Skip to content

Releases: travis-ci/worker

1.3.0 (November 30th, 2015)

30 Nov 19:13
Compare
Choose a tag to compare

IMPROVEMENTS:

  • backend/docker: Tests added by @jacobgreenleaf 😻
  • utils/package-*, bin/travis-worker-install: Rework and integration into
    base CI suite.
  • Switch to using gvt for dependency management.
  • backend/bluebox: Removed
  • amqp_job: Send all known timestamps during state updates.
  • backend: Set defaults for all StartAttributes fields, which are also
    exposed via CLI.

BUG FIXES:

  • backend/docker: Use correct HostConfig when creating container
  • image/api_selector: Set is_default=true when queries consist of a single
    search dimension.

v1.2.0

10 Nov 23:56
Compare
Choose a tag to compare

1.2.0 (November 10th, 2015)

IMPROVEMENTS:

  • utils/pkg: Official releases built with go 1.5.1.
  • vendor/*: Updated all vendored dependencies
  • utils/lintall: Set 1m deadline for all linters
  • backend/jupiterbrain: switch to env image selector
  • image/api:
    • Add image selection query with dist + group + language
    • Add last-ditch image selection query with is_default=true

FEATURES:

  • log-writer: Write folded worker info summary

BUG FIXES:

  • backend/gce: Removed wonky instance group feature
  • step/run-script: Do not requeue if max log length exceeded

v1.1.1

11 Sep 01:25
Compare
Choose a tag to compare

1.1.1 (September 10th, 2015)

IMPROVEMENTS:

  • utils/pkg: updated upstart config to copy/run executable as
    /var/tmp/run/$UPSTART_JOB, allowing for multiple worker instances per
    host.

v1.1.0

09 Sep 20:52
Compare
Choose a tag to compare

1.1.0 (September 9th, 2015)

FEATURES:

  • backend/gce:
    • Configurable image selector, defaulting to legacy selection
      method for backward compatibility.
    • Support for reading account JSON from filename or JSON blob.
    • Optionally add all instances to configurable instance group.
  • image/*: New image selection abstraction with env-based and api-based
    implementations.

IMPROVEMENTS:

  • vendor/*: Upgraded all vendored dependencies to latest.
  • utils/pkg:
    • Official releases built with go 1.5.
    • Packagecloud script altered to only use ruby stdlib dependencies,
      removing the need for bundler.
  • backend/gce: Lots of additional test coverage.
  • backend/*: Introduction of Setup func for deferring mutative actions needed
    for runtime.
  • config: Addition of Unset method on ProviderConfig

BUG FIXES:

  • processor: Fix graceful shutdown by using tryClose on shutdown channel.

v1.0.0

19 Aug 18:56
Compare
Choose a tag to compare

1.0.0 (August 19th, 2015)

FEATURES:

  • backend/gce: Add auto implode, which will cause a VM to automatically shut
    down after a hard timeout.

BUG FIXES:

  • logger: Make the processor= field in the logs not be empty anymore
  • sentry: Stringify the err field sent to Sentry, since it's usually parsed
    as a struct, making it just {} in Sentry.

v0.7.0

19 Aug 02:32
Compare
Choose a tag to compare

0.7.0 (August 18, 2015)

FEATURES:

  • backend: The local backend was added

IMPROVEMENTS:

  • backend/jupiterbrain: Add exponential backoff on all HTTP requests
  • sentry: Include stack trace in logs sent to Sentry
  • step/generate-script: Add exponential backoff to script generation

BUG FIXES:

  • backend/gce: Fix a bug causing VMs for a build language ending in symbols
    (such as C++) to error while booting
  • log-writer: Fix a race condition causing the log writer to be closed before
    the logs were fully flushed.
  • log-writer: Minimize locking in the internals of the log writer, making
    deadlocks less likely.
  • processor: Fix graceful and forceful shutdown when there are still build
    jobs waiting.

v0.6.0

23 Jul 20:41
Compare
Choose a tag to compare

0.6.0 (July 23rd, 2015)

FEATURES:

  • backend: The GCE backend was added

IMPROVEMENTS:

  • step/upload-script: Add a timeout for the script upload (currently 1 minute)
  • step/upload-script: Treat connection errors as recoverable errors, and requeue the job
  • backend/jupiterbrain: Per-image boot time and count metrics

BUG FIXES:

  • backend/jupiterbrain: Fix a goroutine/memory leak where SSH connections for cancelled jobs wouldn't get cleaned up
  • logger: Don't print the job UUID if it's blank
  • processor: Fix a panic that would sometimes happen on graceful shutdown

v0.5.2

16 Jul 12:19
Compare
Choose a tag to compare

0.5.2 (July 16th, 2015)

IMPROVEMENTS:

  • config: Use the server hostname by default if no Librato source is given
  • version: Only print the basename of the binary when showing version

BUG FIXES:

  • step/run-script: Print the log timeout and not the hard timeout in the log
    timeout error message [GH-49]

v0.5.1

15 Jul 14:42
Compare
Choose a tag to compare

0.5.1 (July 14th, 2015)

FEATURES:

  • Runtime pool size management. Send SIGTTIN and SIGTTOU signals to increase and decrease the pool size during runtime. [GH-42]
  • Report runtime memory metrics, including GC pause times and rates, and goroutine count. [GH-45]

OTHER:

  • Add more log messages so that all error messages are caught in some way.

v0.5.0

09 Jul 16:03
Compare
Choose a tag to compare

0.5.0 (July 9th, 2015)

BACKWARDS INCOMPATIBILITIES:

  • backend: Sauce Labs backend was removed [GH-36]

FEATURES:

  • backend: Blue Box backend was added [GH-32]
  • main: lifecycle hooks were added [GH-33]
  • the log timeout can be set in the configuration
  • the log timeout and hard timeout can be set per-job in the payload from AMQP [GH-34]