag to armada create.
Base image for services created using DotNET Core 1.0
Template for sample "Hello, world!" REST service.
Base image microservice_python3.5 is now deprecated in favor of upgraded microservice_python3 with python3.6.

Improvements

Separated supervisor configs for armada_agent and register_in_service_discovery to make overriding the latter in services easier.
Python3 version in base image microservice_python3 has been upgraded from python3.4 to python3.6.

Bug fixes

Fixed bug where restarting service started without env was setting MICROSERVICE_ENV environment variable as env.

1.10.0 (2016-12-15)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

You can now point path to Dockerfile in armada build using --file flag (like docker build).

1.9.2 (2016-12-12)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fixed bug where not recovered containers stayed with recovering status.

1.9.1 (2016-11-29)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fixed restarting crashed services.

1.9.0 (2016-11-17)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Build deb, rpm and amazon linux package.
Privilege mode of docker containers can be specified in /etc/default/armada.

Bug fixes

Fix keeping stopped services in crashed state after armada restart.

1.8.2 (2016-11-14)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fixed stopping services in started state.

Improvements

Send request data to sentry, add tag ship_IP to events.

1.8.1 (2016-11-03)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fix getting services list when microservice_env is not set

Improvements

Disable timeout in systemd service

1.8.0 (2016-10-31)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

Log errors to sentry. It can be enable by adding sentry_url={url of project in sentry} in /etc/default/armada config file.

Bug fixes

Armada builds from base image with proper tag.

1.7.1 (2016-10-24)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Bug fixes

Fixed bug in /list endpoint in clusters with old backups in K/V store

1.7.0 (2016-10-18)

We do best effort to support docker versions 1.6.0 - 1.12.1 with this release.

Features

New base image microservice_node6 with node v6.

Improvements

Upgrade go in microservice_go base image from 1.5.3 to 1.6.3.
Optimized /list endpoint
Removed Consul watch checks
Consul watch keyprefix narrowed down to local services
Set locales to UTF-8 version

Bug fixes

Generating unique ID for services while recovering
Fixed autoreloading armada backed in vagrant environment
Fixed detecting stop of armada service in systemd
Fixed recovering containers
Removing services from Consul K/V store while armada shutdown

1.6.0 (2016-10-05)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Features

Storing service restart parameters and status in K/V store.
armada recover by default will try recover crashed services from K/V store

Improvements

Ship names survive joining armada
Ship names must be unique
armada diagnose properly prints last health-check for services based on old microservice image, along with deprecation warning.

1.5.3 (2016-09-20)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Ensure services based on old microservice image do not register themselves back in Consul during armada stop.

1.5.2 (2016-09-09)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fixed an issue with armada ssh ignoring --local flag.
Fixed KeyError exception issue when connecting to remote service.

1.5.1 (2016-09-07)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Improvements

--env parameter armada run by default take MICROSERVICE_ENV environment variable as value.

Bug fixes

Fixed restarting services which have been run using MICROSERVICE_NAME environment variable.

1.5.0 (2016-09-06)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Features

Added flag -s/--squash to armada build. With this option armada try to minimize service image via docker-squash.

Bug fixes

Fixed an issue with armada ssh not accepting container ids.

Improvements

Merge run_health_checks.py and register_in_service_discovery.py into armada_agent.py in order to save resources, register_is_service_discovery.py is left for backward compatibility.
Changed timeout in armada stop from 10s to 60s until SIGKILL is sent to container main process.

1.4.0 (2016-09-02)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Features

Added service selection prompt to armada ssh. Whenever multiple matching services are found, user can easily select which instance they want to ssh into.
Added --no-prompt flag to armada ssh which disables prompting mechanism and results in error if multiple matching services are found.
Added -l/--local flag to armada ssh which limits matched services to local machine.

Improvements

Vagrant Use socat instead of armada-bind to proxy insecure dockyard connection through localhost.

Bug fixes

Properly deregister crashed sub-services.
Fixed a bug which caused armada run in development environment use 4999:80 port mapping, even if port 80 was explicitly overridden.

1.3.3 (2016-08-25)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fixed armada service on systemd init system.

1.3.2 (2016-08-23)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fixed critical bug with setting containers' config paths introduced in 1.3.0.

1.3.1 (2016-08-19)

Bug fixes

Fixed critical bug with mounting containers' volumes introduced in 1.3.0.

