Fluentd 错误：“缓冲区空间有太多数据”

Question

我在我的 kubernetes 集群中使用 fluentd 从 pod 收集日志并将它们发送到 elasticseach。一两天，fluetnd 收到错误：

[warn]: #0 emit transaction failed: error_class=Fluent::Plugin::Buffer::BufferOverflowError error=“buffer space has too many data” location=“/fluentd/vendor/bundle/ruby/2.6.0/gems/fluentd-1.7.4/lib/fluent/plugin/buffer.rb:265:in `write’”

并且 fluentd 停止发送日志，直到我重置 fluentd pod。

如何避免出现此错误？

也许我需要更改配置中的某些内容？

<match filter.Logs.**.System**>
  @type elasticsearch
  host "#{ENV['FLUENT_ELASTICSEARCH_HOST']}"
  port "#{ENV['FLUENT_ELASTICSEARCH_PORT']}"
  scheme "#{ENV['FLUENT_ELASTICSEARCH_SCHEME']}"
  user "#{ENV['FLUENT_ELASTICSEARCH_USER']}"
  password "#{ENV['FLUENT_ELASTICSEARCH_PASSWORD']}"

  logstash_format true
  logstash_prefix system
  type_name systemlog
  time_key_format %Y-%m-%dT%H:%M:%S.%NZ
  time_key time
  log_es_400_reason true
  <buffer>
    flush_thread_count "#{ENV['FLUENT_ELASTICSEARCH_BUFFER_FLUSH_THREAD_COUNT'] || '8'}"
    flush_interval "#{ENV['FLUENT_ELASTICSEARCH_BUFFER_FLUSH_INTERVAL'] || '5s'}"
    chunk_limit_size "#{ENV['FLUENT_ELASTICSEARCH_BUFFER_CHUNK_LIMIT_SIZE'] || '8M'}"
    queue_limit_length "#{ENV['FLUENT_ELASTICSEARCH_BUFFER_QUEUE_LIMIT_LENGTH'] || '32'}"
    retry_max_interval "#{ENV['FLUENT_ELASTICSEARCH_BUFFER_RETRY_MAX_INTERVAL'] || '30'}"
    retry_forever true
  </buffer>
</match>

Answer 1

默认的缓冲区类型是内存检查：https : //github.com/uken/fluent-plugin-elasticsearch/blob/master/lib/fluent/plugin/out_elasticsearch.rb#L63

这种类型的缓冲区有两个缺点——如果 pod 或容器重新启动，缓冲区中的日志将丢失。- 如果所有分配给 fluentd 的 RAM 都被消耗了，日志将不再发送

尝试使用具有以下配置的基于文件的缓冲区

<buffer>
    @type file
    path /fluentd/log/elastic-buffer
    flush_thread_count 8
    flush_interval 1s
    chunk_limit_size 32M
    queue_limit_length 4
    flush_mode interval
    retry_max_interval 30
    retry_forever true
  </buffer>