The procedure for replacing the PCI card can vary depending on whether the PCI card is only connected to the cluster interconnect or connected to the cluster interconnect and GLS Public LAN.
Save hardware configuration information in the file before replacing the cluster interconnect.
# /usr/sbin/prtpicl -v > pre_prtpicl.log
Disconnect the device from the cluster interconnect. Then, make sure "Receptacle" is "disconnected" and "Occupant" is "unconfigured" respectively.
For cfgadm command, see the documentation for devices published by Oracle Solaris.
Example: The device of the PCI card to be replaced is /dev/fjgi7, and the value indicating the location of PCI slot is pcipsy8:R0B00-PCI#slot05.
(For how to calculate the value indicating the location of PCI slot, see the documentation for devices published by Oracle Solaris.)
# cftool -d
Number Device Type Speed Mtu State Configured Address
1 /dev/fjgi1 4 100 1432 UP YES 00.00.0e.25.1a.38
2 /dev/fjgi7 4 100 1432 UP YES 00.00.0e.25.1a.38
# cfrecon -d /dev/fjgi7 Delete Cluster Interconnect: device "/dev/fjgi7" successfully removed from node configuration # cfgadm -c disconnect pcipsy8:R0B00-PCI#slot05 # cfgadm pcipsy8:R0B00-PCI#slot05 Ap_Id Type Receptacle Occupant Condition pcipsy8:R0B00-PCI#slot05 unknown disconnected unconfigured unknown
Edit /etc/default/cluster to delete the settings of the device of the PCI card to be replaced.
[Before change]
nodename node0 clustername PRIMECLUSTER device /dev/fjgi1 device /dev/fjgi7
[After change]
nodename node0 clustername PRIMECLUSTER device /dev/fjgi1
If necessary, execute the following commands.
# cfgadm -c disconnect pcipsy8:R0B00-PCI#slot05
# cfgadm pcipsy8:R0B00-PCI#slot05
Ap_Id Type Receptacle Occupant Condition pcipsy8:R0B00-PCI#slot05 unknown disconnected unconfigured unknown
CE removes the PCI card.
CE installs a new PCI card on the PCI slot.
Determine the value that indicates the PCI slot location at step 4.
For how to request the value, see the documentation for devices published by Oracle Solaris.
Add the replaced PCI card. Specify the value at step 5 after "cfgadm -c configure". Then, make sure "Receptacle" is "connected" and "Occupant" is "configured" respectively.
For cfgadm command, see the documentation for devices published by Oracle Solaris.
Example: The value determined in step 5 is pcipsy18:R0B02-PCI#slot02.
# cfgadm -c configure pcipsy18:R0B02-PCI#slot02 # cfgadm pcipsy18:R0B02-PCI#slot02
Ap_Id Type Receptacle Occupant Condition pcipsy18:R0B02-PCI#slot02 bridge/hp connected configured ok
Save hardware configuration information in a different file than that previously used.
# /usr/sbin/prtpicl -v > post_prtpicl.log
Compare the files and specify the interface name and number of the PCI card.
# diff -c pre_prtpicl.log post_prtpicl.log | more
(snip)
! :devfs-path /pci@9c,4000/SUNW,hme@3,1
! :driver-name hme
! :binding-name SUNW,hme
! :bus-addr 3,1
! :instance 9
! :_class obp-device
! :name SUNW,hme
(The rest is omitted)
Add the PCI card in the cluster interconnect using the interface name and number.
Example: When adding /dev/hme9.
# cfrecon -a /dev/hme9
Add Cluster Interconnect:
device "/dev/hme9" successfully added to node configuration
Make sure the following message is output to the console of each node.
Example: When adding /dev/hme9.
cf:eventlog CF: Route recovery on /dev/hme9 to node node0. (#0000 1 2 2 2)
Make sure the PCI card has successfully been added in the cluster interconnect.
Example: When /dev/hme9 is successfully added.
# cftool -d
Number Device Type Speed Mtu State Configured Address
1 /dev/fjgi1 4 100 1432 UP YES 00.00.0e.25.1a.38
2 /dev/hme9 4 100 1432 UP YES 00.00.0e.25.1a.38
Note
This section describes the operations when the cluster interconnect and GLS public LAN are connected individually to one PCI card that has multiple ports.
In the case of such connection status, you must use the same device name before and after replacing the PCI card by using PCI Hot Plug. Make sure to put the same PCI card in the same PCI slot before and after changing the PCI card.
Disconnect the NIC (PCI card) to be replaced from the GLS redundant configuration.
Since the interface name below will vary depending on the environment, replace the interface name according to the environment.