1.3.0 (2016-08-19)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Improvements

Optionally restrict possible config directories. By default root directory is mounted inside armada container in read-only mode, to verify existence of --config paths. This can be restricted by setting RESTRICT_CUSTOM_CONFIG_DIRS variable to a specific path. (e.g. /home/user/configs)
run and restart commands can get IP address as '--ship' argument.

Features

armada list show services that should be running with additional statuses:
- recovering
- not-recovered if recovering was not successful
- crashed for not running containers
- armada stop to remove service from list

Bug fixes

While default armada network interface is unavailable armada uses host default interface.
Fixed logging of Armada CLI commands

1.2.2 (2016-08-04)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fix armada list always assuming -l/--local flag.
Revert old behavior in armada list that showed 'container_id' instead of 'microservice_id' in ID column.

1.2.1 (2016-08-04)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fix filtering by app_id in armada list.
Fix pyflakes issues.
Fix installation of bash completion file.

1.2.0 (2016-08-03)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Features

Added autocomplete of armada commands, dockyard aliases, services and ships names.

Improvements

'armada list' command support wildcards *, ?.

Bug fixes

Remove HTTP_PROXY header.

1.1.1 (2016-07-19)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Bug fixes

Fixed issue with unhandled exception during the installation process.

1.1.0 (2016-07-19)

We do best effort to support docker versions 1.6.0 - 1.12.0 with this release.

Features

Added logging of Armada CLI
Added support for docker 1.12.*

Improvements

After critical status of service health checks next check is run after 1s,2s,3s... until 10s or first pass of health checks.
Enabled autoreload of Armada API in development Vagrant

1.0.0 (2016-07-08)

Ahoy sailor!
🎉 🎂 Armada is celebrating one year anniversary! 🎂 🎉
Fair winds!
⚓

We do best effort to support docker versions 1.6.0 - 1.10.3 with this release.

Features

🎈 Added indicator of current ship in 'armada info'.

Bug fixes

🎈 Fix log file permission error.
🎈 Fix getting docker pid on centos7.
🎈 Validate response from armada_api.get('version').

Improvements

🎈 Install armada pip package (with hermes) in microservice_python* base images.

0.20.2 (2016-07-04)

We do best effort to support docker versions 1.6.0 - 1.10.3 with this release.

Bug fixes

Do not use local restart if --ship parameter is provided.
Fix issue with fallback to the default dockyard alias if the -d options is not provided.

0.20.1 (2016-06-28)

We do best effort to support docker versions 1.6.0 - 1.10.3 with this release.

Bug fixes

Fix displaying the information about the new version.

0.20.0 (2016-06-28)

We do best effort to support docker versions 1.6.0 - 1.10.3 with this release.

⚠️ UPGRADE WARNING ⚠️

Due to internal changes in Armada, all Armada agents within the same cluster need to be upgraded, otherwise remote runs and restarts will not work, and armada info may contain incorrect names. Ships' names will have to be set once again using armada name command.

Features

Allow verbose output for armada build command.
Added images versioning. Since now it's possible to build, push and run services using service_name:tag notation or :tag notation if the MICROSERVICE_NAME env is provided.
Added warning about using outdated armada version. It can be disabled by adding the check_updates=0 line to the /etc/default/armada config file.

Improvements

Upgrade Consul from version 0.6.3 to 0.6.4.
Store names of Armada ships in Consul's kv store. armada name does not restart Armada service anymore and is much faster.
After joining Armada agent to cluster, Armada will detect running Couriers there and fetch configuration from them.

Bug fixes

Fix build of microservice_python3.5 base image.

0.19.2 (2016-05-31)

Improvements

Retry after failure during stopping of Docker container.

Bug fixes

Fix bug with recovering multiple copies of the same service.
Fix detection of Docker daemon start time, which could cause unnecessary recovery.
Make sure python requests package is not updated in Armada service.
Fix race condition in require_service.py, causing not registering service in local magellan.

0.19.1 (2016-05-25)

Improvements

Update haproxy in microservice image (used by local magellan), from 1.5 to 1.6.
Increase timeout to docker api from 7s to 11s.

Bug fixes

Hold version of python requests package to 2.9.1 in microservice image.
Fixed checking for newest local image of service on restart in vagrant development environment.
Fix import errors in couple of backend scripts.

