hidden:Starting and Stopping HPSS

From Lsdf
Revision as of 17:20, 27 May 2021 by Dorin-daniel.lobontu (talk | contribs) (Miscellaneous)

Activate/deactivate HPSS on either hpsscr04 or hpsscr05

1.Activate HPSS 
run: /hat/core_manual/bin/hpss_activate_node.ksh
This script:
- activates hpss volume groups (see Miscellaneous for details)
- mounts hpss file systems
- sets IPs of the productive HPSS core
- starts hpss and DB2
- copies /hat/core_manual/etc/root_active_crontab over /var/spool/cron/root
2.Deactivate HPSS active core 
run: /hat/core_manual/bin/hpss_deactivate_node.ksh
This script:
- stops HPSS and DB2
- umounts all HPSS file systems
- deactivates all HPSS volume groups
- takes away all HPSS productive IPs
- deletes /var/spool/cron/root

Please note:
- if you solely want to stop HPSS and DB2 but have access to HPSS file systems
please see below how only stop hpss
- if you want to set a cronjob on the active core, write it in 
/hat/core_manual/etc/root_active_crontab
to be activeted at the next run of 
/hat/core_manual/bin/hpss_activate_node.ksh
Beware:
keep /hat/core_manual synchronized on both hpsscr04 and hpsscr05

Stop/Start HPSS and DB2 on the active core without deactivating it

1.Stop all HPSS processes and DB2 in one step
/opt/hpss/bin/rc.hpss stop

2.Alternatively, stop HPSS and DB2 in more steps for a better overview 

2.1.Start the command line on the core server and stop all server 
/opt/hpss/bin/hpssadm.pl -U hpssssm -a /var/hpss/etc/hpss.unix.keytab
hpssadm> server shutdown -all
hpssadm> server list -all  
hpssadm> quit
2.2.Stop startup daemon
/opt/hpss/bin/rc.hpss -d stop
2.3.Stop system manager
/opt/hpss/bin/rc.hpss -m stop
2.4.Stop DB2
2.5.Stop DB2 using rc.hpss:
/opt/hpss/bin/rc.hpss stop
Please note:
rc.hpss will try to stop all hpss processes
rc.hpss will find all hpss processes stopped (had been stopped in step 2.3)
rc.hpss will only stop DB2
DB2 can also be stopped using DB2 manager
su - hpssdb
db2stop
exit
3. Start HPSS and DB2
as user root on the core server
rc.hpss starts all hpss processes and DB2 as well
/opt/hpss/bin/rc.hpss start
check if all servers are running
start hpss admin tool
/opt/hpss/bin/hpssadm.pl -U hpssssm -a /var/hpss/etc/hpss.unix.keytab
check if all servers are running
hpssadm>server list -all  
a list of servers will be showed please note:
"status: connected" and "op state: enabled"  
close hpss admin tool
hpssadm> quit

Alternatively you can stop/start HPSS servers using HPSS GUI

In this case stopping is done in two steps. 
First in the GUI all hpss servers are to be stopped and then the database is to be stopped.

1. open the HPSS GUI on the active Core-Server
/opt/hpss/bin/hpssgui.pl &
2. click on Operations --> Shutdown --> All Non-SSM Servers
3. then Operations --> Shutdown --> System Manager
4. stop DB2:
either using rc.hpss 
/opt/hpss/bin/rc.hpss stop
or DB2 manager
su - hpssdb
db2stop
exit

Starting also in two steps:
start DB2:
su - hpssdb
db2start
exit

start HPSS GUI and start all HPSS servers

Miscellaneous

1.HPSS volumes groups:
at the OS level HPSS and DB2 is using the following volume groups:
db2_backup1_vg      : used for DB2 backups
db2_backup2_vg      : also used for DB2 backups
db2_log_vg          : used for storing active DB2 redologs
db2_logarch1_vg     : used for DB2 redologs archived path 1 
db2_logarch2_vg     : used for DB2 redologs archived path 2
db2_logmirror_vg    : used for mirrored redologs
hpss_tbspc1_vg      : used for DB2 tablespaces
hpss_tbspc2_vg      : also used for DB2 tablespaces
hpss_vg             : used for HPSS specific files
Please note:
DB2 active logs (redologs) cannot be backuped; they have to be actively mirrored
Please also take a look at scripts in /root/software but please used them with CAUTION:
These scripts are intended to be used by a skilled administraor to manually
activate/deactivate/export/import/mount/umount hpss vg/fs
activate_hpss_vg.sh
deactivate_hpss_vg.sh
export_hpss_vg.sh
import_hpss_vg.sh
mount_hpss_fs.sh
umount_hpss_fs.sh





back to HPPS main page