San Francisco

dave spink toolset




E10K SSP ADMIN:

SSP INFO SSP ENV FAILOVER DOMAIN MANAGEMENT
CONTROL BOARDS DYNAMIC RECONFIG

SSP INFO

hostinfo -h ;system boards
hostinfo -p ;power status
hostinfo -F ;fan status
domain_status ;list of domains configured - read existing file
hostinfo -hFtSp ;status of system components
fan -s nominal | fast ;display or control fan power and speed
cat $SSPVAR/.ssp_private/cb_config ;to show main control board
power ;display status
showusage -b 9 -r CPU ;report on system board 9 CPU configuration
showusage -b 9 -r DRAIN ;check if board is being drained
showusage -b 9 -r IO ;check I/O usage
showusage -b 9 -r OBP ;check I/O slots and if memory pageable or not
showusage -b 9 -r MEM ;check if memory pageable or not
showusage -b 9 -r UNSAFE ;check for unsafe devices




SSP ENV

$SSPETC ;/etc/opt/SUNWssp startup scripts
$SSPVAR ;/var/opt/SUNWssp
$SSPOPT ;/opt/SUNWssp executables and man pages
$SSPLOGGER ;/var/opt/SUNWssp/adm, messages
$SSPLOGGER/$mydom/messages ;bringup domain messages
$SSPLOGGER/messages ;for platform
$SSPVAR/adm/$mydom/post ;bringup post messages
$SSPVAR/.ssp_private ;domain_config, cb_config, templates for eeprom images
$SSPVAR/etc ;platform, blacklist, domains, working copies of eeprom images




FAILOVER

showfailover ;display the failover status of the SSP and control board
$SSPOPT/bin/ssp_backup /tempdir ;creates ssp_backup.cpio, must be root
setfailover off ;disable failover state
setfailover force ;force failover
setfailover on ;enables auotmatic failover
setfailover -m ;show virtual memory threshold
setfailover -d ;show disk threshold
ps -ef | grep datasyncd ;keeps files in sync between SSPs on a hourly basis
setdatasync ;for immediate SSP file syncronize




DOMAIN MANAGEMENT

Create Domain
sys_id -f eeprom.image.$mydom -k key -s serial_number ;only used for creating 1st eeprom.image
sys_id -f eeprom.image.$mydom -k key -h hostid ;creating other eeprom.image files
sys_id -d -f $SSPVAR/etc/$myssp/$mydom/eeprom.image ;show ID PROM working copy info - MAC, HostID
domain_create -d $mydom -b 0 1 -o os_version -p platform ;create domains


Starting a Domain
domain_switch $mydom ;sets SUNW_HOSTNAME variable
echo $SUNW_HOSTNAME ;ensure host variable is properly set
power -on ;power on all boards in the domain
power -on -csb 0 1 ;use when starting first domain on E10K, ensures centre plane starts
bringup -A off ;set autoboot off, logs $SSPVAR/adm/$mydom/post read /export/home/ssp/.postrc
netcon ;open console window


Stopping a Domain
domain_switch $mydom ;sets SUNW_HOSTNAME variable
domain_status ;get system board numbers
echo $SUNW_HOSTNAME ;ensure host variable is properly set
init 0 ;take OS to okay prompt
hostinfo -h ;check system shows no OS running
power -off ;power off all boards in the domain
hostinfo -p ;check power is off for this domain


Complete PowerOff E10K
# power -off -B ;power off entire E10K inc. plus breakers (this was cool, you hear click, click, click)
# power -off -all ;power off entire E10K leave breakers on


Accessing a Domain
domain_switch $mydom ;sets SUNW_HOSTNAME variable
echo $SUNW_HOSTNAME ;ensure host variable is properly set
netcon ;open console window, /etc/ssphostname must exist and contain virtual IP or SSP
netcon_server -r ;restart netcon daemon, only run when netcon_server has died
netcon_server -p 0 ;restarts daemon using specified boot processor




CONTROL BOARDS

Switching Active CB
power -off -all ;first check all domains are shutdown, then power off domains
cb_reset $myssp-cb0 ;don't do this with domains running, it will crash domains
setfailover -t cb force ;failover control board
power -on -all ;power on domains
showfailover ;check status


Switching JTAG
setfailover -t cb force ;you can move JTAG with domains running
setfailover -t cb on ;failover JTAG
cb_prom -r -h $myssp-cb0 ;display PROM version




DYNAMIC RECONFIG

Attaching System Boards

% dr
dr> init_attach 5
Initiate board attachment, board 5 to domain d1ua0002.
Adding board 5 to domain_config file...
Board attachment initiated successfully.
Ready to COMPLETE board attachment.

dr> complete_attach 5
Completing attach for board 5.
Board attachment completed successfully.

Detaching System Boards

# ifconfig ge1 unplumb
# vmsa
# /opt/jni/ezfibre/standalone/ezf

Umount EMC file systems, and unload the detach-unsafe or suspend-unsafe driver. If modunload fails you need to reboot -- -r.

# unmountall
# modinfo | grep fcaw
# modunload -i 31			;31 is driver number from modinfo

Check the usage and then perform drain.

% showusage -b 5 -r unsafe
% dr
dr> drshow 5 IO
dr> drain 5
dr> drshow 7 drain