0.19.0 (2016-05-17)

Improvements

In vagrant for armada development (Vagrantfile in armada repository), the code used for armada CLI is now mounted from local workstation.
Moved armada installation script from armada-website repository to armada repository.
Added OpenRC support (Contributed by ryneeverett).

Bug fixes

Fixed issue with passing arguments to armada ssh.

0.18.0 (2016-05-09)

Features

Services can now be moved between ships in a cluster with armada restart command using --ship parameter.
E.g.: armada restart myservice --ship 10.0.0.2
If moved service contains mounted volumes or statically assigned ports, restart will fail unless -f \ --force flag is provided.
⚠️ Use --force with caution. Mounted volumes will not be moved to another ship. Static ports collision will result in service being stopped, and not restarted.

Improvements

Nicified armada restart warning/error output.

0.17.0 (2016-04-29)

⚠️ UPGRADE WARNING ⚠️

Due to changes in armada internal API, all armada agents within the same cluster need to be upgraded, otherwise remote runs and restarts will not work.

Features

Local haproxy binds both on IPv4 and IPv6.
Default memory limit (both resident and swap) for services can now be customized per host. E.g. Add line DEFAULT_CONTAINER_MEMORY_LIMIT=512M to /etc/default/armada on host.

Improvements

Set 0.5s timeout for fetching armada agents' versions in armada info.
$CONFIG_PATH environment variable inside microservice (and mounted volumes for these configs) are now updated during armada restart to pick up the new config directories.

0.16.1 (2016-04-04)

Bug fixes:

Fixed incorrect detection of dev environment during service recovery.

0.16.0 (2016-03-25)

Features

armada run adds configs to CONFIG_PATH for the renamed microservice name in addition to image name.
Config directories provided with armada run -c/--config are additionally appended with combinations of --env and --app_id when generating CONFIG_PATH.

Bug fixes

Fix bug with not recovering services run before armada 0.15.0.

0.15.0 (2016-03-22)

Features

Introduced resource limiting parameters to armada run command. These parameters accept same values as their respective docker parameters.
- --cpu-shares
- --memory
- --memory-swap
- --cgroup-parent

Improvements

hermes for python has been moved to pypi repository. It is available for download with pip install armada and can be imported with from armada import hermes command.
As a result built-in hermes has been marked as deprecated.
Suggest working workaround for accessing http dockyards in case they are running behind main-haproxy.

0.14.1 (2016-03-07)

Bug fixes

Fixed detection of development environment.
Fix bug with pulling image without explicit tag.

0.14.0 (2016-02-23)

Features

Support for dockyard-v2 (based on docker registry v2).
Support for HTTPS dockyards with self-signed certificates.
Detect not working remote HTTP dockyards and suggest workaround.
Added microservice_go base image.

Improvements

Upgrade Consul from version 0.6.0 to 0.6.3.
Run Consul using os.execv instead of spawning new process, which makes Consul's logs available in supervisor.
Store sorted container parameters in /opt/armada/ on Ship, to make checking differences between versions easier.
Allow microservices' local haproxy to forward http connections with Host header.

Bug fixes

Passing -d local to armada run in Armada vagrant box does not break detection of development environment.
Vagrant origin_dockyard_address parameter works properly with local dockyards by running localhost proxy.

0.13.1 (2016-01-08)

Bug fixes

armada ssh invoked for remote services, that were rebuilt from microservice image v0.13.0 threw KeyError - '22/tcp'.

0.13.0 (2015-12-30)

Features

Upgrade Consul from version 0.4.1 to 0.6.0.

Improvements

Don't expose port 22 and install openssh-server in microservice image anymore.
Increase timeouts to docker server, dockyards and local haproxy.
Remove hack for manual installing pip on Ubuntu.

⚠️ UPGRADE WARNING ⚠️

Due to Consul upgrade, the promoted Armada agents (leaders and commanders) with version >= 0.13.0 cannot be mixed with older versions. There is a 15 minute window for upgrade. See more details: https://github.com/hashicorp/consul/commit/f53bd94dc334c56968ac4e33e19d9ca6a2b5aa22

0.12.0 (2015-12-28)

Features

You can now pass options -t/--tty and -i/--interactive to armada ssh, which translate to corresponding docker exec options. Also the default behavior has been changed in case the command is provided. Both options, -t and -i, are turned off then.
New microservice template for armada create - python3 (based on bottle).

