1. Support
  2. Knowledge base
  3. 100001320

How to change the "gab_isolate_time" on Linux

Article: 100001320
Last Published: 2022-02-27
Ratings: 0 0
Product(s): InfoScale & Storage Foundation

Problem

How to change the "gab_isolate_time" on Linux

 

Solution

The issue:
Due to the GAB isolation timeout parameter, HAD (High Availability Daemon) may bring the cluster down when there is a lack of system resources and the system is less responsive.
The default " gab_isolate_time" timeout is 120 seconds.
 
To resolve the issue:
Increasing the timeout may alleviate the issue.
The " gab_isolate_time" timeout parameter is system specific, so must be changed on each node manually.
 
To display the current timeout:
#gabconfig -e
GAB Kernel Tunables:
============================
numnids          : 32
numports          :32
flowctrl          :5
logbufsize        :48100
msglogsize        :256
logflag          : 53
isolate_time      :120000        <<<<<<
kill_ntries      : 5
conn_wait        : 12
 
Note " isolate_time" is displayed as 120000 which is equivalent to 120 seconds.
 
The file that will contain the parameter is /etc/sysconfig/gab.
 
An example unmodified "gab" file:
#cat gab
#This file is sourced from /sbin/init.d/gab.
#
#Set the two environment variables below as follows:
#
#      1 = start or stop gab
#      0 = do not start or stop gab
#
 
GAB_START=1
GAB_STOP=1
 
1) First, make a backup copy of the file:
#cp /etc/sysconfig/gab /etc/sysconfig/gab.<date>
 
2) Add the value for " GAB_ISOLATE_TIME" to the file with 'vi' or similar editor.
If the file /etc/sysconfig/gab doesn't exist, create the file using the example below as a guide.
Example of a modified file, where the value has been modified to 150 seconds:
#This file is sourced from /sbin/init.d/gab.
#
#Set the two environment variables below as follows:
#
#      1 = start or stop gab
#      0 = do not start or stop gab
#
 
GAB_START=1
GAB_STOP=1
GAB_ISOLATE_TIME=150000
 
3) Restart GAB.
This may be accomplished by one of two methods-
 
a) Reboot the node
 
-or-
 
b) Completely stop VCS and unload the modules, then reload the modules and restart VCS.
- Make the configuration writable
#haconf -makerw
- Freeze the Service Groups on each system:
#hasys -freeze <system> -persistent
- Save the configuration
#haconf -dump -makero
- Stop the HA daemon
#hastop -all -force
- Stop clustering on each node:
#vxfenconfig -U
#gabconfig -U
#lltconfig -U
#/etc/init.d/vxfen stop
#/etc/init.d/gab stop
#/etc/init.d/llt stop
- Start clustering on each node:
#/etc/init.d/llt start
#/etc/init.d/gab start
#/etc/init.d/vxfen start
#hastart
 
- Verify the new " gab_isolate_time" value has taken effect:
#gabconfig -e
 
The output of  ' gabconfig -e' should show the new value.
 
- Unfreeze the Service Groups on each system:
#haconf -makerw
#hasys -unfreeze <system> -persistent
- Save the configuration
#haconf -dump -makero

Was this content helpful?

Article Languages