Fast switching mode
For Solaris 10
# /opt/FJSVhanet/usr/sbin/hanetnic delete -n sha0 -i fjgi7
# /usr/sbin/ifconfig fjgi7 unplumb
# /usr/sbin/ifconfig fjgi7 inet6 unplumb
For Solaris 11 or later
# /opt/FJSVhanet/usr/sbin/hanetnic delete -n sha0 -i net7
# /usr/sbin/ipadm delete-ip net7
Note
To return to the multiplex configuration, the IP address and the netmask of the interface specified by the ipadm delete-ip command must be set.
Note the output result of the ifconfig command.
# /usr/sbin/ifconfig net7
NIC switching mode
Depending on the status of the NIC to be replaced, the following steps will be necessary.
If the NIC is the operation NIC, it is necessary to manually switch the NIC before changing NIC.
In Solaris 11 or later, if the NIC is the primary interface, it is necessary to delete the IP address settings before changing the NIC.
Checking the primary interface name and the operation NIC
Check the primary interface name. The primary interface name is displayed at the beginning of "Interface List" field. In this example, it is "net1."
# /opt/FJSVhanet/usr/sbin/hanetconfig print
[IPv4,Patrol] Name Hostname Mode MAC Adder/Phys ip Interface List +-----------+---------------+----+-----------------+-------------------------+ sha0 192.168.70.1 d 192.168.70.2 net1,net7 sha1 - p 00:00:00:00:00:00 sha0 [IPv6] Name Hostname/prefix Mode Interface List +-----------+---------------------------------+----+-------------------------+
Check the operation NIC name. The operation NIC is the interface displayed with "(ON)" in Device field. In this example, it is "net1."
# /opt/FJSVhanet/usr/sbin/dsphanet
[IPv4,Patrol] Name Status Mode CL Device +----------+--------+----+----+----------------------------------------------+ sha0 Active d ON net1(ON),net7(OFF) sha1 Active p OFF sha0(ON) [IPv6] Name Status Mode CL Device +----------+--------+----+----+----------------------------------------------+
Stopping the HUB monitoring function and the patrol monitoring function
# /opt/FJSVhanet/usr/sbin/hanetpoll off
# /opt/FJSVhanet/usr/sbin/stpptl -n sha1
Switching the physical interface
Execute this step only if the NIC to be replaced is the operations NIC.
# /opt/FJSVhanet/usr/sbin/hanetnic change -n sha0
Deleting IP address settings
Execute this step only in Solaris 11 or later and NIC to be replaced is the primary interface.
# /usr/sbin/ifconfig net7 # /usr/sbin/ipadm delete-ip net1
Note
To return to the multiplex configuration, the IP address and the netmask of the interface specified by the ipadm delete-ip command must be set.
Note the output result of the ifconfig command.
# /usr/sbin/ifconfig net7
GS/SURE linkage mode
Delete the virtual interface bundled with the NIC to be replaced from the redundant configuration.
# /opt/FJSVhanet/usr/sbin/hanetnic delete -n sha0 -i sha2
Disconnect the device from the cluster interconnect. Then, make sure "Receptacle" is "disconnected" and "Occupant" is "unconfigured" respectively.
For cfgadm command, see the documentation for devices published by Oracle Solaris.
Example: The device of the PCI card to be replaced is /dev/fjgi7, and the value indicating the location of PCI slot is pcipsy8:R0B00-PCI#slot05.
(For how to calculate the value indicating the location of PCI slot, see the documentation for devices published by Oracle Solaris.)
# cftool -d
Number Device Type Speed Mtu State Configured Address
1 /dev/fjgi1 4 100 1432 UP YES 00.00.0e.25.1a.38
2 /dev/fjgi7 4 100 1432 UP YES 00.00.0e.25.1a.38
# cfrecon -d /dev/fjgi7 Delete Cluster Interconnect: device "/dev/fjgi7" successfully removed from node configuration # cfgadm -c disconnect pcipsy8:R0B00-PCI#slot05 # cfgadm pcipsy8:R0B00-PCI#slot05 Ap_Id Type Receptacle Occupant Condition pcipsy8:R0B00-PCI#slot05 unknown disconnected unconfigured unknown
Edit /etc/default/cluster to delete the settings of the device of the PCI card to be replaced.
[Before change]
nodename node0 clustername PRIMECLUSTER device /dev/fjgi1 device /dev/fjgi7
[After change]
nodename node0 clustername PRIMECLUSTER device /dev/fjgi1
If necessary, execute the following commands.
