In DB2 pureScale, we use a RSCT peer domain and a GPFS domain. These domains have couple of machines networked together. In democracy, we choose someone to be a leader / master / manager to create an order. The same type of democracy exists in RSCT and GPFS where the surviving machines elect a leader / master / manager to do the work. When some computer dies, they assemble together at the watercooler and again elect a leader / master / manager.
There are some information that you can only see on the node that is either a leader / master / manager.
node02:~ # mmlsmgr file system manager node ---------------- ------------------ db2data2 192.168.142.103 (node03) db2data4 192.168.142.103 (node03) db2log 192.168.142.103 (node03) db2data1 192.168.142.104 (node04) db2data3 192.168.142.104 (node04) db2fs1 192.168.142.104 (node04) Cluster manager node: 192.168.142.103 (node03)
RSCT Recovery Master
node02:~ # lssrc -ls IBM.RecoveryRM | grep Master Master Node Name : node03 (node number = 2)
RSCT Configuration Leader
node02:~ # lssrc -ls IBM.ConfigRM | grep Leader GroupLeader: node03, 0x7903784ce20081d5, 2 TieBreaker GroupLeader: node03, 0x7903784ce20081d5, 2