Skip to content

Latest commit

 

History

History
250 lines (182 loc) · 8.53 KB

Changes.md

File metadata and controls

250 lines (182 loc) · 8.53 KB
Raw

Faktory Changelog

Changelog: Faktory || Faktory Enterprise

1.9.0

  • Faktory now refuses further connections on :7419 if the connection count is larger than allowed, 2000 in Faktory or your licensed amount in Faktory Enterprise. [#476]
  • The INFO response has changed slightly, all elements now have well-defined types [#474] Deprecate client.Info Go API, add new client.CurrentState Go API which is almost identical to client.Info but has a strongly-typed response.
  • Implement native TLS support [#469] Put public.cert.pem and private.key.pem in your config directory and Faktory will automatically enable TLS on port 7419 and 7420.
  • Unwrap and display ActiveJob class names [#460, ibrahima]
  • Add Go client API so batches can push jobs in bulk [#437, tylerian]
  • Go 1.22, go-redis 9.5.1, alpine 3.18

1.8.0

  • Upgrade to Go 1.21 and go-redis 9.2.0.
  • Don't double encode HTML entities for display in Web UI [#440]
  • Validate and limit days parameter for Dashboard [#431] CVE-2023-37279
  • Validate and limit timeInterval refresh parameter for Dashboard

1.7.0

  • Upgrade all internal APIs to propogate Context
  • Upgrade redis/go-redis to the new v9 driver, requires Redis v6.0+ and RESP3
  • Upgrade datadog-go to the new v5 driver
  • Upgrade to Go 1.20 and Alpine 3.17

1.6.2

  • Add client.RemoveQueues(...) to delete existing queues via API
  • Add IO deadlines to client I/O [#373]
  • Upgrade to Go 1.19

1.6.1

  • Add per-worker connection count on Busy page
  • Add QUEUE REMOVE [queue_names...] subcommand to remove one or more queues [#399]
  • Fixes for linting warnings from gocritic and golangci-lint
  • Upgrade to Go 1.18.

1.6.0

  • The core Job struct in Go was modified slightly so the Retry element is now a int* rather than int. This was necessary so code could distinguish between the default value "0" and disabling retries "0"; the default value is now "nil" and will result in the default retry policy of "25". [#385]
  • Add remaining to the Job Failures struct to track retries remaining [#374]
  • Add new PUSHB PUSH Bulk jobs command. You send an Array of Jobs rather than a single Job with PUSH. There is no limit to the Array size but we recommend 1000 at a time as a best practice. Returns a Map of JID to ErrorMsg for jobs which failed to push. [#386]
PUSHB [{job},{job},...] => Map<JID, ErrorMsg>

1.5.5

  • Fix worker heartbeat monitoring which led to premature worker disassociation

1.5.4

  • Remove invalid client-side deadline which lead to lots of I/O errors [#375]
  • Upgrade to Go 1.17.

1.5.3

  • Remove trailing "-1" on Faktory versions and tags. Future versions will use the standard form of "vX.Y.Z" except for Linux DEB/RPM packages.
  • Implement client-side I/O deadlines, which will look like this if the network is slow or misconfigured: read tcp 127.0.0.1:63027->127.0.0.1:7419: i/o timeout
  • Build fixes for ARM support [#370]

1.5.2

  • Extend Faktory's Redis load timeout to allow for huge datasets [#225]
  • Upgrade Bootstrap from v3.3.7 to v5.0.1 [#360, saurori]
  • Support username in FAKTORY_URL, useful for connection routing proxies.
  • Publish arm64 Docker images [#366]

1.5.1

  • Change license from GPLv3 to AGPLv3. This is intended to ensure Faktory remains open source, even if someone forks Faktory and provides it as a service. No worker or application code is covered by this license.
  • Add support for native arm64 builds. Docker images still in progress.
  • Fixed Windows build.

1.5.0

  • Add QUEUE PAUSE and QUEUE RESUME commands, add buttons to the /queues page [#336]
  • The Busy page can now display current RSS for Worker processes [#339]
  • The /debug page now displays Redis latency [#337]
  • Add Dialer support for custom faktory.Client connections [#330]
  • Upgrade to Go 1.16 and alpine 3.13; remove go-bindata dependency [#338]

1.4.2

  • Allow FAKTORY_ENV to switch environment without flags [#325]
  • Fix jobs with Retry:0 not running FAIL middleware [#317]
  • Fix frequent "Bad connection EOF" log messages due to k8s probes
  • Upgrade to Go 1.15 and alpine 3.12.

