fluent-bit
fluent-bit copied to clipboard
IMDSv2 calls have no timeouts
Bug Report
Describe the bug
In order to harden our security in our kubernetes clusters we recently set the http-put-response-hop-limit to 1 instead of 2 on our instances so that containers wouldn't be able to assume the host instance's role.
This caused fluentbit (running in containers without host networking) to silently stop sending logs because the calls to imdsv2 are dropped by AWS if the number of hops are greater than the limit and because the aws filter calls to imdsv2 have no timeouts making fluentbit wait forever.
To Reproduce
Use the AWS filter in a container running on an instance having http-put-response-hop-limit set to 1.
Expected behavior
AWS filter fails with error if it is not able to reach imdsv2 endpoint.
Screenshots
Your Environment
- Version used: 2.32.5
- Configuration:
- Environment name and version (e.g. Kubernetes? What version?):
- Server type and version:
- Operating System and version:
- Filters and plugins:
Additional context
The version provided is not an OSS version, I'm guessing you're on an AWS specific version so I would raise it there or use an OSS version: https://github.com/aws/aws-for-fluent-bit
Please also follow the template, there is no actual config included.
This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 5 days. Maintainers can add the exempt-stale label.
@patrick-stephens you're right, I am using aws-for-fluent-bit v2.35.5 which uses fluentbit v1.9.10.
I've looked at fluentbit code and it seems fluentbit has its own HTTP client implementation, hasn't it ?
I've looked and I did not find any mention of timeout.
This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 5 days. Maintainers can add the exempt-stale label.
This issue was closed because it has been stalled for 5 days with no activity.