5.1.0

New Feature

• --log-formatter-type now support 'json-cee' as format.
  Thanks to @pdelagrave for the PR

5.0.0

First let's thanks @aeijdenberg for his hard work.

New feature / Improvement

• Load all apps before starting to read from nozzle. In this manner we won't fill our buffer by falling behind before
  we have apps populated.
• Instead of fetching all apps on scheduled basic, instead lazily re-fetch each app or space or org based on a
  randomized TTL.
• Allow for in-memory cache instead of bolt if not persisting the database (which by default bosh release is not, as
  the default path is not in a persistent location).
• strip-app-name-suffixes Add options to strip app name suffixes. This is useful for blue/green and zero
  downtime pushes where apps typically have temporary suffixes during renames. Since this data is often
  incorrect anyway, we may as well strip it.
• Stop using cf-client for important API calls, as it seems to often want to get smart and make successive calls for
  extra info that we don't need, contributing to latency and CPU load on API servers.

Of course as always we may have missed some corner case so please don't hesitate to open Github issue.

4.1.1

New feature / Improvement

• Remove noisy logs "cf_ignored_app": "false"
• Boltdb potential issue with the slice. Thanks to @aeijdenberg for the PR

4.1.0

Well a lot a rewrite into this release.

New feature / Improvement

• Orgs Filtering --orgs="" Now you can choose which orgs logs you wish to forward to your syslog endpoint. thanks to @datianshi for the PR.
• Skip Client SSL verification in case of SSL tcp --skip-ssl-validation-syslog

Fix

• --enable-stats-server is now disable by default, means if you want to enable stats server you need to do it explicitly.

4.0.3

Last release on the road to remove IRD=2 call to the CloudController

#167

Improvement

• Pulling by APP
  Remove the IRD=2 call when loading one unknow app.

4.0.2

First release on the road to remove IRD=2 call to the CloudController

#167

Improvement

• Pulling CC
  Remove the IRD=2 call when loading all apps.

--cc-rps is a new a new flag which let you put a RPS (client rate limiting) to avoid flooding the CC (50 by default)

4.0.1

New feature / Improvement*

• CPU usage thanks to @aleliaert the busy loop will sleep a little avoiding it to consume all cpu

4.0.0

Well a lot a rewrite into this release.

New feature / Improvement

• Buffering Same as loggregator itself, we use diodes (ring buffering) so now
  Firehose message are bufferize before being send to syslog, buffer size could be parametize --logs-buffer-size=10000
• Draining Due to the buffering we now drain the remaining message when the app being stop
  Draining will timeout after 30s (not configurable)
• Retrying connection to Firehose in case of deconnexion
• Embed Stats Server
  Provide information about buffer size and so on at http://IP:8080/stats
  Means you can use http healthcheck with CF
  Could be enable by using --enable-stats-server
  Still in early phase
• Unwanted event are now earlier filtered out means the buffer only contain wanted event.

Know issue

• Yes some test are missing ! Need more time to add them
• Race(s) condition in the Stats Server no real impact but still
• Drain timeout is not configurable
• Stats Server port is not configurable

3.3.2

Bug Fix

• The environment variables for max-retry-delay and max-retry-count were both set to read from MIN_RETRY_DELAY. #160 thanks @j-kaplan

3.3.1

Adding sirupsen