How to stop and restart the VCS Cluster Engine (HAD), GAB & LLT without taking production cluster groups, or resources, offline in Veritas Storage Foundation HA for Windows

Article:TECH53516  |  Created: 2007-01-12  |  Updated: 2014-08-04  |  Article URL http://www.symantec.com/docs/TECH53516
Article Type
Technical Solution

Product(s)

Environment

Issue



When HAD is stopped, without using the "-force" argument, it enters the "LEAVING" state and waits for all groups to be taken offline on the node. Appending "-force" the end of the hastop command allows HAD to be stopped without taking any production service groups, or resources, offline on the system.

This allows you to stop, and restart, VCS without bringing production offline.
 


Solution



Use the commands in Table 1 to stop and restart a VCS cluster. To stop the cluster, start at the top of the table and go down. To start the cluster, start at the bottom and go up.

 

Table 1 - A list of commands and scripts that are used to manually stop or start VCS

Port Going Down (Stop)  ↓ Going Up (Start) ↑  Application Cluster Network
-- Use haconf to save and 
close the cluster 
configuration:

haconf -dump -makero

Note: If the 
configuration is already 
closed, a "Cluster not 
writeable" message will 
be returned.
Use hastatus or hasys to 
confirm that the cluster 
starts successfully:

hastatus -sum

hasys -state
 

 
h (had) hastop [-all | -local] [-force]

Note: Using the -force 
switch causes had to 
simply stop without 
bringing any resources 
offline. Otherwise, it 
brings the cluster 
resources offline.
hastart  

 

 

vcscomm net stop vcscomm net start vcscomm

Note: Running hastart
automatically starts
this service.
-- --
a (gab) net stop gab net start gab

Note: Running hastart
automatically starts
this service.
   

llt net stop llt net start llt

Note: Running hastart
automatically starts
this service.
   

Port Going Down (Stop) ↓ Going Up (Start) ↑  Application Cluster Network


Notice that the table is divided into "Application" and "Cluster Network." In most cases, restarting the application is all that is necessary. Issues that require modifying, or restarting, the network drivers, or configuration, typically require you to stop and restart both the application and cluster network drivers (GAB and LLT) as well as VCSComm, which lies between these layers. 

Note: Use the "-force" argument to avoid bringing production groups offline. If the "-force" is not used, all cluster groups on this node will be brought offline, or moved to another node. The "Cluster Service" group will gracefully stop when HAD is stopped, even if the "-force" argument is used. The Cluster Service group does not need to be online for the other resources to remain online. If this is a global cluster (GCO) the "wide-area connector" resource will brought offline when the Cluster Service group is taken offline, temporarily stopping communication with remote clusters. Since VCS is able to detect the difference between a graceful stop, and an outage, the remote sites will not interpret the outage as a "failure."


Supplemental Materials

Value1734

Legacy ID



291216


Article URL http://www.symantec.com/docs/TECH53516


Terms of use for this information are found in Legal Notices