9/12/2017 HALDOM Cluster 43
Home OVMfor SPARC OVMforx86 OE WOpscenter Solaris 11. Zoner_—‘Oraclo upercluster—_Exadata
HA LDOM Cluster 4.3
How to create a HA-LDOM cluster using Oracle Solaris CLuster 4.3
Setup
Se
2x T6320 blades (1 x CPU, 64GB mem, 1 x dual 4Gbit FC adapter wth dual 1Gbe adapter, 2 x NEM }
‘Storage: Sun Storage 2540 with dual controllers
08: Oracle Solaris 4.3
Cluster Software: Oracle Solaris Cluster 4.3
Configuration:
Below is a picture of the sotup:
ca ory
Re EreelIi cy
ie
Creee ee)
Cluster4.3
The basic layout is as follows. The lwo master servers are 76320 blades (SPARC) in a SunBlade 6000 chassis. Each blade has a
Express Module that contains 2 x Gbe network ports and 2 x 4Gbit fc ports. This is @ combo EM adapter at the back of the SB6000,
‘chassis. The blade chassis also has 2 x Network Express Modules at the back that contains 1Gbe ports,
‘So each blade has 4 x 1Gbe network ports and 2 x fe ports. Two ports, nett and net3 are used for public network and uses OLMP.
‘The other 2 x nets is used as the cluster interconnect. | used straight cables connected direcly to the ports on each blade at the
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp 199/12/2017 HALDOM Cluster 43
back. | aid this because the two blades are next to each other in the chassis so i's not needed to use a switch, Ifyour nodes are far
apart then you would need to connect the nodes via a switch or use tagged vians on the public net forthe cluster interconnect
The storage used is @ Sun Storage 2540 array with dul controllers, Each controler has 2 x 4 Gbit FC ports, Luns were created and
are visible to both nodes.
Cluster Setup
This page does not discuss how | installed the OS or the cluster sotware. | wil have lis inthe future on how to do thal. | basically
used the EIS checklist of April 2016 fo achieve the instal of the OS and cluster.
Making a HA-LDOM configuration.
Please note that some of the commands are executed on both and some only on the “master” node. In this document the two nodes
are nodet (calsto) and node2 (ananke).
(On Both Nodes (callisto and ananke)
Justa note or two here. In the example | used DLMEP forthe virtua switch (vsw) on the primary. You could also use Link
‘Aqaregation for the vsw or physical net ports. This Is just an example and you would probably change the vsw to whatever you
decide on ste
Create the primary domains. This must be done on both nodes. l only show the output of the first node callisto.
rootéeallisto # 1dm add-vconscon port-range=5000-$100 primary-vee0 primary
rootgcallisto # sveadn enable sve: /1doms/vnted:default
rootécallisto # 1dm add-vewitch net-dev-dimp0 primary-vew0 primary
rootécallisto # 1dn add-vdiskserver primary-vde0 primary
rootcallisto # 1dm List-bindings primary
rootécallisto # pkg info Idonsmanager | grep -i version
rootécallisto # svcadm enable lind
rootécallisto # 1dm set-core 2 primary
unit 7 was removed from domain primary
6 was semoved from domaii
3 wae removed from domain primary
crypto unit 2 was removed from domain primary
rootécallisto # ldm list
primary active se 16 6S408e 0.68 Sim
rootécallisto # 1dm start-reconf primary
nitiating a delayed reconfiguration operation on the primary domain.
All configuration changes for other domains are disabled until the primary
Gomain reboots, at xhich time the new configuration fer the primary domain
will alsa take =:
rootécallisto # 1dm set-menory 166 primary
cet The primary domain is in 55 of a delayed reconfiguration
rootécallisto # idm add-config initial
rooticallisto # 1dn list-contig
rootfcallisto # shutdown -y -i6 -g0
Set the failover policy to reset on both nodes
rootécallisto # 1dm set-domain failure-polic
¢ primary
Setup the Idem. Only on nodet
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp 299/12/2017 HALDOM Cluster 43
dm add-domain fldom)
dm set-core 2 fldomi
dm set-men 16g f1doat
dm list £ldom
nactive 16166
dm list-services primary
dm add-vnet vnetO primary-vew0 fldomt
dm set-domain master=primary fidom1
dm set-var auto-boot?=false fidemt
dn list -o domain fldomt
65 UTIL
out
S047e5£1-78d6-4145-a176-bhda27e6efld
faiiure-policy
extended-napin-space
ads-policy=
aute-hoot?-£a
(On Both Nodes add the backend device. Must use d2s2
rootfcallisto # 1dm add-vdsdev /dev/did/rdsk/d2s2 fldomi_boot@primary-vds0
root@ananke # ldm add-visdev /dev/did/rdsk/d2s2 fldom1_bootéprimary-vds0
(n first node nodet add the vdisk, bind the domain and boot. Install the guest.