1.4.1

  • Fix crash with invalid mutate API usage [#313]
  • Fix form handler on Scheduled Job and Retry Job UI pages [#236]
  • Refactor scheduled job processing to handle large job counts [#309]

1.4.0

  • Faktory Enterprise now supports Job Tracking.
  • Integrate golangci-lint and numerous minor changes for issues it raised
  • Support staging environment -e staging

1.3.0

  • Allow workers to send their current process state as part of BEAT [#266]
  • Faktory will now dump all thread stacktraces when sent the TTIN signal
  • Remove apex/log dependency [#289]
  • Upgrade to Go 1.14
  • Enable SLOWLOG in Faktory's Redis instance

1.2.0

  • Faktory Enterprise is now available with Batches and Queue Throttling.
  • Upgrade Docker image to Alpine 3.10
  • Add connection pool support for client Go package.
  • Refactor manager package for Pro/Ent integration.
  • Scheduled jobs now go through push middleware when pushed [#260]

1.1.0

  • Upgrade Go runtime from 1.10 to 1.13.
  • Faktory, Faktory Pro and faktory_worker_go are now using Go modules [#216, ClaytonNorthey92]
  • Fix "Add to Queue" button on the Scheduled page [#236]
  • Rework fetching jobs from queues to be more efficient. [#235]
  • Add several helper APIs to configure Faktory Pro features. [#251, tylerb]
  • You can now put nginx in front of Faktory's Web UI using proxy_pass:
location /faktory {
   proxy_set_header X-Script-Name /faktory;

   proxy_pass   http://127.0.0.1:7420;
   proxy_set_header Host $host;
   proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
   proxy_set_header X-Scheme $scheme;
   proxy_set_header X-Real-IP $remote_addr;
}

1.0.1

  • Pause Faktory boot and wait upon any LOADING errors from Redis

1.0.0

  • Support known, specific errors to be returned to clients [#208]
  • All jobs now default to "retry":25 if not specified when pushed [#211]
  • Update job arguments within Web UI to show as JSON, not native Go [#210]

0.9.7

0.9.6

  • Remove legacy job priority from APIs and Job struct
  • Improve display of job arguments and custom elements in Web UI [#199]

0.9.5

  • Add queue sizes to the INFO stats [#197, thapakazi]
  • Fix infinite loop with sorted set iteration [#196, antoinefinkelstein]

0.9.4

  • More aggressive Redis persistence [#195]
  • Fix possible race condition panic on new connection

0.9.3

  • Increase maximum client count to 1000.

0.9.2

  • Fix crash under load [#187]
  • First Faktory Pro release

0.9.1

  • Fix crash on startup in Linux in development mode
  • Close all associated connections when a worker process expires [#182]
  • Shutdown Redis cleanly if Faktory panics (Linux only)

0.9.0

  • Switch Faktory from RocksDB to Redis for storage. [#160]
  • Implement Faktory-internal middleware hooks [#168]
  • Integrate TOML config system [#169]

0.8.0

  • Workaround for negative busy/retry/scheduled/dead counts [#148]
  • Fix slow index page rendering under heavy load [#156]
  • Upgrade to Go 1.10.3
  • Upgrade to RocksDB 5.14.2

0.7.0

  • Upgrade RocksDB from 5.7.3 to 5.9.2
  • Add CSRF protection to Web UI [#92, vosmith]
  • Faktory in production mode now requires a password by default [#113]
  • Orphaned jobs now use the retry process so they don't re-enqueue forever [#99]
  • Implement automated backups, default once per hour and keeping 24. [#106]
  • Implement purge command for faktory-cli, to remove old backups.

0.6.1

  • Fix job reservation [#94, jweslley]
  • Send exhausted jobs to the morgue [#86, jweslley]

0.6.0

  • Add support for job priorities [#68, andrewstucki] Jobs may now include "priority" with values 1-9, 9 being highest priority. Push a job with "priority":9 and it will effectively go to the front of the queue. The default priority is 5.
{"jid":"12o31i2u3o1","jobtype":"FooJob","args":[1,2,3],"priority":8}
  • Various protocol changes [#72]
  • Remove TLS support in server [#76]
  • Fix heartbeat pruning so old workers disappear from Busy page [#37]
  • Add Docker image [#13]
  • Add Homebrew install [#10]
  • Lots of polish and code cleanup from cdrx, agnivade, adomokos, ustrajunior, jwsslley and others. Thank you!

0.5.0

  • Initial release