in_tail: Serialize DB operation for preventing resource collisions on threaded mode

[cosmo0920]

Closes #11357.

---

Enter [N/A] in the box, if an item is not applicable to your change.

Testing
Before we can approve your change; please submit the following in a comment:

• Example configuration file for the change

[SERVICE]
    Log_Level debug
    Flush 1
    Daemon Off
    # Parsers_File parsers.conf

[INPUT]
    Name tail
    Tag access
    Path access.log
    Path_Key filepath
    Key data
    Read_from_Head true
    DB file-access.db
    DB.locking true
    Refresh_Interval 10
    Threaded true

[OUTPUT]
    Name stdout
    Match *

[OUTPUT]
    Name file
    Match access
    Path ./output/
    File access.log
    Mkdir True

On another terminal before launching fluent-bit:

$ echo "line1" > access.log; echo "line2" >> access.log; echo "line3" >> access.log

after fluent-bit launched:

$ echo "line4" >> access.log; mv access.log access.log.1; echo "line5" > access.log

There's no duplicated lines:

% cat output/access.log
access: [1770973676.233236000, {"filepath":"access.log","data":"line1"}]
access: [1770973676.233240000, {"filepath":"access.log","data":"line2"}]
access: [1770973676.233240000, {"filepath":"access.log","data":"line3"}]
access: [1770973705.990725000, {"filepath":"access.log","data":"line4"}]
access: [1770973706.238184000, {"filepath":"access.log","data":"line5"}]

• Debug log output from testing the change

Fluent Bit v5.0.0
* Copyright (C) 2015-2025 The Fluent Bit Authors
* Fluent Bit is a CNCF graduated project under the Fluent organization
* https://fluentbit.io

______ _                  _    ______ _ _           _____  _____           _            
|  ___| |                | |   | ___ (_) |         |  ___||  _  |         | |           
| |_  | |_   _  ___ _ __ | |_  | |_/ /_| |_  __   _|___ \ | |/' |______ __| | _____   __
|  _| | | | | |/ _ \ '_ \| __| | ___ \ | __| \ \ / /   \ \|  /| |______/ _` |/ _ \ \ / /
| |   | | |_| |  __/ | | | |_  | |_/ / | |_   \ V //\__/ /\ |_/ /     | (_| |  __/\ V / 
\_|   |_|\__,_|\___|_| |_|\__| \____/|_|\__|   \_/ \____(_)\___/       \__,_|\___| \_/


