fluent/fluentd · error · Fluent::ConfigError

timekey should be greater than 0. current timekey: #{@timeke

Error message

timekey should be greater than 0. current timekey: #{@timekey}

What it means

`timekey` was configured but parses to zero or negative seconds. timekey is a time-typed parameter, so `timekey 0`, `timekey -1h`, or a value that evaluates to 0 passes config parsing and is then rejected by the explicit `@timekey <= 0` check, because a non-positive window cannot bucket records.

Source

Thrown at lib/fluent/plugin/output.rb:335

                    false
                  end
                end
              rescue => e
                raise Fluent::ConfigError, "in chunk_keys: #{e.message}"
              end
            }
            raise Fluent::ConfigError, "chunk_keys specification includes invalid char"
          else
            @chunk_key_accessors = Hash[@chunk_keys.map { |key| [key.to_sym, Fluent::PluginHelper::RecordAccessor::Accessor.new(key)] }]
          end

          if @chunk_key_time
            raise Fluent::ConfigError, "<buffer ...> argument includes 'time', but timekey is not configured" unless @buffer_config.timekey
            Fluent::Timezone.validate!(@buffer_config.timekey_zone)
            @timekey_zone = @buffer_config.timekey_use_utc ? '+0000' : @buffer_config.timekey_zone
            @timekey = @buffer_config.timekey
            if @timekey <= 0
              raise Fluent::ConfigError, "timekey should be greater than 0. current timekey: #{@timekey}"
            end
            @timekey_use_utc = @buffer_config.timekey_use_utc
            @offset = Fluent::Timezone.utc_offset(@timekey_zone)
            @calculate_offset = @offset.respond_to?(:call) ? @offset : nil
            @output_time_formatter_cache = {}
          end

          if (@chunk_key_tag ? 1 : 0) + @chunk_keys.size >= CHUNKING_FIELD_WARN_NUM
            log.warn "many chunk keys specified, and it may cause too many chunks on your system."
          end

          # no chunk keys or only tags (chunking can be done without iterating event stream)
          @simple_chunking = !@chunk_key_time && @chunk_keys.empty?

          @flush_mode = @buffer_config.flush_mode
          if @flush_mode == :default
            if has_flush_interval
              log.info "'flush_interval' is configured at out side of <buffer>. 'flush_mode' is set to 'interval' to keep existing behaviour"

View on GitHub (pinned to dd45c6e18d)

Solutions

1. Set a positive window such as `timekey 5m`, `timekey 1h`, or `timekey 1d`
2. If you do not want time chunking, remove `time` from the chunk keys instead of zeroing timekey
3. Check env-substituted values actually resolve (e.g. `timekey #{ENV['FLUENT_TIMEKEY']}`)

Example fix

# before
<buffer time>
  @type file
  timekey 0
</buffer>

# after
<buffer time>
  @type file
  timekey 1h
</buffer>

fluentd --dry-run -c /etc/fluent/fluent.conf
# if timekey comes from env: echo "timekey=#{ENV['FLUENT_TIMEKEY']}" and assert > 0

When it happens

Trigger: `timekey 0` (sometimes written intending 'no window'), `timekey -300`, or a computed/env-substituted value that resolves to 0. Raised at output.rb:335-337 during configure.

Common situations: Attempting to disable time chunking by zeroing the key; using an environment variable (`timekey #{ENV['WINDOW']}`) that is empty and coerces to 0; copy/paste from an example with a placeholder value.

Related errors

• chunk_keys specification includes invalid char
• <buffer ...> argument includes 'time', but timekey is not co
• this plugin '#{self.class}' cannot handle arguments for <buf
• this plugin '#{self.class}' allows <buffer tag> only
• time_slice_format only with %Y or %m is too long

AI-assisted analysis of fluent/fluentd@dd45c6e18d (2026-08-21). Data as JSON: /api/errors/83d33eaf5a214136. Report an issue: GitHub.