- Update fluentd docker image to
v1.17.0-1.0
- Update rake requirement from ~> 0.9 to ~> 13.0
- Update yajl-ruby from 1.4.1 to 1.4.3
- Update rexml from 3.2.4 to 3.2.5
- Update fluentd from 1.11.1 to 1.14.6
- Update fluentd corresponding dependencies
- Include plugin gem name and version in the "User-Agent" header when uploading logs.
- Improve handling of records cannot fit within the configured
max_request_buffer. Single large records will now have the message field truncated to fit within a request if possible, and dropped otherwise.
- Updates to automated release deployment to have
latesttag in dockerhub updated correctly.
- Add a new configuration option
parserfor setting a per-event "parser" field without having to use filters. This is intended to be used in cases where configuring a filter is more difficult, such as Fargate.
-
Update the plugin so we ignore any unicode deserialization related errors which may arise serializing events to JSON in case event attribute string value contains bad or partial unicode escape sequence.
Previously, in cases like that, the plugin would throw an exception and such event would not be processed.
Now in cases like this, we recursively sanitize the event values and strip out any bad or partial unicode escape sequences.
- Change default value of
ssl_ca_bundle_pathtonil, causing the plugin to default to system certs. - Publish a Docker image of fluentd with this plugin installed to the tag
scalyr/fluentd.
- Add support for fluentd multiple process workers feature which allows users to configure fluentd to launch multiple fluentd workers to utilize multiple CPUs. This comes handy under heavy traffic scenarios where you want to split the load across multiple worker processes. (#14)
- Remove the monotonically increasing timestamp requirement which is not required by the server anymore. This allowed us to get rid of cross-thread synchronization which should simplify things. (#15)