Describe the bug

I get weird timestamp lines in the log, seems like the sub-seconds part of the timestamps gets printed as a fraction, like in the following:
2021-07-19 12:56:20 +0200 [warn]: #0 failed to flush the buffer. retry_time=9 next_retry_seconds=2021-07-19 13:00:22 24648004639507749129/34359738368000000000 +0200 chunk="5c777a670a11ecf6fe7fe48328099911

this line is produced from fluentd-1.12.3/lib/fluent/plugin/output.rb:1238

To Reproduce

happens all the time when failing to flush buffers

Expected behavior

A sane timestamp, like next_retry_seconds=2021-07-19 12:52:09.802562942 +0200

Your Environment

- Fluentd version: 1.12.3 (b6e76c09d60ee866de6470f2bdd2551e7c8591d3
- TD Agent version: 4.1.1
- Operating system: Ubuntu 20.04.2 LTS
- Kernel version: 5.4.0-73-generic #82-Ubuntu

Your Configuration

<match **>
  @type forward
  <server>
    host some-host.example.com
  </server>
  <buffer>
    @type file
    path /srv/fluentd/sensor-queue-1.buffer
    chunk_limit_size 128k
    total_limit_size 1M
  </buffer>
</match>

Your Error Log

2021-07-19 12:56:20 +0200 [warn]: #0 failed to flush the buffer. retry_time=9 next_retry_seconds=2021-07-19 13:00:22 24648004639507749129/34359738368000000000 +0200 chunk="5c777a670a11ecf6fe7fe48328099911

Additional context

Nothing more