[2026/02/13 18:07:56.229900000] [ info] Configuration:
[2026/02/13 18:07:56.229910000] [ info]  flush time     | 1.000000 seconds
[2026/02/13 18:07:56.229917000] [ info]  grace          | 5 seconds
[2026/02/13 18:07:56.229924000] [ info]  daemon         | 0
[2026/02/13 18:07:56.229929000] [ info] ___________
[2026/02/13 18:07:56.229934000] [ info]  inputs:
[2026/02/13 18:07:56.229938000] [ info]      tail
[2026/02/13 18:07:56.229943000] [ info] ___________
[2026/02/13 18:07:56.229947000] [ info]  filters:
[2026/02/13 18:07:56.229952000] [ info] ___________
[2026/02/13 18:07:56.229956000] [ info]  outputs:
[2026/02/13 18:07:56.229960000] [ info]      stdout.0
[2026/02/13 18:07:56.229965000] [ info]      file.1
[2026/02/13 18:07:56.229969000] [ info] ___________
[2026/02/13 18:07:56.229976000] [ info]  collectors:
[2026/02/13 18:07:56.230140000] [ info] [fluent bit] version=5.0.0, commit=5857b7dfe8, pid=6313
[2026/02/13 18:07:56.230149000] [debug] [engine] coroutine stack size: 36864 bytes (36.0K)
[2026/02/13 18:07:56.230204000] [ info] [storage] ver=1.5.4, type=memory, sync=normal, checksum=off, max_chunks_up=128
[2026/02/13 18:07:56.230219000] [ info] [simd    ] NEON
[2026/02/13 18:07:56.230223000] [ info] [cmetrics] version=1.0.7
[2026/02/13 18:07:56.230249000] [ info] [ctraces ] version=0.7.0
[2026/02/13 18:07:56.230341000] [ info] [input:tail:tail.0] initializing
[2026/02/13 18:07:56.230345000] [ info] [input:tail:tail.0] storage_strategy='memory' (memory only)
[2026/02/13 18:07:56.232637000] [debug] [input:tail:tail.0] flb_tail_fs_stat_init() initializing stat tail input
[2026/02/13 18:07:56.232647000] [debug] [input:tail:tail.0] scanning path access.log
[2026/02/13 18:07:56.232706000] [debug] [input:tail:tail.0] db file set called for access.log inode=88672876
[2026/02/13 18:07:56.232908000] [debug] [input:tail:tail.0] inode=88672876 with offset=0 appended as access.log
[2026/02/13 18:07:56.232913000] [debug] [input:tail:tail.0] scan_glob add(): access.log, inode 88672876
[2026/02/13 18:07:56.232917000] [debug] [input:tail:tail.0] 1 new files found on path 'access.log'
[2026/02/13 18:07:56.232945000] [ info] [input:tail:tail.0] db: delete unmonitored stale inodes from the database: count=0
[2026/02/13 18:07:56.232951000] [debug] [input:tail:tail.0] [thread init] initialization OK
[2026/02/13 18:07:56.232959000] [ info] [input:tail:tail.0] thread instance initialized
[2026/02/13 18:07:56.232967000] [debug] [tail:tail.0] created event channels: read=45 write=46
[2026/02/13 18:07:56.232976000] [debug] [stdout:stdout.0] created event channels: read=49 write=50
[2026/02/13 18:07:56.233038000] [debug] [file:file.1] created event channels: read=58 write=59
[2026/02/13 18:07:56.233074000] [ info] [output:stdout:stdout.0] worker #0 started
[2026/02/13 18:07:56.233114000] [debug] [router] match rule tail.0:stdout.0
[2026/02/13 18:07:56.233119000] [debug] [router] match rule tail.0:file.1
[2026/02/13 18:07:56.233118000] [ info] [output:file:file.1] worker #0 started
[2026/02/13 18:07:56.233163000] [ info] [sp] stream processor started
[2026/02/13 18:07:56.233204000] [ info] [engine] Shutdown Grace Period=5, Shutdown Input Grace Period=2
[2026/02/13 18:07:56.233300000] [debug] [input:tail:tail.0] [static files] processed 18b
[2026/02/13 18:07:56.233308000] [debug] [input:tail:tail.0] inode=88672876 file=access.log promote to TAIL_EVENT
[2026/02/13 18:07:56.233320000] [debug] [input:tail:tail.0] [static files] processed 0b, done
[2026/02/13 18:07:57.236570000] [debug] [task] created task=0x6000037b4000 id=0 OK
[2026/02/13 18:07:57.236673000] [debug] [output:stdout:stdout.0] task_id=0 assigned to thread #0
[2026/02/13 18:07:57.236760000] [debug] [output:file:file.1] task_id=0 assigned to thread #0
[0] access: [[1770973676.233236000, {}], {"filepath"=>"access.log", "data"=>"line1"}]
[1] access: [[1770973676.233240000, {}], {"filepath"=>"access.log", "data"=>"line2"}]
[2] access: [[1770973676.233240000, {}], {"filepath"=>"access.log", "data"=>"line3"}]
[2026/02/13 18:07:57.237305000] [debug] [output:file:file.1] creating directory ./output
[2026/02/13 18:07:57.237433000] [debug] [out flush] cb_destroy coro_id=0
[2026/02/13 18:07:57.239142000] [debug] [out flush] cb_destroy coro_id=0