Improvements

Added link to docker docs about configuring insecure registries when connection to such registry is attempted. (Contributed by ryneeverett)

Bug fixes

Fix bug with armada ssh always returning exit code 0.
Fix recovering services after Docker/host restart.

0.11.1 (2015-12-17)

Improvements

Start local_magellan only when require_service is invoked.

Bug fixes

Fix hanging Armada on start, when using Docker version 1.9.1.
Use commands python2 and pip2 on Armada host in case the default python is 3.x. (Contributed by ryneeverett)
Fix detection of external Armada IP on some systems. (Contributed by ryneeverett)

0.11.0 (2015-12-07)

Improvements

Increased size of health checks logs. (previously: 2 files up to 1MB each. now: 3 files up to 5MB each)
Revamped armada ssh command by moving sshd process from microservice image to Armada container.
- docker exec is now used to enter a container on the same host.
- In order to enter container on another host, ssh is used to connect to remote Armada container.
- Containers created with new microservice image won't be accessible without updating Armada

Features

Added microservice_python3.5 base image.

0.10.0 (2015-11-12)

Improvements

Default health-check (main-port-open) passes when main service's port is open on any interface, and not only 127.0.0.1. (Contributed by zerofudge)
Changed armada create command syntax from armada create {template_name} [--name/-n {service_name}] to armada create {service-name} [--base-template/-b {template}]. Default value for --base-template is python.
Available templates are python and node. E.g.: armada create my-service -b node creates a directory with new service called my-service based on node template (using microservice_node base image).

Features

Added node.js microservice template to armada create command with node as a base_template name.

Bug fixes

armada create command now works properly in any directory on system.

0.9.0 (2015-10-21)

Features

Armada supports registering UDP services in catalog.
New armada service - example-udp that demonstrates registering UDP services and discovering it by others.

0.8.0 (2015-09-22)

Features

New base image microservice_node4 with node v4.0.0.
New armada service - Static file server

Improvements

Changed order of steps executed on armada restart. Instead of stop, pull, run now we do: pull, stop, run so that service's downtime is minimized.
Changed way of determining time when Docker daemon started to prevent unnecessary services' recoveries.

0.7.4 (2015-09-16)

Improvements

Updated node.js in microservice_node base image from v0.10.25 to v0.12.7.

0.7.3 (2015-09-14)

Bug fixes

Fixed a bug where microservices run with --rename parameter did not recover properly.
Fixed docker error when running restart-armada development script.

0.7.2 (2015-09-03)

Improvements

PID-1 zombie reaping problem resolved. Changed main system process inside containers from microservice init script to supervisor. Containers will no longer spawn multiple zombie processes e.g. after armada ssh command.
Note: This change requires rebuilding services created before this fix.

0.7.1 (2015-08-28)

Improvements

Access Consul agent from Armada commands by localhost, instead of private Docker IP, as it sometimes causes hanging requests.

0.7.0 (2015-08-21)

Features

Added parameter --rename to armada run.

0.6.5 (2015-08-06)

Improvements

armada stop and armada restart will now display microservice name and container id regardless of passed parameter.

0.6.4 (2015-07-17)

Bug fixes

Fixed armada run command on systems running python 2.6.

0.6.3 (2015-07-01)

Improvements

armada run now waits for the dockyards list to be restored from runtime_settings.json.

0.6.2 (2015-06-24)

Bug fixes

Fixed setting first dockyard as the default.

0.6.1 (2015-06-19)

Bug fixes

Fixed race condition bug when one service wants to discover multiple other services using local magellan (e.g. multiple calls to require_service.py in supervisor configs).

Improvements

Removed deprecated way of doing health checks. Armada doesn't have to run health checks by itself anymore. Instead, services based on microservice image report health checks directly to Consul.

0.6.0 (2015-05-14)

Features

Added custom health checks support for multi-service containers. Custom health check can be now specified with register_in_service_discovery script (-c / --health-check).

Improvements

If CONFIG_DIR environment variable is defined in Dockerfile, then hermes automatically will update CONFIG_PATH to include provided directory.
Armada commands now work properly with multi-service containers.
Services' initial health check is no longer set to 'passing'. Instead health checks are run on startup after a slight delay.

0.5.3 (2015-05-11)

Bug fixes

Updated docker-py library to 1.2.2 to fix authorization bug on docker pull.

