Cannot use compressed buffer with elasticsearch data stream

[OranShuster]

• read the contribution guideline

Problem

I am ingesting very big logs (30MB) so i want to use a compressed buffer + ES compression to reduce the size
When i set the buffer to be gzip compressed i get the following errors

2022-10-12 14:26:08 +0000 [warn]: #0 [elasticearch_output] failed to flush the buffer. retry_times=0 next_retry_time=2022-10-12 14:26:09 +0000 chunk="5ead67778ec810a53837e51b712620ee" error_class=Zlib::GzipFile::Error error="not in gzip format"

Steps to replicate

  <match **>
    @type elasticsearch_data_stream
    @id elasticearch_output
    data_stream_name fluentd-k8s-master-audit
    data_stream_template_name fluentd-k8s-master-audit-template
    data_stream_ilm_policy default-ilm
    hosts "#{ENV['ES_HOSTS']}"
    logstash_format false
    verify_es_version_at_startup false
    default_elasticsearch_version 7
    include_timestamp true
    compression_level default_compression
    <buffer>
      @type file
      path /usr/share/fluentd-k8s-master/buffers/k8s-master-audit
      flush_thread_count 3
      chunk_limit_size 10mb
      compress gzip
    </buffer>
  </match>

Expected Behavior or What you need to ask

Using a gzip compressed buffer shouldn't lead to errors

Using Fluentd and ES plugin versions

• OS version N/A
• Kubernetes 1.20
• Fluentd v1.15
• ES plugin 5.2.4
• ES version 7.16.2