-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Segfault in cb_modifier_filter #4013
Comments
This might indeed explain why I did hit this issue and why it disappeared. The log statement definitely surpassed the 128 top level fields on occasion. With the change they always stay below 128 fields. Good to know that this limit currently exists, I have workloads that run at risk to hit that limit in the future. |
The patch #3982 is merged and it will be released as v1.8.7. |
Can we close this issue ? |
Closing @andreas-schroeder could you see if the new release helps? |
It probably does, but I took care for now that the workload respects the original field limit. Still good to know an upgrade option exists in case my workload runs into this issue again. Thanks for reaching out! |
Bug Report
Fluentbit segfaults on some JSON inputs. I'm opening this bug report as followup to this issue: aws/aws-for-fluent-bit#224
To Reproduce
I'm not entirely sure about what is causing the problem, and I don't have a local setup available to try to reproduce it. However, the segfaults stopped once we changed the field names in the logs from a dotted notation to proper nested JSON objects. I.e., we had log messages including (among other fields) fields with names of the form
with up to around 10 names.
Expected behavior
Not crashing ;)
Your Environment
Parsers File
Additional context
This setup works fine for three workloads, and crashes for a fourth workload when enabling trace logs (i.e. logs that are larger in size, and have a richer structure). The logs are all generated with logback logstash encoder, so I think it's valid to assume they are proper JSON.
I ran fluentbit with valgrind, and this is the stacktrace/issues it found.
The text was updated successfully, but these errors were encountered: