DOWNSTREAM: carry: OCPBUGS-103516: Replace bbolt with patched fork to remove MADV_RANDOM by hasbro17 · Pull Request #395 · openshift/etcd

hasbro17 · 2026-08-04T20:42:11Z

Summary

Testing-only PR — the actual downstream fix should land via openshift/bbolt once that fork is synced to upstream v1.4.3.

This replaces the bbolt dependency with a patched fork hasbro17/bbolt@remove-madv-random that removes the madvise(MADV_RANDOM) call from bbolt's mmap path.

Since Linux 6.4 (torvalds/linux@8788f678), MADV_RANDOM causes the kernel to aggressively evict mmap'd pages by
short-circuiting the second-chance LRU mechanism. This results in high major page fault rates on bbolt's database file, particularly during compaction which sees order-of-magnitude duration increases. On OCP
5.0 FIPS clusters (RHCOS10, kernel 6.12), this manifests as a ~30% WAL fdatasync P99 latency regression.

This is the same root cause as OCPBUGS-50521 (OCP 4.19), which was fixed kernel-side by reverting the behavior change in RHEL 9.6. The RHEL kernel team has stated this behavior will not be reverted in RHEL 10,
so the fix must be in bbolt.

Next steps

Validate with perfscale CI on RHCOS10 FIPS clusters
Sync openshift/bbolt to upstream v1.4.3 and apply the patch there
Update this PR's replace directive to point to openshift/bbolt (or open a new PR)
Push the fix upstream to etcd-io/bbolt (#939)

References

JIRA: OCPBUGS-103516
Prior instance: OCPBUGS-50521
Upstream issue: etcd-io/bbolt#939
Prior proposed fix: etcd-io/bbolt#940
bbolt patch: hasbro17/bbolt@remove-madv-random

Summary by CodeRabbit

Chores
- Updated underlying system and synchronization components to newer versions.
- Improved embedded database compatibility through an updated implementation mapping.
- No user-facing features or behavior changes.

…to remove MADV_RANDOM Since Linux 6.4, MADV_RANDOM causes the kernel to aggressively evict mmap'd pages, resulting in major page faults on bbolt's database file and order-of-magnitude compaction duration increases. The RHEL 10 kernel will not revert this behavior change, so the fix must be in bbolt. This adds a go.mod replace directive pointing to a patched bbolt fork with the MADV_RANDOM madvise hint removed. The replace should be dropped once the fix is merged upstream (etcd-io/bbolt#939). Ref: etcd-io/bbolt#939 Ref: https://redhat.atlassian.net/browse/OCPBUGS-103516 Assisted-by: Claude Code (Opus 4.6)

openshift-ci-robot · 2026-08-04T20:42:19Z

@hasbro17: This pull request references Jira Issue OCPBUGS-103516, which is invalid:

expected the bug to target the "5.0.0" version, but no target version was set

Comment /jira refresh to re-evaluate validity if changes to the Jira bug are made, or edit the title of this pull request to link to a different bug.

The bug has been updated to refer to the pull request using the external bug tracker.

Details

In response to this:

Summary

Testing-only PR — the actual downstream fix should land via openshift/bbolt once that fork is synced to upstream v1.4.3.

This replaces the bbolt dependency with a patched fork hasbro17/bbolt@remove-madv-random that removes the madvise(MADV_RANDOM) call from bbolt's mmap path.

Since Linux 6.4 (torvalds/linux@8788f678), MADV_RANDOM causes the kernel to aggressively evict mmap'd pages by
short-circuiting the second-chance LRU mechanism. This results in high major page fault rates on bbolt's database file, particularly during compaction which sees order-of-magnitude duration increases. On OCP
5.0 FIPS clusters (RHCOS10, kernel 6.12), this manifests as a ~30% WAL fdatasync P99 latency regression.

This is the same root cause as OCPBUGS-50521 (OCP 4.19), which was fixed kernel-side by reverting the behavior change in RHEL 9.6. The RHEL kernel team has stated this behavior will not be reverted in RHEL 10,
so the fix must be in bbolt.

Next steps

Validate with perfscale CI on RHCOS10 FIPS clusters

Sync openshift/bbolt to upstream v1.4.3 and apply the patch there

Update this PR's replace directive to point to openshift/bbolt (or open a new PR)

Push the fix upstream to etcd-io/bbolt (#939)

References

JIRA: OCPBUGS-103516

Prior instance: OCPBUGS-50521

Upstream issue: etcd-io/bbolt#939

Prior proposed fix: etcd-io/bbolt#940

bbolt patch: hasbro17/bbolt@remove-madv-random

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

coderabbitai · 2026-08-04T20:42:31Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository: openshift/coderabbit/.coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 98020688-63df-43e2-90cf-165b30921132

📥 Commits

Reviewing files that changed from the base of the PR and between 64f8851 and 76de772.

