Skip to content



Create Network Interface Backup (NIB) adapter

mkvdev -lnagg entX -attr backup_adapter=entY
mkvdev -lnagg entX entY -attr mode=8023ad hash_mode=src_dst_port use_jumbo_frames=yes

Create Shared Ethernet Adapter (SEA)

mkvdev -sea entX -vadapter entX,entY -default entX -defaultid XXX -attr ha_mode=sharing
chdev -dev entX -perm -attr jumbo_frames=yes large_receive=yes largesend=1
chdev -dev entX -perm -attr  accounting=enabled

Check SEA mapping

lsmap -all -net

List active VLAN's being sent over the SEA

SEA en* device needs to be up in ifconfig, but doesn't need an IP address

tcpdump -i en19 -s 1500 -nn -e vlan

Manually change SEA failover mode

chdev -dev entX -attr ha_mode=<auto|sharing|standby>

Reset ethernet adapter statistics

entstat -reset entX

SEA statistics

Requires accounting to be enabled on adapter

seastat -d ent5 [-c] [-n search_criteria=value]
  • -c resets statistics
  • -n search on particular criteria
  • mac=<value>
  • ip=<value>
  • vlan=<value>
  • host=<value>

Preallocate virtual ethernet buffers

lsdev | awk '/Virtual I\/O Ethernet Adapter/{ print $1 }' | while read -r VETH; do chdev -l "${VETH}" -a max_buf_tiny=4096 -a min_buf_tiny=4096 -a max_buf_small=4096 -a min_buf_small=4096 -a max_buf_medium=2048 -a min_buf_medium=2048 -a max_buf_large=256 -a min_buf_large=256 -a max_buf_huge=128 -a min_buf_huge=128 -P; done


NPIV map client adapter to server FCS port

vfcmap -vadapter vfchost0 -fcp fcs0

To unassign the mapping, just leave the physical adapter off the command

vfcmap -vadapter vfchost0 -fcp

View NPIV mappings

Specific client adapter mapping

lsmap -vadapter vfchost0 -npiv

All client adapter mappings (CSV formatting)

lsmap -all -npiv -fmt ,

Show all client WWPN's from the VIOS

echo "svfcCI; svfcPrQs; svfcva" | kdb -i

Verbose adapter stats

echo "svfcCI; svfcPrQs; svfcva vfchost1" | kdb -i`

Replace FCS adapter being used with NPIV

1. Verify port(s) that are offline

The output below shows fcs5 isn't listed as NPIV capable. Even after a port down/up by the SAN team, the port still shows as offline.

$ lsnports
name             physloc                        fabric tports aports swwpns  awwpns
fcs0             U78CA.001.CSS134Y-P1-C2-C1-T1       1     64     30   3088    2984
fcs1             U78CA.001.CSS134Y-P1-C2-C1-T2       1     64     29   3088    2981
fcs4             U78CA.001.CSS134Y-P1-C4-C1-T1       1     64     30   3088    2976
fcs6             U78CA.001.CSS112R-P1-C2-C1-T1       1     64     27   3088    2965
fcs7             U78CA.001.CSS112R-P1-C2-C1-T2       1     64     26   3088    2962
fcs10            U78CA.001.CSS112R-P1-C4-C1-T1       1     64     37   3088    3005
fcs11            U78CA.001.CSS112R-P1-C4-C1-T2       1     64     37   3088    3005

2. Put vfchosts into a Defined state

As these are two port adapters, fcs4 which is still functional has active NPIV maps, so they need to be put into a Defined state. The below command shows all the current mappings down fcs4 and fcs5. The mappings down fcs5 wont show if the port isn't available.

lsmap -all -npiv -fmt : | awk -F: '$7 ~ /fcs[45]/'

Loop through above vfchost adapters and put them into a Defined state.

lsmap -all -npiv -fmt : | awk -F: '$7 ~ /fcs[45]/{ print $1 } | while read -r VFC; do rmdev -dev ${VFC} -ucfg; done

3. Put the physical ports of the card being replaced into a Defined state

rmdev -dev fcs4 -ucfg -recursive
rmdev -dev fcs5 -ucfg -recursive

4. Replace adapter and bring it back online

Once the card is verified as being back online, bring the devices back into an Available state. This should also put back all the above vfcmap's.