rootécallisto # 1dn add-vdisk fldoni_boot fidoml_bootéprimary-vde0 #1dom1
rootfcallisto # 1dm bind fldom1
rootécallisto # 1dm start fidom
rootfcallisto # telnet localhost 5000
‘The below seting is for when you want to boot from an Automated Installer (A) server, If you have added an iso image then you
‘would install from that image. If you want to boo! fom an iso image then you need to assign the image as a virtual disk tothe Idom
Ike this (ldm add-vdedev /var/tmp/solarisli_3.iso fldoml_isolprimary-vde0 primary; 1dn add-vdisk
fidomi_iso fldoml_Asotprinary-vds0 fldomi). You would then boot using boot #idomi_iso.
Boot from the network and install from Al
ok setenv network-boot-argunents host~ip=192.168.50.121,router-ip=192.168.50.1,subnet~
mask=255.255.255.0,file=http: //192.168.50.40:5555/cgi-bin/wanboot-cgi
ok boot vnet0 - install
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp9/12/2017 HALDOM Cluster 43
You may wish ¢
reboot ¢
automated ation finished successfully
the system ooted now
Please refer to the /systen/volatile/install_ log file for deta
Aster reboot it will be located at /var/log/install/inatal.
solaris console login: root
Paseword: solaris
01:16:29 solaris login: ROOT LOGIN /dev/console
cle Corporation sunos 5.1 11.3 March 2016
roottsolaris:~# reboot
(On both nodes check If ha-ldom data service is installed. I not Install it
Footecalliste # pkg info ha-cluster/data-service/na-idom
rootdananke # pkg info ha-cluster/data-sezvice/ha-ldom
Sunnary: Oracle fer HA Cor xVM x86-66/SPARC Guest Domains
beseription aria Cluster HA for xVM Server x86-64/SPARC Guest
version: 4 fe Solaris Cluster 4.3.0.24.0
Release
Branch: 0.24.0
Packaging Date: August 26, 2015 11:32:36 OM
MRI: pkg? //ha-cluster/ha-cluster/data-service/na-idowt.3,5.
2o1soazer2332362
Check on both nodes that first entries in /etcinsswitch is cluster. If not change it.
rootfcallisto # svecfg -s nane-service/awitch listprop config/host
aster files"
config/host astring
rootécalliste # svcefg -s name-service/switch Listprop config/ipnodes
rootfcallisto # svecfg -s nane-service/switch listprop config/netmask
To set a proporty use the folloiwng command:
rootécallisto # svecfg -s name-service/switch setprop config/ipnodes = astring:
‘(eluater files ans)’
rootfcalliste # svecfg -s name-service/switch listprop config/ipnodes
Create resource group node 1 only
rootécalliste #| clzg create -n callisto,ananke fldemi-rg
Register HAStoragePlus
rootdcalliste # clrt register SUNW.HAStoragePlue
Create HAStoragePlus resource for boot disk
rootécallisto # clrs create -g fldonl-rg -t SUNW.HAStoragePlus -p
GlobalDeviceraths=/dev/global/dsk/d22 f1dom1-has-r=
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp9/12/2017 HALDOM Cluster 43
Enable resource fidomt-rg
rootfcallisto # clzg online -M -n callisto fldeml-rg
rootécallisto # clrg status
Group Name Node Nane suspended status
Sidon sallisto No online
Register Idom data service
rooticallisto # clrt register SUNW.1don
rootécalliste # crt list
SUNW. Shareaadaire:
SUN, HAStoragePlue:11
sun. 1dom:7
Create password file for failover Idoms (This only works if SUNW.ldom:6 or 7)
rootdcallisto # clpstring create -t resource -b fldoml-re fldom1-rg_fldom1~
Create fldomt resource
rootécallisto #| clra create -g fldonl-rg -t SUNW.1dom -p Domain name=fidomi -p
resource dependencies=fldoni-has-rs fldoni-re
Check migration type property. It must be MIGRATE. If not set it to MIGRATE
rootcalliste # clr show -v fldoml-ra | grep Migration type
Migration type MIGRATE
(Zo change do the following)
cles set -p Migration type-MIGRATE fldoni-rs
To stop Idom resource disable it
rootécallisto # clrs status
Node Name state
fidomi-rs ealliste online Online ~ fldoni is active (nozmal)
ananke
Stdomi-has-re cailisto
ananke
rootfcallisto # clrs disable fldom1-rs
rootécallisto # clrs status
de Name ates Kes!
