vrrp: restore multicast membership on interface up #2681
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
When an interface goes down and back up (e.g., network driver restart), the kernel drops multicast group membership. However, keepalived doesn't re-add the membership because the VRRP sockets remain open (fd_in != -1). This causes VRRP instances to stop receiving advertisements and stay stuck in BACKUP state indefinitely, even though the interface is back up. Fix by closing and reopening VRRP sockets in interface_up() to restore multicast group membership. This mirrors the socket handling logic in cleanup_lost_interface() but triggers on interface state change rather than interface deletion. Fixes: acassen#1847 Co-Authored-By: Claude <[email protected]> Signed-off-by: Aleksei Sviridkin <[email protected]>
This was referenced Dec 21, 2025
lexfrei
added a commit
to lexfrei/charts
that referenced
this pull request
Dec 21, 2025
…161) # Pull Request ## Description Add fully customizable probe configuration and trackInterface option for the vipalived chart. **Key features:** - Configurable `livenessProbe.exec.command` (default: `pgrep keepalived`, overridable) - New `startupProbe` support for initial VIP acquisition scenarios - New `readinessProbe` support - `enabled` flag for all probes (livenessProbe enabled by default) - `trackInterface` option as workaround for keepalived interface recreation bug ## Type of change - [ ] Bug fix (non-breaking change which fixes an issue) - [x] New feature (non-breaking change which adds functionality) - [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected) - [ ] Documentation update - [ ] Chart version bump ## Checklist ### Required for all PRs - [x] I have tested these changes locally - [x] My code follows the style guidelines of this project - [x] I have performed a self-review of my own code ### Required for chart changes - [x] Chart version has been bumped in `Chart.yaml` - [x] **`Chart.yaml` annotations updated with changelog entries** (`artifacthub.io/changes`) - [x] `values.schema.json` has been updated (if new values were added) - [x] `README.md` has been updated (if new values were added) - [x] All new values have proper descriptions/comments - [x] Tests have been added/updated in `tests/` directory - [x] All tests pass locally (`helm unittest charts/<chart-name>`) - [ ] Schema validation passes (`check-jsonschema --schemafile charts/<chart-name>/values.schema.json charts/<chart-name>/values.yaml`) - [x] Helm lint passes (`helm lint charts/<chart-name>`) ### If adding new templates - [x] Templates follow Helm best practices - [x] Templates use proper indentation (`nindent` instead of `indent`) - [x] Conditional rendering uses `{{- if }}` to control whitespace ### If modifying existing functionality - [x] Changes are backward compatible OR breaking changes are documented - [x] Default values maintain existing behavior ## Additional context Closes #160 This implementation follows TDD approach with 12 new tests added (96 total). **Backward compatible:** Existing configurations work without changes. Default livenessProbe behavior is preserved (`pgrep keepalived`). **Related upstream fix:** acassen/keepalived#2681 --------- Signed-off-by: Aleksei Sviridkin <[email protected]> Co-authored-by: Claude <[email protected]>
lexfrei
added a commit
to lexfrei/k8s
that referenced
this pull request
Dec 21, 2025
- Upgrade vipalived chart from 0.5.2 to 0.6.0 - Add trackInterface workaround for keepalived issue 1847 - Configure startupProbe to verify MASTER state (60s timeout) - Configure livenessProbe to check VIP presence on interface This ensures automatic pod restart if keepalived gets stuck in BACKUP state after network interface recreation. Refs: lexfrei/charts#160, acassen/keepalived#2681 Co-Authored-By: Claude <[email protected]> Signed-off-by: Aleksei Sviridkin <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
When a network interface goes down and back up (e.g., network driver restart like Mellanox mcab), the kernel drops multicast group membership. However, keepalived doesn't re-add the membership because VRRP sockets remain open (
fd_in != -1).This causes VRRP instances to stop receiving advertisements and stay stuck in BACKUP state indefinitely, even though the interface is back up and functional.
Changes
interface_up()to close and reopen VRRP sockets when interface comes back upIP_ADD_MEMBERSHIP)cleanup_lost_interface()but triggers on interface state change rather than deletionTest plan
ip link set eth0 downip link set eth0 upip maddr showshould show 224.0.0.18Related
Fixes: #1847