DM-Multipath Configuration

This guide covers configuring Linux Device-Mapper Multipathing (multipathd) for redundant SAN paths, automatic failover, and I/O load balancing across dual-controller iSCSI storage arrays.

Prerequisites

# Install multipath-tools
apt-get install -y multipath-tools

# Load kernel modules
modprobe dm-multipath
modprobe dm-round-robin
modprobe dm-service-time

# Start multipathd
systemctl enable --now multipathd

Basic Configuration

multipath.conf

# Create multipath configuration
cat > /etc/multipath.conf <<EOF
defaults {
    user_friendly_names yes
    find_multipaths smart
    path_checker tur
    path_grouping_policy group_by_prio
    prio alua
    failback immediate
    no_path_retry queue
}

blacklist {
    devnode "^sd[a-z]"
    devnode "^hd[a-z]"
    devnode "^cciss.*"
}

devices {
    device {
        vendor "LIO-ORG"
        product "*"
        path_checker tur
        hardware_handler "1 alua"
        path_grouping_policy group_by_prio
        prio alua
        failback immediate
        no_path_retry 10
    }
}
EOF

# Restart multipathd
systemctl restart multipathd

ALUA Configuration

Configure ALUA on Target

# On iSCSI target (LIO)
targetcli /iscsi/iqn.2024-01.hu.pilab:iscsiman.target-01/tpg1 set attribute alua_support=1

# Create ALUA group
targetcli /iscsi/iqn.2024-01.hu.pilab:iscsiman.target-01/tpg1/alua create name=default_alua

# Set ALUA state
targetcli /iscsi/iqn.2024-01.hu.pilab:iscsiman.target-01/tpg1/alua/default_alua set alua_access_state=0

Configure ALUA on Initiator

# multipath.conf
devices {
    device {
        vendor "LIO-ORG"
        product "*"
        hardware_handler "1 alua"
        prio alua
        path_grouping_policy group_by_prio
    }
}

Path Policies

Round Robin

# multipath.conf
defaults {
    path_grouping_policy multibus
    path_selector "round-robin 0"
    rr_min_io_rq 100
}

Service Time

# multipath.conf
defaults {
    path_grouping_policy group_by_prio
    path_selector "service-time 0"
}

Queue Length

# multipath.conf
defaults {
    path_grouping_policy group_by_prio
    path_selector "queue-length 0"
}

Failover Configuration

Immediate Failover

# multipath.conf
defaults {
    failback immediate
}

Manual Failover

# multipath.conf
defaults {
    failback manual
}

Delayed Failover

# multipath.conf
defaults {
    failback 30  # 30 second delay
}

Path Checking

TUR (Test Unit Ready)

# multipath.conf
defaults {
    path_checker tur
}

Direct I/O

# multipath.conf
defaults {
    path_checker directio
}

Read Sector 0

# multipath.conf
defaults {
    path_checker readsector0
}

Monitoring

Check Multipath Status

# List multipath devices
multipath -ll

# Output example:
# mpatha (360050764008181c60000000000000001) dm-0 LIO-ORG,iscsiman.target-01
# size=100G features='0' hwhandler='1 alua' wp=rw
# |-+- policy='service-time 0' prio=50 status=active
# | `- 6:0:0:0 sdb 8:16 active ready running
# `-+- policy='service-time 0' prio=10 status=enabled
#   `- 7:0:0:0 sdc 8:32 active ready running

Check Path Status

# Check path status
multipath -l | grep -A10 "mpatha"

# Check individual paths
multipathd show paths

Check Performance

# Check I/O stats
iostat -x /dev/dm-0 1

# Check throughput
fio --name=multipath-test --ioengine=libaio --iodepth=16 \
  --rw=randread --bs=4k --direct=1 --size=1G \
  --filename=/dev/dm-0

Troubleshooting

Paths Not Detected

# Check iscsi sessions
iscsiadm -m session -P3

# Check multipathd status
systemctl status multipathd

# Check multipathd logs
journalctl -u multipathd -f

# Reconfigure multipath
multipathd reconfigure

Path Failures

# Check path status
multipath -ll | grep failed

# Check path checker
multipathd show paths | grep failed

# Check logs
journalctl -u multipathd -f | grep -i fail

Performance Issues

# Check path priorities
multipathd show paths

# Check path grouping
multipath -l | grep "policy="

# Check I/O scheduler
cat /sys/block/dm-0/queue/scheduler

Best Practices

1. Use ALUA

Always use ALUA for optimal path selection:

# multipath.conf
devices {
    device {
        hardware_handler "1 alua"
        prio alua
    }
}

2. Configure Failover

Configure appropriate failover policy:

# multipath.conf
defaults {
    failback immediate
    no_path_retry queue
}

3. Monitor Paths

Regularly monitor path status:

# Check path status
multipath -ll

# Set up monitoring
echo "*/5 * * * * root multipath -ll | grep failed" >> /etc/crontab

4. Test Failover

Regularly test failover scenarios:

# Disable path
echo offline > /sys/block/sdb/device/state

# Check multipath
multipath -ll

# Re-enable path
echo running > /sys/block/sdb/device/state