MC ServiceGuard - Reasons for TOC

quinta-feira, 29 de dezembro de 2011

Transfer of Control (TOC)

MCSG will TOC a system in order to release system resources and to ensure data integrity. There are several scenarios in which MCSG will TOC a system. They are:

A two-node cluster loses heartbeat at which time a single node cluster will form. The system that loses the race to the lock disk will TOC.

Checking that:

# cmviewconf

Cluster information:cluster name: testcluster
version: 0
flags: 12 (single cluster lock)
heartbeat interval: 1.00 (seconds)
node timeout: 8.00 (seconds)
heartbeat connection timeout: 16.00 (seconds)
auto start timeout: 600.00 (seconds)
network polling interval: 2.00 (seconds)
first lock vg name: /dev/vglock
second lock vg name: (not configured)
Cluster Node information:Node ID 1:
Node name: node1
first lock pv name: /dev/dsk/c0t4d4
first lock disk interface type: c720
Network ID 1:

 mac addr: 0x080009fd4375
hardware path: 8/16/6
network interface name: lan0
subnet: 15.140.8.0
subnet mask: 255.255.248.0
ip address: 15.140.10.236
flags: 1 (Heartbeat Network) 
bridged net ID: 1

# lanscan

Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
8/16/6 0x080009FD4375 0 UP lan0 snap0 1 ETHER Yes 119
8/8/2/0 0x00108318AFEE 2 UP lan2 snap2 2 ETHER Yes 119
8/8/1/0 0x00108318AFED 1 UP lan1 snap1 3 ETHER Yes 119

# cmscancl -n node -o /tmp/scan.log

Check the "link-level connectivity" in the log.

The cluster daemon, cmcld, dies for any reason

msg in the log:
Serviceguard: Unable to maintain contact with cmcld daemon.Performing TOC to ensure data integrity.

NODE_FAIL_FAST=YES is set in a package configuration file.

The cluster lvm daemon, cmlvmd, dies for any reason.

System safety time is disabled via the cmsetsafety command.

SERVICE_FAIL_FAST_ENABLED = YES is set (causes reboot).

You can confirm TOC searching in the /etc/shutdownlog  for something like that..

18:23 Thu Apr 24 2003. Reboot after panic: SafetyTimer expired, ...