0.5.2 (2015-05-07)

Improvements

armada recover now retries up to 5 times and returns list of those that failed to restart within retry limit if there were any.

0.5.1 (2015-04-29)

Improvements

Insert environment variables to crontab inside the container, one by one, to avoid not passing variables at all if one hits the crontab's limit of 1000 characters per line and warn if that's the case.
Skip passing internal Armada's environment variables to crontab.

0.5.0 (2015-04-22)

Improvements

Compatibility with docker version > 1.3.0 . Warning with help message will appear when trying to use insecure registry (dockyard over HTTP).
Improved local magellan with 'env inheritance' denoted by '/'. Now when using require_service script, if a service with matching MICROSERVICE_ENV is not found, local magellan will look for a service with 'parent' MICROSERVICE_ENV. e.g. if running a service with production/my-app/my-name env, local magellan will first look for a service with production/my-app/my-name env, then production/my-app env and finally production env.

0.4.4 (2015-03-10)

Improvements

microservice - Local magellan now configures HAProxy to actively refuse connections when required service is not available.

0.4.3 (2015-03-06)

Bug fixes

Fixed armada shutdown behaviour on CentOS.

0.4.2 (2015-02-27)

Bug fixes

Local magellan: require_service.py can now be overridden with empty --env or --app_id.

0.4.1 (2015-02-12)

Bug fixes

Fixed running containers from dockyards with authentication required, once again.

0.4.0 (2015-02-11)

Features

Changed domain armada.ganymede.eu to armada.sh.
service armada stop now doesn't remove runtime info about ship name, dockyards and connected ships. Previous behaviour is available with command service armada shutdown.
New base image: microservice_python3.

Bug fixes

Fixed armada push for default dockyard.

0.3.3 (2015-01-26)

Bug fixes

Fixed running containers from dockyards with authentication required.

0.3.2 (2015-01-22)

Bug fixes

'Created (UTC)' column in armada list -u now shows time in UTC instead of local.

Improvements

No retries when docker build command failed, as it usually indicates problem with Dockerfile or that user pressed Ctrl + c.

0.3.1 (2015-01-21)

Bug fixes

armada push now handles dockyards' authentication.

0.3.0 (2015-01-21)

Features

armada stop and armada restart now have option -a/--all to restart all services with given name.

Bug fixes

Detect failures of pulling from dockyards and return non-zero code to system.

0.2.0 (2015-01-19)

Features

Armada agent now recovers local running containers after machine failure or docker daemon restart.
armada recover command for manual recovery of running services from previously saved state.
Running containers' parameters are backed up every 1h in /opt/armada/saved_containers_backup/ for armada recover use.

0.1.0

Initial Armada version.

Files

CHANGELOG.md

Latest commit

History

CHANGELOG.md

File metadata and controls

Changelog

Unreleased

Improvements

2.11.3 (2021-03-10)

Bug fixes

2.11.2 (2021-03-02)

Bug fixes

2.11.1 (2021-02-11)

Bug fixes

2.11.0 (2021-02-02)

Improvements

Bug fixes

2.10.0 (2019-01-31)

Improvements

2.9.1 (2018-11-02)

Bug fixes

2.9.0 (2018-10-31)

Features

Bug fixes

2.8.1 (2018-05-09)

Bug fixes

2.8.0 (2018-04-24)

Features

2.7.1 (2018-03-29)

Bug fixes

2.7.0 (2018-03-21)

Features

Improvements

Bug fixes

2.6.0 (2018-03-09)

Features

Bug fixes

2.5.1 (2018-02-12)

Improvements

Bug fixes

2.5.0 (2018-02-05)

Features

Improvements

2.4.3 (2018-01-11)

Improvements

Bug fixes

2.4.2 (2018-01-02)

Improvements

2.4.1 (2017-12-01)

Improvements

2.4.0 (2017-11-17)

Features

2.3.0 (2017-11-03)

Features

Improvements

Bug fixes

2.2.0 (2017-10-19)

Features

Improvements

Bug fixes

2.1.2 (2017-09-20)

Bug fixes

2.1.1 (2017-09-13)

Improvements

2.1.0 (2017-07-25)

Features

Improvements

2.0.0 (2017-07-10)

Breaking changes

Features

Improvements

Bug fixes

1.17.0 (2017-06-20)

Features

Improvements

1.16.2 (2017-05-29)

Bug fixes

1.16.1 (2017-05-24)

Bug fixes