# cfgadm -c disconnect pcipsy8:R0B00-PCI#slot05
# cfgadm pcipsy8:R0B00-PCI#slot05
Ap_Id Type Receptacle Occupant Condition pcipsy8:R0B00-PCI#slot05 unknown disconnected unconfigured unknown
CE replaces the PCI card.
Add the replaced PCI card. Make sure "Receptacle" is "connected" and "Occupant" is "configured" respectively.
For cfgadm command, see the documentation for devices published by Oracle Solaris.
Example: The value indicating the location of replaced PCI slot is pcipsy8:R0B00-PCI#slot05.
# cfgadm -c configure pcipsy8:R0B00-PCI#slot05 # cfgadm pcipsy8:R0B00-PCI#slot05 Ap_Id Type Receptacle Occupant Condition pcipsy8:R0B00-PCI#slot05 bridge/hp connected configured ok
Execute the commands below and add the replaced NIC (PCI card) in the GLS redundant configuration.
Since the interface name below will vary depending on the environment, replace the interface name according to the environment.
Fast switching mode
For Solaris 10
# /opt/FJSVhanet/usr/sbin/hanetnic add -n sha0 -i fjgi7
For Solaris 11 or later
# /opt/FJSVhanet/usr/sbin/hanetnic add -n sha0 -i net7
When setting the environment for fast switching mode, execute the following commands to activate the replaced interface in advance:
If using IPv4 address
- For Solaris 10
# /usr/sbin/ifconfig fjgi7 plumb
# /usr/sbin/ifconfig fjgi7 ipaddress netmask + broadcast + -trailers up
- For Solaris 11 or later
# /usr/sbin/ipadm create-ip net7
# /usr/sbin/ipadm create-addr -T static -a ipaddress/netmask net7/v4
If using IPv6 address
- For Solaris 10
# /usr/sbin/ifconfig fjgi7 inet6 plumb up
- For Solaris 11 or later
# /usr/sbin/ipadm create-ip net7 # /usr/sbin/ipadm create-addr -T addrconf net7/v6
NIC switching mode
Depending on the status of the replaced NIC (PCI card), the steps below will be necessary.
In Solaris 11 or later, if the NIC is the primary interface, follow step 1 to manually fail back the NIC and reset the IP address after changing NIC. If not, go to step 2.
Failing back the physical interface and resetting the IP address
- If using IPv4 address
# /usr/bin/touch /var/opt/FJSVhanet/tmp/disable_watchif
# /opt/FJSVhanet/usr/sbin/hanetnic change -n sha0 # /usr/sbin/ipadm delete-ip net1 # /usr/sbin/ipadm create-ip net1 # /usr/sbin/ipadm create-addr -T static -a ipaddress/netmask net1/v4
# /usr/bin/rm -f /var/opt/FJSVhanet/tmp/disable_watchif
- If using IPv6 address
# /usr/bin/touch /var/opt/FJSVhanet/tmp/disable_watchif
# /opt/FJSVhanet/usr/sbin/hanetnic change -n sha0 # /usr/sbin/ipadm delete-ip net1
# /usr/sbin/ipadm create-ip net1
# /usr/sbin/ipadm create-addr -T addrconf net1/v6 # /usr/bin/rm -f /var/opt/FJSVhanet/tmp/disable_watchif
Starting the HUB monitoring function and the patrol monitoring function
# /opt/FJSVhanet/usr/sbin/strptl -n sha1
# /opt/FJSVhanet/usr/sbin/hanetpoll on
Failing back the physical interface
In Solaris 10, if you need to return the operation NIC to the primary interface, fail back the physical interface.
# /opt/FJSVhanet/usr/sbin/hanetnic change -n sha0
GS/SURE linkage mode
Include the virtual interface that is bundled with the replaced NIC in the redundant configuration.
# /opt/FJSVhanet/usr/sbin/hanetnic delete -n sha0 -i sha2
Add the replaced PCI card to the cluster interconnect.
Example: The device of the replaced PCI card is /dev/fjgi7.
# cfrecon -a /dev/fjgi7
Add Cluster Interconnect:
device "/dev/fjgi7" successfully added to node configuration
Make sure the following message is output to the console of each node.
Example: The device of the replaced PCI card is /dev/fjgi7.
cf:eventlog CF: Route recovery on /dev/fjgi7 to node node0. (#0000 1 2 2 2)
Make sure the PCI card has successfully been added in the cluster interconnect.
Example: The device of the replaced PCI card is /dev/fjgi7.
# cftool -d
Number Device Type Speed Mtu State Configured Address
1 /dev/fjgi1 4 100 1432 UP YES 00.00.0e.25.1a.38
2 /dev/fjgi7 4 100 1432 UP YES 00.00.0e.25.1a.38