ROMFASTSQL/oracle/standby-server-scripts/DR_VM_MIGRATION_GUIDE.md

Oracle DR VM - Migration Between Proxmox Nodes

Purpose: How to migrate VM 109 between Proxmox nodes while maintaining backup access Scenario: Move VM from pveelite (10.0.20.202) to pvemini (10.0.20.201) or vice versa

---

📋 OVERVIEW

Current Setup:

• VM 109 runs on pveelite (10.0.20.202)
• Backups stored on pveelite: /mnt/pve/oracle-backups
• VM has mount point: qm set 109 -mp0 /mnt/pve/oracle-backups
• Mount appears in Windows as *F:* (E:\ already used)

Challenge:

• Mount points are node-local - path /mnt/pve/oracle-backups exists only on pveelite
• If you migrate VM to pvemini, mount point breaks

Solution:

• Create same directory structure on destination node
• Sync backups between nodes
• Mount point works identically on new node

---

🔄 MIGRATION PROCEDURE

PRE-MIGRATION CHECKLIST

• VM 109 is powered OFF
• You have root SSH access to both Proxmox nodes
• You know which node you're migrating TO
• Backups are current (check timestamp)

---

STEP 1: Prepare Destination Node (pvemini)

On pvemini (10.0.20.201):

ssh root@10.0.20.201

# Create identical directory structure
mkdir -p /mnt/pve/oracle-backups/ROA/autobackup
chmod 755 /mnt/pve/oracle-backups
chmod 755 /mnt/pve/oracle-backups/ROA
chmod 755 /mnt/pve/oracle-backups/ROA/autobackup

# Verify structure
ls -la /mnt/pve/oracle-backups/ROA/autobackup

---

STEP 2: Sync Backups from Source to Destination

Option A: Full Sync (first time migration)

# On pvemini, sync all backups from pveelite
rsync -avz --progress \
    root@10.0.20.202:/mnt/pve/oracle-backups/ \
    /mnt/pve/oracle-backups/

# This copies all backup files (~15 GB, takes 2-3 minutes on 1Gbps network)

Option B: Incremental Sync (if you already synced before)

# On pvemini, sync only new/changed files
rsync -avz --progress --update \
    root@10.0.20.202:/mnt/pve/oracle-backups/ \
    /mnt/pve/oracle-backups/

# Much faster - only copies new backups

Verify sync:

# Check file count matches
ssh root@10.0.20.202 "ls /mnt/pve/oracle-backups/ROA/autobackup/*.bkp | wc -l"
ls /mnt/pve/oracle-backups/ROA/autobackup/*.bkp | wc -l

# Should be same number

---

STEP 3: Migrate VM via Proxmox

Option A: Online Migration (VM stays running)

# From Proxmox CLI on source node (pveelite):
qm migrate 109 pvemini --online

# This uses live migration - VM doesn't stop
# Takes 5-10 minutes depending on RAM/disk

Option B: Offline Migration (VM must be stopped)

# Stop VM first
qm stop 109

# Migrate
qm migrate 109 pvemini

# Faster than online, but requires downtime

Option C: Via Proxmox Web UI

1. Select VM 109 on pveelite
2. Click "Migrate"
3. Select target node: pvemini
4. Choose migration type: online or offline
5. Click "Migrate"

---

STEP 4: Verify Mount Point After Migration

After migration completes:

# On pvemini, check VM config includes mount point
qm config 109 | grep mp0

# Expected output:
# mp0: /mnt/pve/oracle-backups,mp=/mnt/oracle-backups

# If missing, add it:
qm set 109 -mp0 /mnt/pve/oracle-backups,mp=/mnt/oracle-backups

---

STEP 5: Start VM and Verify Access

# Start VM on new node
qm start 109

# Wait for boot
sleep 180

# Check mount in Windows
ssh -p 22122 romfast@10.0.20.37 "Get-PSDrive F"

# Should show F:\ with Used/Free space

# Verify backup files accessible
ssh -p 22122 romfast@10.0.20.37 "Get-ChildItem F:\ROA\autobackup\*.bkp | Measure-Object"

# Should show backup file count

---

STEP 6: Update PRIMARY Transfer Scripts

On PRIMARY (10.0.20.36):

Backup transfer scripts need to know which node to send to.