Virtual Media Library

Define a VML

$ mkrep -sp rootvg -size 10G
Virtual Media Repository Created
Repository created within "VMLibrary" logical volume

Add ISO image(s) to VML

# ls -l /var/vio/VMLibrary/AIX-6100-06-01-Disk1.iso
-r--------    1 root     system   4236148736 Apr 06 18:07 /var/vio/VMLibrary/AIX-6100-06-01-Disk1.iso
$ lsrep
Size(mb) Free(mb) Parent Pool         Parent Size      Parent Free
   10198     6158 rootvg                   279552           236288

Name                                    File Size Optical         Access
AIX-6100-06-01-Disk1.iso                     4040 None            ro

Create File Backed Optical Device for LPAR

$ mkvdev -fbo -vadapter vhost0
vtopt0 Available

Load image into the vtopt0 device

$ loadopt -vtd vtopt0 -disk AIX-6100-06-01-Disk1.iso
$ lsmap -vadapter vhost1
SVSA            Physloc                                      Client Partition ID
--------------- -------------------------------------------- ------------------
vhost0          U9133.55A.066EB4H-V1-C34                     0x00000000

VTD                   vtopt0
Status                Available
LUN                   0x8200000000000000
Backing device        /var/vio/VMLibrary/AIX-6100-06-01-Disk1.iso
Mirrored              N/A

Load a second disk

Can be used if you're prompted to insert disk 2 (for example)

loadopt -f -vtd vtopt0 -disk AIX-6100-06-01-Disk2.iso

Unload the image

unloadopt -vtd vtopt0

Shared Storage Pools

The following parameter are used when working with Shared Storage Pools (SSPs)

Parameter Description
clustername Name of the cluster
repopvs Disk which will contain the repository (Around 1GB LUN)
spname Name of the storage pool
sppvs Disks in the storage pool
hostname FQDN of the VIOS

Create cluster

cluster -create -clustername <name> -repopvs <hdiskX> -spname <sp_name> -sppvs <hdiskY> <hdiskX> -hostname <vios_hostname1> <vios_hostname2>

Remove the cluster

cluster -delete -clustername <name>

Add node to cluster

cluster -addnode -clustername <name> -hostname <vios_hostname3>

List Shared Storage Pools in a cluster

lssp -clustername clusterA

Allocate disk to VM

mkbdsp -clustername <name> -sp <sp_name> xGB -bd <device_name> -vadapter <vhost> [-thick]
  • <device_name> is the virtual disk name, and can be whatever you want

Remove disk from VM

rmbdsp -clustername <name> -sp <sp_name> -bd <device_name>
  • <device_name> The disk udid can be used here instead

Monitor disk pool

lssp -clustername <name> -sp <sp_name> -bd
lssp -clustername <name>

Add more LUN's to a pool

chsp -add -clustername <name> -sp <sp_name> <hdiskX> <hdiskY>

List nodes in a cluster

cluster -listnode -clustername <cluster_name>

Clean disk of all cluster information

Repository disk

cleandisk -r <hdisk>

Storage Pool Disk

cleandisk -f -s <hdisk>

List cluster storage interfaces

lscluster -d

Storage Pool threshold alerts

alert -list -clustername <cluster_name> -spname <sp_name>
alert -set -clustername <cluster_name> -spname <sp_name> -type threshold -value <value>


Setup PS1 prompt

Append the follow in /usr/ios/cli/.profile

export ENV=/home/padmin/.kshrc

Append the following in /home/padmin/.kshrc

export EDITOR=/usr/bin/vi
export HOST="$(/usr/bin/uname -n)"
export ENTSTAT_MODE=closed.error

alias aix="ioscli oem_setup_env"
alias history='fc -t'

if [ "$(whoami)" != "root" ]; then
  export PS1="[$(whoami)@${HOST}]$ "
  export PS1="[$(whoami)@${HOST}]# "

set -o vi

Overall monitoring of all LPARs

topas -cecdisp
  • Press p to see the Shared Processor Pools
  • Press ~ to toggle between nmon and topas
  • Press E to see SEA stats

See the underlying AIX commands being run from padmin shell

export CLI_DEBUG=33

Schedule backup of virtual and logical configuration

viosbr -backup -file `hostname` -frequency daily -numfiles 7



Determine the physical switch name and port number by sniffing CDP packets.