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp9/12/2017 HALDOM Cluster 43
sidomi-zs calliste offline offline - Successfully stopped fidont
ananke offline offline
fdomi-has-rs cailisto online online
ananke ores offline
rootécallisto # ldm list
NAME STATE “FLAGS CONS-«VCPU MEMORY «UTIL UPTIME
primary active neve SP 16165084 Th Sim
idom snactive ae 16s
‘To Start enable the resource
rooticallisto # clrs enable fldoni-rs
rootfcallisto # crs status
Cluster Resources
Resource Name Node Name State status Message
calliste online Online ~ fldomi is active (noxmal)
ofeline ossline
‘Sidomi-has-re catlisto online online
ananke offline offline
rootécallisto # dm list
wae stare cons. VCPU MEMORY «UTIL UPTIME
primary active se 16 2.48 th S2m
siden active sooo 16 ae zis
‘Switch over the resource to node 2 (ananke)
rootfcallisto # clrg switch -n ananke fidomi-rg
root¢ananke # ldm list
NAME stare cons VCPU MEMORY UTIL UPTIME
primary acti se 66s 0.58 Ih Sem
Sidon active $000 16 6s 1S hen
rootananke # cirs status
== clus} —
Resource Name Node Name state Status Message
Aidoni-es caltiste offline offline
ananke online online ~ Service is online
Sidowi-has-re cailisto offline offline
anke online online
Switch back to nedet (callisto)
root@ananke # clrg switch -n calliste fidomi-rg
rootcallisto # 1dm list
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp9/12/2017 HALDOM Cluster 43
prinary active se 166s. 2.88 2h an
iden active =n 500016 165.14 10m
rootfcallisto # clrs status
a+ cluster Resources ===
Resource Nane ode Name state
Eldomi-rs calliste online Online - Sezvice is online.
ananke offline offline
cidomi-has-re cailisve online OnLine
ananke offline offline
‘To test node failure stop the server from ILOM using the -f option and check whether the Idom fails over to the other node.
From nodet (callisto ILOM)
> step -£ /svs
Ave you sure you want to inmediately stop /SY¥S (y/a)? y
Stopping /S¥S innediately
Check status from node? (ananke)
rootananke # clrs statue
=== cluster Resources
Resource Nane Node Name state Status Message
Sidomi-rs eallisto offline offline
ananke Starting Unknown - Starting
Eidomi-has-rs ealliste oftline ogeline
ananke online Degraded - Device service dsk/d2 associated with
path /dev/global/dsk/d2s2 is Degraded.
dom show up and running from node2 (ananke)
root@ananke # clrs status
cluster Resources
Resource Name Node Name State Status Message
fidomi-rs caltiste offline offline
ananke ontine Online ~ fldoml is active (normal)
Eldomi-has-re. callisto offline offline
ananke online Degraded - Device service dsk/d2 associated with
path /dev/giobai/dsk/d292 is Degraded.
root¢ananke # ldm list
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp 79.9/12/2017
primary
sadona
‘Start nodet (callisto) again and migrate back to primary
-> start /s¥s
Are you sure you want te start /S¥S (y/a)? y
Check nodes after nodet started
rootananke # cldev status
o-- cluster
D devices
Device Instan:
/dev/di/rdsk/al ananke
calliste
/dev/aid/rdsk/e2 ananke
callisto
/ev/ald/rask/as calliste
/dev/aid/rdsk/ae callisto
/dev/did/rdsk/as callisto
Jdev/aid/eask/a6 callisro
Jdeu/ahd/edsk/d? ananke
/eev/did/rask/ae ananke
/dev/did/rdsk/as ananke
callisto
root@ananke # cirs status
Cluster Resources
Resource Nane Node Nane State
fidomi-rs ealiiate offline
ananke online
Eldom-has-rs calliste offline
ananke online
‘Switch the resource group back to nodot (callisto)
rootécallisto # clrs status
cluster Resources
hiipsos-chronicles.yolasite,comma-Kdom-clusterphp
1616s 0
HALDOM Cluster 4.3
2h an,
16 16s 0.18 an
ok
0K
Unnonitered
Unnonitozed
Unnonitoree
ok
Uunnonitored
Unnonitered
ok
ok
Status Message
offline
Online ~ fldori is active (normal)
offline
online9/12/2017 HALDOM Cluster 43
ealtiate online ~ fldori is active (normal)
ananke offline
sidomi-has-re cailisto online online
ananke offline DesLine
Koop in mind that if you want to add additional disks tothe Idom you need to ad only the vdsdev to both nodes and add the vdlisk to
the ldom. The cluster wil take care ofthe rest
[NBNBNENB: Save your config on both nodes withthe add-config option to kim.
rootécallisto # 1dn add-config ha-ldom-latest
rootécolliste # dm List-contig
root@ananke # ldm add-config ha-ldom-latest
root@ananke # ldm list-config
factory-defauit
snitial
you don't then you have to go and do this all over again!
Back to Home Page
Back to OVM for SPARC
hiipsos-chronicles.yolasite,comma-Kdom-custerphp