Repository navigation
Conversation
…s inaccessible When deleting encrypted (LUKS) resources, the LUKS layer may close its device before DRBD layer attempts to adjust the resource or regenerate the res file. This causes 'Failed to adjust DRBD resource' errors. This fix adds checks before regenerateResFile() and drbdUtils.adjust() to verify that child layer devices are accessible. If a child device doesn't exist or is not accessible, these operations are skipped and adjustRequired flag is cleared, allowing resource deletion to proceed successfully. Fixes issue where encrypted resources could not be deleted due to DelayedApiRcException during DRBD resource adjustment.
Add physical device path existence check before calling lsblk in updateDiscGran(). This prevents race condition when drbdadm adjust temporarily brings devices down/up and the kernel hasn't created the device node yet. Issue: After satellite restart with patched code, some DRBD resources ended up in Unknown state because: 1. drbdadm adjust successfully completes (brings devices up) 2. updateDiscGran() immediately tries to check discard granularity 3. /dev/drbd* device node doesn't exist yet (kernel hasn't created it) 4. lsblk fails with exit code 32 "not a block device" 5. StorageException interrupts DeviceManager cycle 6. DRBD device remains in incomplete state Solution: Check Files.exists(devicePath) before calling lsblk. If device doesn't exist yet, skip the check - it will be retried in the next DeviceManager cycle when the device node is available. Co-Authored-By: Claude <[email protected]> Signed-off-by: Andrei Kvapil <[email protected]>
The previous implementation blocked `drbdadm adjust` whenever child device paths were unavailable, even for operations that don't require disk access (like network reconnect from StandAlone to Connected). After node reboot, DRBD resources often remain in StandAlone state because: 1. updateResourceToCurrentDrbdState() correctly detects StandAlone and sets adjustRequired=true 2. However, canAdjust check fails because child volumes may have devicePath=null (due to INACTIVE flag, cloning state, or initialization race) 3. adjust is skipped → adjustRequired=false → resources stay StandAlone Root cause: The canAdjust check was added to protect LUKS deletion scenarios but was applied to ALL cases, including network reconnect which doesn't need disk access. Fix: Check child device accessibility only when disk access is actually required (volume creation, resize, metadata operations). Network reconnect operations (StandAlone → Connected) now proceed without checking child devices. This ensures automatic DRBD reconnection after reboot while preserving protection for LUKS deletion scenarios. Co-Authored-By: Claude <[email protected]> Signed-off-by: Andrei Kvapil <[email protected]>
kvaps
force-pushed
the
fix/encrypted-resource-deletion
branch
2 times, most recently
from
January 12, 2026 13:40
16c8a8e to
f0eaa4c
Compare
kvaps
added a commit
to cozystack/cozystack
that referenced
this pull request
Jan 12, 2026
## What this PR does This PR updates piraeus-server patches to address several critical production issues with DRBD resources and LUKS encryption: 1. **Add fix-duplicate-tcp-ports.diff** - Prevents duplicate TCP ports after toggle-disk operations (upstream PR #476) 2. **Update skip-adjust-when-device-inaccessible.diff** - Comprehensive fix for multiple issues: - Resources stuck in StandAlone state after node reboot - Unknown state race condition during satellite restart - Encrypted LUKS resource deletion failures - Network reconnect blocked by unavailable child device checks These patches resolve scenarios where DRBD resources fail to automatically reconnect after node reboots and improve LUKS resource lifecycle management. Upstream PRs: - LINBIT/linstor-server#476 - LINBIT/linstor-server#477 ### Release note ```release-note [linstor] Fix DRBD resources stuck in StandAlone state after reboot and encrypted resource deletion issues ``` <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Prevents duplicate TCP port conflicts after disk toggle operations * Fixes resources stuck in StandAlone or Unknown state after reboot * Resolves issues with encrypted resource deletion * Improves handling of temporarily inaccessible storage devices <sub>✏️ Tip: You can customize this high-level summary in your review settings.</sub> <!-- end of auto-generated comment: release notes by coderabbit.ai -->
This was referenced Jan 14, 2026
