Can't start rac package on a node

[chuanpeng.wang]

The package run normally at first.
But the package can't start now.
The file of package had been modified.

# cmmodpkg -e -n odser1 ods1
cmmodpkg: Completed successfully on all packages specified

# cmrunpkg ods1
Running package ods1 on node odser1
Node odser1 is currently unable to run package ods1.
Unable to run package ods1 on node odser1
Check the syslog and pkg log files for more detailed information
cmrunpkg: Unable to start some package or package instances


cat pkg1.cntl.log
/etc/cmcluster/pkg1/pkg1.cntl: line 1803: integer: command not found
/etc/cmcluster/pkg1/pkg1.cntl: line 1848: print: command not found
/etc/cmcluster/pkg1/pkg1.cntl: line 575: integer: command not found
/etc/cmcluster/pkg1/pkg1.cntl: line 576: integer: command not found
/etc/cmcluster/pkg1/pkg1.cntl: line 577: integer: command not found
/etc/cmcluster/pkg1/pkg1.cntl: line 578: set: -A: invalid option

[chuanpeng.wang]

The file of package had NOT been modified.

[Steven E. Protter]

Shalom,

Check the environment and shell of the package pkg1.cntl

There appears to be a scripting error in there. The PATH variable is almost certainly wrong among other issues.

SEP

[chuanpeng.wang]

# grep -v '^#' pkg1.cntl|more



. ${SGCONFFILE:=/etc/cmcluster.conf}


PATH=$SGSBIN:/usr/bin:/usr/sbin:/etc:/bin

VGCHANGE="vgchange -a s" # Default

CVM_ACTIVATION_CMD="vxdg -g \$DiskGroup set activation=exclusivewrite"

VG[1]="vgods1"
VG[2]="vgods2"
VG[3]="vgods3"







DEACTIVATION_RETRY_COUNT=0


KILL_PROCESSES_ACCESSING_RAW_DEVICES="NO"

VXVOL="vxvol -g \$DiskGroup startall" # Default

FS_UMOUNT_COUNT=1

FS_MOUNT_RETRY_COUNT=0

CONCURRENT_VGCHANGE_OPERATIONS=1

CONCURRENT_FSCK_OPERATIONS=1

CONCURRENT_MOUNT_AND_UMOUNT_OPERATIONS=1







log_file=${SG_SCRIPT_LOG_FILE:-$0.log}



function customer_defined_run_cmds
{
: # do nothing instruction, because a function must contain some command.

test_return 51
}


function customer_defined_halt_cmds
{
: # do nothing instruction, because a function must contain some command.
test_return 52
}




EVFS_SCRIPT="${0%/*}/evfs_sg.${EVFS_SCRIPT_EXTENSION:-sh}"

function verify_evfs
{

if [[ -x $EVFS_SCRIPT ]]
then

$EVFS_SCRIPT $1 ${LV[@]}

#
# The evfs_sg.sh script has the return values as follow:
#
# 0 - success
# 1 - fails
#
test_return 54
fi
}

......
......

[chuanpeng.wang]

They had installed gcc in the period.
I don't know whether it is the cause.

[chuanpeng.wang]

the syslog:
Jan 10 17:07:37 odser1 cmcld[4099]: Request from node odser1 to enable node swit
ching for package ods1 on node odser1.
Jan 10 17:07:37 odser1 cmcld[4099]: Enabled package ods1 on node odser1.
Jan 10 17:07:37 odser1 cmcld[4099]: Request from node odser1 to start package od
s1 on node odser1.
Jan 10 17:07:37 odser1 cmcld[4099]: Executing '/etc/cmcluster/pkg1/pkg1.cntl st
art' for package ods1, as service PKG*48897.
Jan 10 17:07:37 odser1 cmcld[4099]: Service PKG*48897 terminated due to an exit(
2).
Jan 10 17:07:37 odser1 cmcld[4099]: Package ods1 run script exited with RESTART.

Jan 10 17:07:37 odser1 cmcld[4099]: Examine the file /etc/cmcluster/pkg1/pkg1.cn
tl.log for more details.
Jan 10 17:07:37 odser1 cmcld[4099]: Disabled node odser1 from running package od
s1.
Jan 10 17:07:37 odser1 cmcld[4099]: Package ods1 cannot run on this node because
switching has been disabled for this node

[Julio Yamawaki]

Hi,

It seems that your run command is using the wrong shell:

/etc/cmcluster/pkg1/pkg1.cntl: line 578: set: -A: invalid option

The correct shell is posix.
As this shell script is executed by root, I think you could check root's shell.

Regards,

[chuanpeng.wang]

I have check the shell is /sbin/sh

[Julio Yamawaki]

Hi,

Is it possible for you to post entire file pkg1.cntl?

Regards,

[chuanpeng.wang]

the customer mv the /usr/bin/sh to /usr/bin/sh.bak.

move it back, package can startup.