Closed ayushiaks closed 3 years ago
First, please use v1.12.4 or later, or v1.11.x. v1.12.0 - v1.12.3 has some bugs in in_tail. e.g.) #3393
When it's reproduces even if such versions, please check following things:
Thanks for pointing this out @ashie, after investigation we found out that few logs are not being parsed. We're using two different logging libraries and they're cutting each other's logs in between, rendering them non-parseable. We're trying to figure out how to handle multithreaded logging scenarios like these.
Thanks a lot for your help, closing this issue. Although, do let me know if you've any leads on how to solve the issue.
Check CONTRIBUTING guideline first and here is the list to help us investigate the problem. Describe the bug
We've been missing logs after they are rotated very frequently (high load on application). We're working with Azure Kubernetes Service. Upon investigating, we confirmed the following:
Our rotate_wait param is set to default (5 secs). Logs are being rotated properly as per fluentd logs. Seeing the missed logs in compressed files, is the position file not being updated correctly? Which doesn't make sense either with logs missing from random places.
To Reproduce
Expected behavior
Logs should not be missing ideally
Your Environment
fluentd --version
ortd-agent --version
fluentd 1.12.3cat /etc/os-release
uname -r
5.4.0-1047-azureIf you hit the problem with older fluentd version, try latest version first.
Your Configuration
Kubernetes.conf
fluentd.conf
Your Error Log
Additional context