Pakge Auto-run disable service guard

[Mutum]

I have success setting oracle datbase,using EMCT, and all runing fine.
When I halt the pakage in one node it's not strating automatically in the second node, I have issued the command #cmmodpke -e , to enable the auto-run, and start the pakge, while the pkage is runing, and veiw the cluster, I can see that pakge auto-run is enabled, but when I halt it again it's not strat auto- and I view the cluster it show again that pakage auto-run is disabed, in the pakage control file, pakage auto-run is set to Yes.
Any help will be apprecaited.

[Mark Nieuwboer]

Hi Muatasim,

This i normal. Because you give the cmhaltpkg the cluster sees you halted it. If you want to check if the packages is falling over you have to kill 1 of the processes of the software in the package.

good luck Mark

[Frank de Vries]

Indeed it is normal.

Another good test (other then to kill )
is to unplug the
network cable so long it will exceed
the time-out you specified in your
cluster configuration file.

The switch-over should then happen
over the fail-over lan.

Rgds,

[melvyn burnard]

This is designed behaviour.
When you issue the cmhaltpkg command this automatically disables global package switching.

[Mutum]

Thanks to all, okay getting auto-run disabled when pkg halted.

But I have kill on of the oracle datbase service, while pkg is running on the first node, pkg the pkg stop on the first node, but it's not able to failover to second node, it till me that pkg should not restart on node2, and then again I have to issue #cmmodpkg -e , and start it manually on the first node.

I don't know why it's not able to run on node2.

[RAC_1]

First enable package switching.
cmmodpkg -ev "pkg_name"

Then kill/stop monitor service or you can just the lan cable off. Now check if package fails over or not. If not whay does package log file says??

[Mutum]

Thanks RAC

This is not work also.
pkg log file says:
NSS2:Unable to strat DB_Service.
NSS2: ask NSS1 to disable PKG golobaly.

DB_SERVICE is the monitoring service I have defined in control and config pkg configuation .

[RAC_1]

Does following work??

cmrunpkg -v -n "other_node"

What does log files say??

[Mutum]

Hi RAC,

Also not work, when I isssue the cmd #cmrunpkg -v -n

it saying that Pkg should not restart on .
unable to start DB_SEVICE.
Then it will umount file system, and remove the IP.

[Henk Geurts]

hallo Muatasim.
please make sure service is configured without typo's in both .cntl and .conf file
and
make sure service is applied...
use the commands
cmcheckconf -P /etc/cmcluster//.conf
and
cmapplyconf -P /etc/cmcluster//.conf

then try to start the package again.....
good luck

[RAC_1]

cmmodpkg -ev -n "other_node" "pkg_name"

Now, cmrunpkg -v -n "other_node" "pkg_name"

What does log files say??

[melvyn burnard]

It sounds like you still have LOCAL package switching disabled on the second node.

What does cmviewcl -v -p tell you for the switching on the nodes?

[Mutum]

Helo Henk,

I didn't get you, what you mean typo's ?

#cmviewcl -v -P

it show Auto-run is enabled when pkg is runing.
how to enable local switching ?

[Henk Geurts]

i meant that syntax-errors (typing errors) can cause service not being started in the package ...

can you post your output of
cmviewcl -v -p ?
it should be something like :
PACKAGE STATUS STATE AUTO_RUN NODE
xxxxx up running enabled oraxxxx

Policy_Parameters:
POLICY_NAME CONFIGURED_VALUE
Failover configured_node
Failback manual

Script_Parameters:
ITEM STATUS MAX_RESTARTS RESTARTS NAME
Service up 3 0 xxxxxx_mon
Subnet up 10.xxx.xx.xx

Node_Switching_Parameters:
NODE_TYPE STATUS SWITCHING NAME
Primary up enabled oraxxxxx (current)
Alternate up enabled orayyyyy

[Mutum]

Dear All

pls see the log file of the second node

[melvyn burnard · Accepted solution]

Well from the log file:

> /etc/cmcluster/pkg/MXP/MXP.cntl[2]: > /etc/cmcluster/pkg/MXP/toolkit.sh: Autorisation d'excution refuse. <<<< This indiocates it is not executable?

> dc 19 21:59:29 - Node "NSS2": Starting service DB_SERVICE using "/etc/cmcluster/pkg/MXP/toolkit.sh monitor"
usage: cmrunserv [-v] [-r number_of_restarts | -R] service_name service_command_string

So the above shows a useage error in your script!


ERROR: Function start_services
ERROR: Failed to start service DB_SERVICE
dc 19 21:59:29 - Node "NSS2": Halting service DB_SERVICE
cmhaltserv : Service name DB_SERVICE is not running. <<< this tells you that because it could not start the service from above, it will halt th epackage!!

/etc/cmcluster/pkg/MXP/MXP.cntl[2]: /etc/cmcluster/pkg/MXP/toolkit.sh: Autorisation d'excution refuse. <<
This failure will automatically DISABLE the package from LOCAL switching, i.e. turn of the ability for this node to run the package.

Compare the toolkit.sh file between the working an dnon-working nodes.

[Henk Geurts]

follow Melvin's suggestion.
check for execute bits on the monitorscript.

[RAC_1]

Just eager to know, how does cmclconfig file looks on both nodes??
ll on it and compare the sizes and cksum.

Also are config and control files and monitoring scrips same on both nodes??

I think you may need to re-appy the changes to this node from working node.

cmapplyconf -P pkg_name -n other_node

[Mutum]

melvyn ,

Thanks problem is solved, as per your advise.
I have change the execute permission for toolkit.sh on the second node, and also transfer the control fine again to second node, now all runiing fine.

Thanks

[Mutum]

Thanks to All,

Correcting the execute permission on the conntrol file and monitoring script (toolkit.sh) on the second node.