⛔ Files ignored due to path filters (4)

etcdutl/go.sum is excluded by !**/*.sum
go.sum is excluded by !**/*.sum
server/go.sum is excluded by !**/*.sum
tests/go.sum is excluded by !**/*.sum

📒 Files selected for processing (4)

etcdutl/go.mod
go.mod
server/go.mod
tests/go.mod

Walkthrough

The pull request updates golang.org/x/sys and golang.org/x/sync versions in Go module files. It also adds a pinned replacement for go.etcd.io/bbolt in each module.

Changes

Module dependency alignment

Layer / File(s)	Summary
Module requirements and bbolt replacements `go.mod`, `etcdutl/go.mod`, `server/go.mod`, `tests/go.mod`	The modules update `golang.org/x/sys` to `v0.47.0`. `tests/go.mod` updates `golang.org/x/sync` to `v0.22.0`. Each module maps `go.etcd.io/bbolt` to the pinned `github.com/hasbro17/bbolt` revision.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Suggested reviewers: ivanvc, ar21sm

🚥 Pre-merge checks | ✅ 15

✅ Passed checks (15 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the downstream change: replacing bbolt with a patched fork to remove MADV_RANDOM.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Stable And Deterministic Test Names	✅ Passed	The commit changes only Go module metadata and checksums; no Go test files or Ginkgo test titles were added or modified.
Test Structure And Quality	✅ Passed	The PR changes only Go module manifests and sums; no Ginkgo test code, setup, waits, or assertions changed, so this check is not applicable.
Microshift Test Compatibility	✅ Passed	The PR changes only eight go.mod/go.sum files; the parent-to-HEAD diff adds no Ginkgo tests or MicroShift-incompatible API references.
Single Node Openshift (Sno) Test Compatibility	✅ Passed	The pull request changes only eight Go module and checksum files; it adds no Ginkgo e2e tests or other test declarations requiring SNO compatibility review.
Topology-Aware Scheduling Compatibility	✅ Passed	The PR changes only go.mod and go.sum files; no deployment manifests, operator code, controllers, or scheduling constraints were added or modified.
Ote Binary Stdout Contract	✅ Passed	The PR changes only go.mod/go.sum dependency metadata; it adds no Go source or process-level stdout writes.
Ipv6 And Disconnected Network Test Compatibility	✅ Passed	The patch changes only four go.mod and four go.sum files; it adds no Go source, Ginkgo tests, or network operations.
No-Weak-Crypto	✅ Passed	The commit changes only Go module metadata; it adds no Go source, weak-crypto API, custom crypto, or secret comparison, and the added bbolt replacement is non-cryptographic.
Container-Privileges	✅ Passed	The PR changes only Go module and checksum files; the patch contains no privileged, host namespace, SYS_ADMIN, root, or allowPrivilegeEscalation settings.
No-Sensitive-Data-In-Logs	✅ Passed	The commit changes only go.mod/go.sum dependency metadata; no logging code or sensitive-data values were added.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

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

openshift-ci · 2026-08-04T20:42:35Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hasbro17

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~OWNERS~~ [hasbro17]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

hasbro17 · 2026-08-04T20:43:04Z

/jira refresh

openshift-ci-robot · 2026-08-04T20:43:11Z

@hasbro17: This pull request references Jira Issue OCPBUGS-103516, which is valid.

3 validation(s) were run on this bug

bug is open, matching expected state (open)
bug target version (5.0.0) matches configured target version for branch (5.0.0)
bug is in the state POST, which is one of the valid states (NEW, ASSIGNED, POST)

Details

In response to this:

Summary

Testing-only PR — the actual downstream fix should land via openshift/bbolt once that fork is synced to upstream v1.4.3.

This replaces the bbolt dependency with a patched fork hasbro17/bbolt@remove-madv-random that removes the madvise(MADV_RANDOM) call from bbolt's mmap path.

Since Linux 6.4 (torvalds/linux@8788f678), MADV_RANDOM causes the kernel to aggressively evict mmap'd pages by
short-circuiting the second-chance LRU mechanism. This results in high major page fault rates on bbolt's database file, particularly during compaction which sees order-of-magnitude duration increases. On OCP
5.0 FIPS clusters (RHCOS10, kernel 6.12), this manifests as a ~30% WAL fdatasync P99 latency regression.