Option A: Update scripts to point to new node

# Edit transfer scripts
cd D:\rman_backup

# Find and replace in transfer scripts:
# ÎNAINTE:
$DRHost = "10.0.20.202"  # pveelite

# DUPĂ:
$DRHost = "10.0.20.201"  # pvemini

Option B: Use DNS/hostname (RECOMMENDED)

# In transfer scripts, use hostname instead of IP:
$DRHost = "pvedr"  # DNS name

# Then update DNS to point to active node:
# pvedr → 10.0.20.201 (currently pvemini)
# When you migrate back, just update DNS

---

🔄 ONGOING SYNC STRATEGY

If VM Stays on New Node Long-Term

Setup automated sync from PRIMARY → new node:

Just update transfer scripts as in Step 6 above. Backups will now go directly to pvemini.

Old backups on pveelite:

• Can be deleted after verification
• Or kept as additional backup copy (recommended)

# On pveelite, cleanup old backups after 7 days
find /mnt/pve/oracle-backups/ROA/autobackup -name "*.bkp" -mtime +7 -delete

---

If You Migrate VM Back and Forth

Scenario: VM moves between nodes frequently

Solution 1: Sync in both directions

# Cronjob on pveelite (every 6 hours)
0 */6 * * * rsync -az root@10.0.20.201:/mnt/pve/oracle-backups/ /mnt/pve/oracle-backups/

# Cronjob on pvemini (every 6 hours)
0 */6 * * * rsync -az root@10.0.20.202:/mnt/pve/oracle-backups/ /mnt/pve/oracle-backups/

Solution 2: Shared Storage (NFS/CIFS)

Use Proxmox shared storage instead of local paths:

• Setup NFS server on one node
• Both nodes mount same NFS share
• /mnt/pve/oracle-backups points to shared storage
• VM migration doesn't require backup sync

---

📊 MIGRATION CHECKLIST

Before Migration:

• VM 109 is stopped (or prepared for online migration)
• Destination node has directory: /mnt/pve/oracle-backups/ROA/autobackup
• Backups synced to destination node (rsync completed)
• You have tested restore recently (weekly test passed)

During Migration:

• VM migration initiated (online or offline)
• Migration progress monitored (no errors)
• Migration completed successfully

After Migration:

• VM 109 shows as running on new node
• Mount point configured: qm config 109 | grep mp0
• VM started successfully
• F:\ drive accessible in Windows: Get-PSDrive F
• Backup files visible: Get-ChildItem F:\ROA\autobackup\*.bkp
• PRIMARY transfer scripts updated (point to new node IP)
• Test restore completed successfully

---

⚠️ TROUBLESHOOTING

Mount Point Not Visible in VM After Migration

Symptom: F:\ drive missing in Windows after migration

Solution:

# On new node, verify mount point config
qm config 109 | grep mp0

# If missing, add it
qm set 109 -mp0 /mnt/pve/oracle-backups,mp=/mnt/oracle-backups

# Restart VM
qm stop 109
qm start 109

Backup Files Not Accessible

Symptom: F:\ exists but shows as empty

Cause: Backups not synced to new node

Solution:

# Re-sync backups from old node
rsync -avz root@10.0.20.202:/mnt/pve/oracle-backups/ /mnt/pve/oracle-backups/

# Verify files exist
ls -lh /mnt/pve/oracle-backups/ROA/autobackup/*.bkp

PRIMARY Still Sending to Old Node

Symptom: New backups not appearing on new node

Cause: Transfer scripts still point to old node IP

Solution: Update $DRHost in transfer scripts on PRIMARY (see Step 6)

---

🎯 MIGRATION TIMELINE

Task	Duration	Downtime
Prepare destination node	5 min	None
Sync backups (full, ~15GB)	3 min	None
Migrate VM (offline)	5 min	5 min
Verify and start VM	3 min	3 min
Update PRIMARY scripts	2 min	None
Total	18 min	8 min

With online migration: 0 minutes downtime (VM keeps running during migration)

---

📞 QUICK REFERENCE

Current Setup:

• Source node: pveelite (10.0.20.202)
• Destination node: pvemini (10.0.20.201)
• VM: 109 (oracle-dr-windows)
• Backup path: /mnt/pve/oracle-backups
• Windows mount: F:\ (not E:\ - already used)

Key Commands:

# Sync backups
rsync -avz root@SOURCE:/mnt/pve/oracle-backups/ /mnt/pve/oracle-backups/

# Migrate VM
qm migrate 109 DESTINATION --online

# Check mount
qm config 109 | grep mp0

# Add mount if missing
qm set 109 -mp0 /mnt/pve/oracle-backups,mp=/mnt/oracle-backups

---

Generated: 2025-10-09 Version: 1.0 Status: Ready for use See Also: DR_UPGRADE_TO_CUMULATIVE_PLAN.md