[2026/02/13 18:07:57.239280000] [debug] [task] destroy task=0x6000037b4000 (task_id=0)
[2026/02/13 18:08:06.236668000] [debug] [input:tail:tail.0] scanning path access.log
[2026/02/13 18:08:06.236734000] [debug] [input:tail:tail.0] scan_blog add(): dismissed: access.log, inode 88672876
[2026/02/13 18:08:06.236746000] [debug] [input:tail:tail.0] 0 new files found on path 'access.log'
[2026/02/13 18:08:16.236506000] [debug] [input:tail:tail.0] scanning path access.log
[2026/02/13 18:08:16.236528000] [debug] [input:tail:tail.0] scan_blog add(): dismissed: access.log, inode 88672876
[2026/02/13 18:08:16.236533000] [debug] [input:tail:tail.0] 0 new files found on path 'access.log'
[2026/02/13 18:08:26.236736000] [debug] [input:tail:tail.0] inode=88672876 rotated: access.log => /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1
[2026/02/13 18:08:26.236821000] [debug] [input:tail:tail.0] inode=88672876 rotated access.log -> /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1
[2026/02/13 18:08:26.236936000] [debug] [task] created task=0x6000037a8000 id=0 OK
[2026/02/13 18:08:26.236891000] [ info] [input:tail:tail.0] inode=88672876 handle rotation(): access.log => /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1
[2026/02/13 18:08:26.236985000] [debug] [output:stdout:stdout.0] task_id=0 assigned to thread #0
[2026/02/13 18:08:26.237020000] [debug] [output:file:file.1] task_id=0 assigned to thread #0
[0] access: [[1770973705.990725000, {}], {"filepath"=>"access.log", "data"=>"line4"}]
[2026/02/13 18:08:26.237351000] [debug] [out flush] cb_destroy coro_id=1
[2026/02/13 18:08:26.237718000] [debug] [input:tail:tail.0] db file set called for access.log inode=88673054
[2026/02/13 18:08:26.238015000] [debug] [input:tail:tail.0] inode=88673054 with offset=0 appended as access.log
[2026/02/13 18:08:26.238064000] [debug] [input:tail:tail.0] scanning path access.log
[2026/02/13 18:08:26.238119000] [debug] [input:tail:tail.0] scan_blog add(): dismissed: access.log, inode 88673054
[2026/02/13 18:08:26.238134000] [debug] [out flush] cb_destroy coro_id=1
[2026/02/13 18:08:26.238139000] [debug] [input:tail:tail.0] 0 new files found on path 'access.log'
[2026/02/13 18:08:26.238222000] [debug] [task] destroy task=0x6000037a8000 (task_id=0)
[2026/02/13 18:08:26.238652000] [debug] [input:tail:tail.0] [static files] processed 6b
[2026/02/13 18:08:26.238775000] [debug] [input:tail:tail.0] inode=88673054 file=access.log promote to TAIL_EVENT
[2026/02/13 18:08:26.238928000] [debug] [input:tail:tail.0] [static files] processed 0b, done
[2026/02/13 18:08:27.238113000] [debug] [task] created task=0x6000037a8000 id=0 OK
[2026/02/13 18:08:27.238206000] [debug] [output:stdout:stdout.0] task_id=0 assigned to thread #0
[0] access: [[1770973706.238184000, {}], {"filepath"=>"access.log", "data"=>"line5"}]
[2026/02/13 18:08:27.238292000] [debug] [output:file:file.1] task_id=0 assigned to thread #0
[2026/02/13 18:08:27.238485000] [debug] [out flush] cb_destroy coro_id=2
[2026/02/13 18:08:27.239346000] [debug] [out flush] cb_destroy coro_id=2
[2026/02/13 18:08:27.239514000] [debug] [task] destroy task=0x6000037a8000 (task_id=0)
[2026/02/13 18:08:31.236829000] [debug] [input:tail:tail.0] inode=88672876 purge rotated file /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1 (offset=24 / size = 24)
[2026/02/13 18:08:31.236912000] [debug] [input:tail:tail.0] inode=88672876 removing file name /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1
[2026/02/13 18:08:31.237333000] [debug] [input:tail:tail.0] db: file deleted from database: /Users/hiroshihatake/GitHub/fluent-bit/build/access.log.1
^C[2026/02/13 18:08:34] [engine] caught signal (SIGINT)
[2026/02/13 18:08:34.81773000] [debug] [input:tail:tail.0] thread pause instance
[2026/02/13 18:08:34.81836000] [ info] [output:stdout:stdout.0] thread worker #0 stopping...
[2026/02/13 18:08:34.81946000] [ info] [output:stdout:stdout.0] thread worker #0 stopped
[2026/02/13 18:08:34.82048000] [ info] [output:file:file.1] thread worker #0 stopping...
[2026/02/13 18:08:34.82152000] [ info] [output:file:file.1] thread worker #0 stopped
[2026/02/13 18:08:34.82257000] [debug] [input:tail:tail.0] inode=88673054 removing file name access.log
[2026/02/13 18:08:34.83517000] [debug] [input:tail:tail.0] thread exit instance

