Could anyone help on this issue: https://github.com/fluent/fluent-bit/issues/11847?

Hi All, Facing this issue while installing fluent-bit for my services

16:43:41  W: Failed to fetch <https://packages.fluentbit.io/debian/bullseye/dists/bullseye/InRelease>  Certificate verification failed: The certificate is NOT trusted. The certificate chain uses expired certificate.  Could not handshake: Error in the certificate verification. [IP: 104.21.80.176 443]
16:43:41  W: Some index files failed to download. They have been ignored, or old ones used instead.

I'm not sure how to go about this..could someone let me know if they've encountered a similar issue and if so, how did you resolve it?

Hi team, I use s3 output in my project, I noticed upload was slow compared to input, and saw s3 option

workers

, which could help with my situation. I added

workers: 4

and started facing issue, I reverted change and it started working. I found issues on github around the same - ISS 10287, 10570. I see that these issues are waiting for triage.

Hi, Could you please review these PR when you have time? - https://github.com/fluent/fluent-bit/pull/11889 PR #11889 adds Azure Blob service principal authentication and includes testing details, debug logs, Valgrind output, and the related docs PR. Thanks!

Fluent Bit v5.0.7 is out https://fluentbit.io/announcements/v5.0.7/

Hi there, there is a CVE in fluent bit image. All 9 OpenSSL findings (incl. the Critical CVE-2026-34182) are already fixed in Debian trixie via

openssl 3.5.6-1~deb13u2

/ DSA-6335-1, published 2026-06-09. Can new image be published for fluent bit which addressed these CVEs. cc: @Rajeev Kumar Kallempudi

@Tianno Ross-Gray has left the channel

could anyone take a look at this? github.com/fluent/fluent-bit/issues/10821 -- no response from a dev since submitted Sept 2025

Hi everyone! I have a question about Fluent Bit's retry engine. Does Fluent Bit have any internal limit on the number of retry jobs/tasks, even when

Retry_Limit

is set to

no_limits

? I'm trying to understand whether there is any engine-level behavior that causes an output (or input) plugin to stop sending new logs when the retry queue is full, effectively waiting until there's enough space for the next retry task. If such a mechanism exists, could you point me to the relevant documentation or source code? Thanks!

Hi All, Has anyone faced this error in fluent-bit

2026/07/21 12:33:56.499] [ warn] [output:s3:s3.1] Could not buffer chunk. Data order preservation will be compromised [2026/07/21 12:33:56.500] [ warn] [engine] failed to flush chunk '<tel:11784637203|1-1784637203>.399331802.flb', retry in 20 seconds: task_i

d=337, input=tail.2 > output=s3.1 (out_id=1) [2026/07/21 12:33:57.455] [error] [output:s3:s3.1] Buffer is full: current_buffer_size=18446744073522589696, new_data=1855

9, store_dir_limit_size=2000000000 bytes

current buffer seems to be a very big data but that much data is not tailed..

Hi Team, Can we please have windows package for fluent bit 5.0.9 released as well. Github issue: github.com/fluent/fluent-bit/issues/12147

@Karan Agrawal has left the channel

Hi I am facing with this error

[2026/09/01 06:18:00.088] [error] % Failed to produce to topic pod-logs: Broker: Message size too large

Obviously the log lines are larger than the Redpanda topic’s

message.max.bytes

But I don't have the control on the log size and would not want to change the Redpanda settings as it is a central multi-tenant cluster. What I could do here on the FluentBit's configuration?

Hi All. We have configured

retry_limit 5

and

retry_exhausted_action delete

for the S3 output. In Fluent Bit 5.0.7 and 5.0.9, buffered files were successfully deleted from

store_dir

once the retry limit was reached. However, after upgrading to 5.1.1,we could see some of the files remain in the buffer path and continue to accumulate. As a result, buffer usage keeps increasing even after the retry limit has been exceeded. We are also unable to use 5.0.7/5.0.9 due to the underflow error issue. Has anyone observed this behaviour in 5.1.1. Can anyone help on this.

Fluent Bit v5.1.2 is out: fluentbit.io/announcements/v5.1.2

Hiya, got a little stuck setting routing with more than 10 entries. Is github.com/fluent/fluent-bit/issues/10294 a bug or by design please?

I can go back to routing with tags (as I did under Fluentd) but it seemed like leaning into conditional routing/labels etc was better/the new better design

obviously, I've got more than 10 entries so I'm just trying to understand if I should: 1. Add the max 10 per list to docs? 2. Wait 3. See how much C I remember

I'm just missing any context as to why it seems "designed" to die after 10 and can imagine there is "don't allocate something XXX" going on

Looks like the arbitrary 10 limit is still in place in latest version as well, I'm guessing it was just a number picked initially. @hhatake can you remember if it was specifically chosen?

Can Fluent Bit's in_forward or a custom plugin decode Filebeat's Lumberjack protocol directly (similar to how in_elasticsearch handles Bulk API payloads) to output plain logs?

hi folks, we are trying to track down a "memory leak." We had hoped that #11846 would resolve the problem, but unfortunately not. We queried 6 days of Thanos history for four Fluent Bit pods under normal cluster log volume. All four grew linearly and never recovered: t92nz: 728.7 MiB to 1602.0 MiB — 139.6 MiB/day, R² 0.999 xl5r8: 603.0 MiB to 1256.6 MiB — 105.3 MiB/day, R² 0.999 hk58z: 522.1 MiB to 957.2 MiB — 69.3 MiB/day, R² 0.998 72jnz: 500.1 MiB to 929.0 MiB — 68.1 MiB/day, R² 1.000 R² of 0.998–1.000 over six days is effectively a perfect straight line with no plateau — the leak signature from #11846, at the ~100 MB/day rate reporters describe. All four pods were running 5.1.1, confirmed from the image label on the historical series. This matches the dissenting reports on #11846 (users ivan-bushmarinov, TyranVdmKingMakers) rather than the "fixed in 5.1.0" consensus. Please let me know if there is any information that would be useful for tracking this down.

Hello I have a quick question about the new release v5.1.2, specially about the

Multiline
make the new built-in JSON multiline parser available to the filter and Tail input

The provided link brings us to this page docs.fluentbit.io/manual/pipeline/filters/multiline-stacktrace where I can read

The built-in json multiline parser groups pretty-printed JSON objects split across lines. When reading from a log file, prefer setting it on the Tail plugin rather than on this filter.

My english is bad so I don't really get it. So it is available for Tail plugin or not? Can somebody explain what does it mean please?

Hello. We have the store_dir_limit space available but we could still see buffer path full.

As I understand it, github.com/fluent/fluent-bit/pull/12403 fixes the append path, but cio_meta_read/_size/_cmp are still unguarded in chunkio, so other callers (fstore, chunk trace, backlog) can still hit a down chunk. Is a chunkio-level guard planned, or should that go up as a follow-up issue?

Hi team! Could this pr be reviewed? https://github.com/fluent/fluent-bit/pull/12436

Hi Team, we are seeing tail paused (mem buf overlimit) error and all other tail which does not have memory buffer limit configuration also stopped tailing. We could not see fluent-bit processing the logs anymore. Also we could see the memory limit of fluent bit container reached limit. 4.0,3 version we are currently using. Any suggestion to handle this.

Hi! I backported a commit from master to 5.1.x: github.com/fluent/fluent-bit/pull/12691. Could this be reviewed?

Hey , Just a clarification: We see that

.conf

format is being deprecated by end of 2026—does that mean it will eventually be removed entirely, or will it remain supported (just not receive new features)? Trying to gauge migration urgency. Thanks! cc: @Rajeev Kumar Kallempudi