This is the same root cause as OCPBUGS-50521 (OCP 4.19), which was fixed kernel-side by reverting the behavior change in RHEL 9.6. The RHEL kernel team has stated this behavior will not be reverted in RHEL 10,
so the fix must be in bbolt.

Next steps

Validate with perfscale CI on RHCOS10 FIPS clusters

Sync openshift/bbolt to upstream v1.4.3 and apply the patch there

Update this PR's replace directive to point to openshift/bbolt (or open a new PR)

Push the fix upstream to etcd-io/bbolt (#939)

References

JIRA: OCPBUGS-103516

Prior instance: OCPBUGS-50521

Upstream issue: etcd-io/bbolt#939

Prior proposed fix: etcd-io/bbolt#940

bbolt patch: hasbro17/bbolt@remove-madv-random

Summary by CodeRabbit

Chores

Updated underlying system and synchronization components to newer versions.

Improved embedded database compatibility through an updated implementation mapping.

No user-facing features or behavior changes.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

openshift-ci-robot · 2026-08-04T20:43:11Z

@hasbro17: This pull request references Jira Issue OCPBUGS-103516, which is valid. The bug has been moved to the POST state.

3 validation(s) were run on this bug

bug is open, matching expected state (open)
bug target version (5.0.0) matches configured target version for branch (5.0.0)
bug is in the state New, which is one of the valid states (NEW, ASSIGNED, POST)

Details

In response to this:

/jira refresh

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

afcollins · 2026-08-05T00:57:37Z

/pj-rehearse pull-ci-openshift-etcd-main-perfscale-control-plane-6nodes

afcollins · 2026-08-05T00:58:29Z

/test perfscale-control-plane-6nodes

openshift-ci · 2026-08-05T04:04:02Z

@hasbro17: The following tests failed, say /retest to rerun all failed tests or /retest-required to rerun all mandatory failed tests:

Test name	Commit	Details	Required	Rerun command
ci/prow/upstream-e2e	`76de772`	link	false	`/test upstream-e2e`
ci/prow/perfscale-control-plane-6nodes	`76de772`	link	false	`/test perfscale-control-plane-6nodes`
ci/prow/upstream-integration	`76de772`	link	false	`/test upstream-integration`

Full PR test history. Your PR dashboard.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here.

mcornea · 2026-08-05T08:18:43Z

/payload-job periodic-ci-openshift-eng-ocp-perfscale-main-aws-5.0-nightly-x86-control-plane-fips-24nodes

openshift-ci · 2026-08-05T08:19:27Z

@mcornea: trigger 1 job(s) for the /payload-(with-prs|job|aggregate|job-with-prs|aggregate-with-prs) command

periodic-ci-openshift-eng-ocp-perfscale-main-aws-5.0-nightly-x86-control-plane-fips-24nodes

See details on https://pr-payload-tests.ci.openshift.org/runs/ci/45e4cef0-90a6-11f1-84e2-697c00fba475-0

openshift-ci Bot requested review from deads2k and dusk125 August 4, 2026 20:42

openshift-ci Bot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Aug 4, 2026

openshift-ci-robot added jira/valid-bug Indicates that a referenced Jira bug is valid for the branch this PR is targeting. and removed jira/invalid-bug Indicates that a referenced Jira bug is invalid for the branch this PR is targeting. labels Aug 4, 2026

Uh oh!

Conversation

hasbro17 commented Aug 4, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Next steps

References

Summary by CodeRabbit

Uh oh!

openshift-ci-robot commented Aug 4, 2026

Summary

Next steps

References

Uh oh!

coderabbitai Bot commented Aug 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

openshift-ci Bot commented Aug 4, 2026

Uh oh!

hasbro17 commented Aug 4, 2026

Uh oh!

openshift-ci-robot commented Aug 4, 2026

Summary

Next steps

References

Summary by CodeRabbit

Uh oh!

openshift-ci-robot commented Aug 4, 2026

Uh oh!

afcollins commented Aug 5, 2026

Uh oh!

afcollins commented Aug 5, 2026

Uh oh!

openshift-ci Bot commented Aug 5, 2026

Uh oh!

mcornea commented Aug 5, 2026

Uh oh!

openshift-ci Bot commented Aug 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

hasbro17 commented Aug 4, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Aug 4, 2026 •

edited

Loading