• Attached Valgrind output that shows no leaks or memory corruption was found

If this is a change to packaging of containers or native binaries then please confirm it works for all targets.

• Run local packaging test showing all targets (including any new ones) build.
• Set ok-package-test label to test for all targets (requires maintainer to do).

Documentation

• Documentation required for this feature

Backporting

• Backport to latest stable release.

---

Fluent Bit is licensed under Apache 2.0, by submitting this pull request I understand that this code will be released under the terms of that license.

Summary by CodeRabbit

• Bug Fixes
  • Added synchronization around database operations to prevent concurrent access conflicts and improve data integrity; locks now released on all success and error paths to avoid deadlocks.
  • Added debug logging for database operations to aid troubleshooting and visibility.
  • Extended lock retry logic to treat an additional transient error as retryable, improving robustness of lock acquisition.

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds DB lock/unlock helpers and wraps tail plugin database operations (open, file set, offset, rotate, delete, stale-file delete) with explicit locking and unlock-on-error; updates flb_tail_db_open signature to accept ctx and inserts debug logs for DB-targeted file/inode operations.

Changes

Cohort / File(s)	Summary
Tail DB locking & ops plugins/in_tail/tail_db.c	Add tail_db_lock/tail_db_unlock helpers; change flb_tail_db_open signature to include ctx; acquire/release DB lock around flb_tail_db_file_set, flb_tail_db_file_offset, flb_tail_db_file_rotate, flb_tail_db_file_delete, flb_tail_db_stale_file_delete; add debug logs and ensure unlock on all error paths.
Lock retry behavior src/flb_lock.c	Treat EBUSY as transient in flb_lock_acquire retry logic (alongside EAGAIN), causing additional retry attempts before failing.

Sequence Diagram(s)

sequenceDiagram
    participant Tail as Tail Worker
    participant Lock as DB Lock (tail_db_lock / tail_db_unlock)
    participant DB as SQLite DB

    Tail->>Lock: request lock (tail_db_lock)
    Lock-->>Tail: lock acquired
    Tail->>DB: perform DB operation (open/set/offset/rotate/delete)
    DB-->>Tail: operation result (ok / error)
    Tail->>Lock: release lock (tail_db_unlock)
    Lock-->>Tail: lock released

Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related issues

• in_tail fails silently when SQLite DB is locked (offset update failure) #11416: Changes add explicit locking and unlock-on-error around flb_tail_db_file_offset and related DB ops, which align with fixing offset-read/consistency problems described in the issue.

Suggested labels

backport to v4.0.x

Suggested reviewers

• edsiper
• koleini
• fujimotos

Poem

🐇 I nibble bytes and guard the rows,
I clutch the lock where sqlite grows,
No skipped offsets, no tangled thread,
I hop, unlock, and keep logs fed,
Tiny paws, big syncs — onward we go! 🥕

🚥 Pre-merge checks | ✅ 5 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 62.50% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main objective: adding serialization of database operations in in_tail to prevent resource collisions in threaded mode, which directly addresses the linked issue #11357.
Linked Issues check	✅ Passed	The changes implement serialization via inline locking helpers (tail_db_lock/tail_db_unlock) and add locking around all DB operations, directly addressing issue #11357's requirement to maintain file position tracking and prevent re-reading in threaded mode.
Out of Scope Changes check	✅ Passed	All changes are scoped to addressing the linked issue: DB operation serialization in in_tail plugin and EBUSY error handling in flb_lock are both necessary for fixing threaded mode file position tracking.
Merge Conflict Detection	✅ Passed	✅ No merge conflicts detected when merging into master

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch cosmo0920-serialize-db-operation-on-on_tail

---

No actionable comments were generated in the recent review. 🎉

🧹 Recent nitpick comments

src/flb_lock.c (1)

76-93: Consider: flb_lock_release retry on EAGAIN may be dead code for non-recursive mutexes.

pthread_mutex_unlock doesn't return EAGAIN or EBUSY for default (non-recursive) mutexes — it returns EPERM on ownership errors. The retry loop here would never actually retry; any error would hit the break. This is pre-existing and not introduced by this PR, but worth noting if this code is being actively maintained.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c2e82e062c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".