Skip to content

Latest commit

 

History

History
110 lines (81 loc) · 5.18 KB

CHANGELOG.md

File metadata and controls

110 lines (81 loc) · 5.18 KB
Raw

CHANGELOG

TO BE RELEASED

1.1.2 - 2/10/2016

  • Allow the "create-file-uploader-user" recipe to fail, to work around mysterious "usermod: no options" chef bug. Does not require anything to be done during a deployment. The recipe always works the first time it's run, it's just the second time that we sometimes see this failure.

1.1.1 - 2/3/2016

  • Allow java debug to be set to true or false via the java_debug_enabled stack custom_json parameter. It currently defaults to "true" (so, enabled) because most clusters are going to be development or testing clusters.

  • Install the newrelic agent when '{"newrelic": {"key": "your key"}' is included in the stack's custom_json. See the main mh-opsworks README for info.

  • REQUIRES MANUAL CHEF RECIPE RUNS Make the mysql dumps more bandwidth efficient by using "--compress" on the mysqldump client and an inline gzip when saving to shared storage. This causes bandwidth during mysql dumps to drop by 4 or 5 times without an appreciable performance penalty. This recipe can by run at any time, be sure to confirm the instance your dumps are happening on by finding this recipe in your active cluster config.

      # This is currently on the "Ganglia" layer in prod,
  # the value below is for dev clusters.
  ./bin/rake stack:commands:execute_recipes_on_layers layers="Admin" recipes="mh-opsworks-recipes::install-mysql-backups"

  • REQUIRES OPTIONAL CHEF RECIPE RUNS Installs the file_uploader user onto the instance of your choice. This allows you to use an rsync backchannel for uploads directly to the matterhorn inbox in combination with a couple cron jobs to copy and maintain these uploads. If you want to use this feature, add it to the setup lifecycle in the appropriate layer in your cluster config and then run this recipe. This recipe does not need to be run at any particular time relative to a deployment.

      # The current value for the admin node
  ./bin/rake stack:commands:execute_recipes_on_layers layers="Admin" recipes="mh-opsworks-recipes::create-file-uploader-user"

1.1.0 - 1/28/2016

  • All clusters should now include a private_assets_bucket_name in their cluster config, which can be the same as the cluster_config_bucket_name as that's a private bucket too and includes some shared assets.

  • REQUIRES CHEF RECIPE RUNS Rotate nginx logs more aggressively than a week, but still keep a year. All this change does is install a new logrotate file, overriding the default. The recipes can be run whenever.

      # After the recipes have been updated. . .
  ./bin/rake stack:commands:execute_recipes_on_layers recipes="mh-opsworks-recipes::configure-nginx-proxy" layers="Admin, Workers"
  ./bin/rake stack:commands:execute_recipes_on_layers recipes="mh-opsworks-recipes::configure-engage-nginx-proxy" layers="Engage"

  • REQUIRES MANUAL CHEF RECIPE RUNS Remove the "ok-action" alarms - they are too chatty and not useful. These recipe runs can happen post-deploy.

      ./bin/rake stack:commands:execute_recipes_on_layers recipes="mh-opsworks-recipes::create-alerts-from-opsworks-metrics"
  ./bin/rake stack:commands:execute_recipes_on_layers layers="Ganglia" recipes="mh-opsworks-recipes::create-mysql-alarms"
  ./bin/rake stack:commands:execute_recipes_on_layers layers="Admin" recipes="mh-opsworks-recipes::install-mysql-backups"
  ./bin/rake stack:commands:execute_recipes_on_layers layers="Admin,Workers,Engage,Utility,Asset Server" recipes="mh-opsworks-recipes::nfs-client"

    Be sure to double-check all layers and recipe combinations.

1.0.7 - 1/21/2016

  • Clean up s3 distribution configuration. Remove asset server hooks.
  • REQUIRES MANUAL CHEF RECIPE RUNS engage nginx uid cookie & usertracking session id + uid log for identifying shopping period viewers
  • REQUIRES MANUAL CHEF RECIPE RUNS Use mo-scaler --scale-available option

1.0.6 - 1/18/2016

  • Configuration for the new Matterhorn S3 distribution service.

1.0.5 - 1/14/2016

  • Hot fix for Otherpubs sevice config

1.0.4 - 1/14/2016

  • REQUIRES MANUAL CHEF RECIPE RUNS Send X-Forwarded-For header from the nginx proxies so that user-tracking gets correct client ip
  • Get rid of the no longer necessary ffmpeg package removal
  • Modify mh_default and otherpubs config for MATT-1822 auto populate admin UI upload

1.0.3 - 1/8/2016

  • Fix find's mtime invocation to ensure we're compressing and reaping files at the correct times. Requires that the MySQL backup installation chef recipe be run on the admin node.
  • Relax the matterhorn_availability alert to require errors over two 2 minute periods instead of just one.
  • Modify the matterhorn init script to check that the daemon started properly, working around the race condition related to bundle loading. See the relevant commit for details.

1.0.2 - 1/6/2016

  • Only set the org.opencastproject.server.maxload property on worker nodes - all other nodes can have an unlimited number of jobs.

1.0.1 - 1/5/2016

  • Allow the default "start matterhorn if it's not running after deploy" to be overridden via the :dont_start_matterhorn_after_deploy chef custom_json attribute.

1.0.0 - 1/4/2016

  • Initial release