# Name:
# Description:
#       Tries to determine the physical switch name and port number by sniffing
#       CDP packets. Only works on Cisco switches and only if the port is configured to 
#       publish CDP packets.
# Usage:
#       Optional:
#                 -c : CSV output format.
# Requirements:
#       timeout - Comes as part of the coreutils RPM which can be installed with yum.
# Note:
#       This script isn't perfect. Sometimes the packet captured isn't always
#       the CDP one, so you may need to run it multiple times to get the switch details when
#       the real adapter is a link aggregation of multiple physical ports.

hostname=$(hostname -s | tr '[:upper:]' '[:lower:]')

# Arguments
while getopts "c" OPTION; do
    case ${OPTION} in
        c) CSVOUT="yes"
        *) CSVOUT="no"

if [[ ${CSVOUT} = "no" ]]; then
    printf '\n%s\n' "Sniffing Shared Ethernet Adapters (SEA) for switch details"
    printf '%s\n' "This may take up to 90 seconds per SEA..."
# Iterate through all SEA's
for sea in $(lsdev -F 'name description' | awk '/Shared Ethernet Adapter/{ print $1 }'); do
    # Get SEA real adapter
    real_adapter=$(lsattr -E -a real_adapter -F value -l "${sea}")

    # If the real adapter is an EtherChannel, get a count of the number of aggregated interfaces
    if [ "$(lsdev -l "${real_adapter}" -F 'description')" = "EtherChannel / IEEE 802.3ad Link Aggregation" ]; then
        ent_count=$(lsattr -E -a adapter_names -F value -l "${real_adapter}" | awk -F ',' '{ print NF }')

    # Bring SEA's en adapter up (if not already)
    sea_en_state=$(lsattr -E -a state -F value -l en"${sea#ent}")
    if [ "${sea_en_state}" != "up" ]; then
      chdev -a state=up -l en"${sea#ent}" > /dev/null 2>&1

    # Capture CDP packets from the SEA interface. If tcpdump is still running after 90 seconds, 
    # we may have hit a switch that either isn't Cisco, or the port doesn't have CDP enabled
    unset switchinfo
    set -A switchinfo $(/home/padmin/scripts/bin/timeout 90s tcpdump -i en"${sea#ent}" -s 1500 -c "${ent_count}" -v 'ether[20:2] = 0x2000' 2>&1 | grep -E '(Device-ID|Port-ID)' | awk -F ':' 'v && NR==n{ gsub(/ /, "", v); gsub(/ /, "", $3); print v","$3}/Device-ID/{ v=$3; n=NR+1 }')

    # If the SEA interface was down when we first checked, lets put it back in that state
    if [ "${sea_en_state}" = "down" ]; then
        chdev -a state=down -l en"${sea#ent}" > /dev/null 2>&1
        ifconfig en"${sea#ent}" down detach

    # Print values
    if [[ ${CSVOUT} = "yes" ]]; then
        if [ "${switchinfo}" = "NONE" ]; then
            printf '%s\n' "${hostname},Switch not Cisco, CDP not enabled, or CDP packet not seen in timeout duration."
            printf '%s\n' "hostname,sea,real_adapter,switchname,switchport"
            while [[ ${count} < ${#switchinfo[*]} ]]; do
                switchname=$(echo "${switchinfo[$count]}" | awk -F ',' '{gsub("\047", "", $1); print $1}')    # \047 is octal for a single quote
                switchport=$(echo "${switchinfo[$count]}" | awk -F ',' '{gsub("\047", "", $2); print $2}')    # \047 is octal for a single quote
                printf "%s\n" "${hostname},${sea},${real_adapter},${switchname},${switchport}"
                (( count+=1 ))
        if [ "${switchinfo}" = "NONE" ]; then
            printf '\n%s\n' "+-----------------------------------+"
            printf '%s\n' "SEA: ${sea}\n"
            printf '%s\n' "    -> Backing adapter: ${real_adapter}"
            printf '%s\n' "+-----------------------------------+"
            printf '%s\n' "Unable to determine switch details."
            printf '%s\n' "  Switch not Cisco, CDP not enabled, or CDP packet not seen in timeout duration."
            printf '\n%s\n' "+-----------------------------------+"
            printf '%s\n' "SEA: ${sea}"
            printf '%s\n' "    -> Backing adapter: ${real_adapter}"
            printf '%s\n' "+-----------------------------------+"
            printf '%-40s %s\n' "Switch" "Port"
            printf '%-40s %s\n' "------" "----"
            while [[ ${count} < ${#switchinfo[*]} ]]; do
                switchname=$(echo "${switchinfo[$count]}" | awk -F ',' '{gsub("\047", "", $1); print $1}')    # \047 is octal for a single quote
                switchport=$(echo "${switchinfo[$count]}" | awk -F ',' '{gsub("\047", "", $2); print $2}')    # \047 is octal for a single quote
                printf "%-40s %s\n" "${switchname}" "${switchport}"
                (( count+=1 ))