beatzaplenty/nixos
Archived
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(ha/cluster-init): use targetcli delete commands for LIO teardown

Browse Source 
clearconfig does not reliably clear kernel LIO configfs state — the kernel
still holds backing-file references, blocking umount.  Replace clearconfig
with explicit targetcli delete commands (/iscsi delete, /backstores/fileio
delete) which do release kernel state.  Also make DRBD secondary demote
idempotent (skip if already Secondary).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HaH1cSGvhogRP5ExoF6nD8
This commit is contained in:
beatzaplenty
2026-07-28 17:55:26 +10:00
co-authored by Claude Sonnet 4.6
parent d31d9fa584
commit 79cde50e27
1 changed files with 12 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/ha/cluster-init.sh
+12 -5
View File
@@ -186,9 +186,14 @@ log "Configuring LIO iSCSI target via targetcli..."
# VIP resource will make the target reachable at the VIP address. # VIP resource will make the target reachable at the VIP address.
# #
# Clear any existing LIO state first (idempotent: re-run after a partial failure). # Clear any existing LIO state first (idempotent: re-run after a partial failure).
if ls /sys/kernel/config/target/iscsi/ 2>/dev/null | grep -q iqn; then # Use specific delete commands — clearconfig does not reliably clear kernel state.
log "Clearing existing LIO targets before reconfiguration..." if ls /sys/kernel/config/target/iscsi/ 2>/dev/null | grep -q "${ISCSI_IQN}"; then
echo "clearconfig confirm=yes" | targetcli 2>/dev/null || true log "Clearing existing LIO target ${ISCSI_IQN} before reconfiguration..."
targetcli "/iscsi delete ${ISCSI_IQN}" 2>/dev/null || true
fi
if ls /sys/kernel/config/target/core/ 2>/dev/null | grep -q "fileio"; then
log "Clearing existing LIO backstore ha-lun0 before reconfiguration..."
targetcli "/backstores/fileio delete ha-lun0" 2>/dev/null || true
fi fi
targetcli <<EOF targetcli <<EOF
/backstores/fileio create name=ha-lun0 file_or_dev=${ISCSI_LUN_FILE} size=0 write_back=false /backstores/fileio create name=ha-lun0 file_or_dev=${ISCSI_LUN_FILE} size=0 write_back=false
@@ -201,7 +206,9 @@ EOF
log "Tearing down LIO kernel objects — Pacemaker will restore via targetctl on the Active node..." log "Tearing down LIO kernel objects — Pacemaker will restore via targetctl on the Active node..."
# LIO holds the backing file open; clear kernel state now so the XFS unmount succeeds. # LIO holds the backing file open; clear kernel state now so the XFS unmount succeeds.
echo "clearconfig confirm=yes" | targetcli 2>/dev/null || warn "LIO clearconfig had errors — umount may fail" # Use specific delete commands (clearconfig does not reliably clear kernel configfs state).
targetcli "/iscsi delete ${ISCSI_IQN}" 2>/dev/null || warn "LIO iscsi delete failed — umount may fail"
targetcli "/backstores/fileio delete ha-lun0" 2>/dev/null || warn "LIO backstore delete failed"
log "Distributing iSCSI saveconfig to $NODE2..." log "Distributing iSCSI saveconfig to $NODE2..."
n2_scp /etc/target/saveconfig.json /etc/target/saveconfig.json n2_scp /etc/target/saveconfig.json /etc/target/saveconfig.json
@@ -210,7 +217,7 @@ log "Unmounting ${XFS_MOUNT} — Pacemaker manages it..."
umount "${XFS_MOUNT}" || { sync; umount -l "${XFS_MOUNT}"; } umount "${XFS_MOUNT}" || { sync; umount -l "${XFS_MOUNT}"; }
log "Demoting DRBD to Secondary — Pacemaker manages primary role..." log "Demoting DRBD to Secondary — Pacemaker manages primary role..."
drbdadm secondary ha-data [[ "$(drbdadm role ha-data 2>/dev/null)" == "Primary/Secondary" ]] && drbdadm secondary ha-data || true
# ── 7. Pacemaker resources ──────────────────────────────────────────────── # ── 7. Pacemaker resources ────────────────────────────────────────────────
log "Configuring Pacemaker cluster properties..." log "Configuring Pacemaker cluster properties..."