LuksLayer was the only layer in LINSTOR that never called setExists(true) after successfully creating and opening a LUKS device. All other layers (Storage, DRBD, Writecache, Cache, BCache) properly set this flag. This caused a critical issue when combined with the child device accessibility check in DrbdLayer.adjustDrbd(): the check "!childVlm.exists()" always evaluated to true for LUKS volumes, preventing .res file regeneration for new DRBD+LUKS+STORAGE resources. As a result, drbdadm create-md failed with "not defined in your config" on all nodes, leaving resources in Unknown state. The fix aligns LuksLayer with every other layer by calling setExists(true) alongside the existing setOpened(true) call after successful device creation/opening. Co-Authored-By: Claude <[email protected]> Signed-off-by: Andrei Kvapil <[email protected]>
Contributor
Author
|
Added a new commit that fixes a critical regression introduced by the child device accessibility check. Root cause: Symptoms: Every newly created resource with the DRBD+LUKS+STORAGE layer stack fails with: Typically 1 out of N nodes succeeds (via the Fix: Added The |
kvaps
added a commit
to cozystack/cozystack
that referenced
this pull request
Feb 19, 2026
## What this PR does Adds the `skip-adjust-when-device-inaccessible.diff` patch (upstream [LINBIT/linstor-server#477](LINBIT/linstor-server#477)) which: - Skips DRBD adjust and .res file regeneration when child layer devices are inaccessible (fixes encrypted resource deletion) - Skips lsblk when device path doesn't physically exist yet (fixes race condition after drbdadm adjust) - Only checks child devices when disk access is actually needed (allows network reconnect from StandAlone) - Fixes missing `setExists(true)` in `LuksLayer` — the root cause of all new DRBD+LUKS+STORAGE resources failing with "not defined in your config" ### Release note ```release-note [linstor] Fix DRBD+LUKS+STORAGE resource creation: all new encrypted volumes were failing because the DRBD .res file was never written due to a missing exists flag in the LUKS layer ``` <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved handling of inaccessible storage devices by adding pre-checks before performing operations * Operations are now skipped when underlying storage devices are unavailable, preventing unnecessary failures * Enhanced error recovery during storage adjustments when devices are temporarily inaccessible <!-- end of auto-generated comment: release notes by coderabbit.ai -->
Contributor
Author
|
Decided not to use these changes for now, closing. Will reopen if the issues arise again. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
fixes #350 (comment)
Problem
This PR fixes three related issues with device availability checks in DRBD layer:
1. Resources stuck in StandAlone after node reboot
After node reboot, DRBD resources remain in StandAlone state instead of automatically reconnecting because child device availability checks block network reconnection operations that don't require disk access.
2. Resources end up in Unknown state after satellite restart
Race condition when
drbdadm adjusttemporarily brings devices down/up: the kernel hasn't created/dev/drbd*device nodes yet, butupdateDiscGran()immediately tries to runlsblk, which fails and interrupts the DeviceManager cycle.3. Encrypted resource deletion fails
When deleting LUKS-encrypted resources, the LUKS layer may close its device before the DRBD layer attempts to adjust or regenerate the res file, causing 'Failed to adjust DRBD resource' errors.
Solution
Commit 1: Skip DRBD operations when child devices are inaccessible during deletion
Add checks before
regenerateResFile()anddrbdUtils.adjust()to verify child layer devices are accessible. If devices don't exist (e.g., LUKS closed during deletion), skip operations and clearadjustRequiredflag.Commit 2: Skip lsblk when device path doesn't physically exist
Check
Files.exists(devicePath)before callinglsblkinupdateDiscGran(). If device node doesn't exist yet, skip the check - it will retry in the next DeviceManager cycle.Commit 3: Only check child devices when disk access is needed
Distinguish between operations requiring disk access (volume creation, resize, metadata) and those that don't (network reconnect from StandAlone). Only check child device accessibility for disk operations, allowing automatic reconnection after reboot.
Testing
drbdadm adjustrace conditions