Check-list for a iAS production node

Quick guide:

• Go directly to validation page and check if the necessary container on the stand-by machine is running
• Make the DNS switch
• Update the banner (.bash_profile) on the host
• If is not a temporary solution, stop all unecessary containers running on that node by dcmctl stop -co <container_name>
• Update machine list on excel file at /afs/cern.ch/project/dbgroup/www/private/ps/

All steps:

Before putting a spare machine in a production environment, don't forget to check the following list in order to see if the follows the configuration.

• Stop application server
• Check existing containers
• Check datasources
• Change init scripts
• Start application server
• Check if endpoints are reachable
• Modify banner
• Put machine in production (DNS switch)
• Modify banner in new spare machine

The commands should be performed from the machine logged in as oracle.

1. Stop application server

To stop application server use:

sudo /etc/init.d/oracle_ias stop

If not all opmn processes are killed, then you can either kill them or just reboot (send an e-mail to Computer.Operations@cern.ch and Sysadmin.Support@cern.ch

or phone 75011 before) the machine:

sudo reboot

Note: After a reboot, the HTTP Server and the VOs that were up and running are automatically restarted. Don't forget to stop them.

2. Check Existing containers

- Check: existing containers and their state:

dcmctl getstate -v

- Check: all with the "In Sync Status" as True. If not, try to run:

dcmctl updateconfig -v -d

- Check: All containers should have 2 applications (LRC and RMC).

dcmctl listapplications -co cms

- If need to undeploy and deploy web-apps.

3. Check Datasources

- Check: datasources point to the right database host and sid, for each VO you should check the file: $ORACLE_HOME/j2ee/<VO>/config/data-sources.xml

- Check: Exists one <data-source> for each catalog (LRC and RMC).

- in the url attributes it specify the database host and the sid. In the example bellow:
url="jdbc:oracle:thin:@lxshare071d:1521:rls1"

- in the username and password attributes you can check if they correspond to the existing database users.

- Compare the information with the one in the Excel sheet. If necessary to change the data-sources check the instructions to Datasource Reconfiguration

4. Change init scripts

Is necessary to edit the file /etc/iastab and put a Y in front of the containers should run on this machine. For example:

home:N
alice:N
atlas:Y
cms:N
lhcb:N

5. Start application server

sudo /etc/init.d/oracle_ias start

If that does not work you can always run:

dcmctl start -co <container>

6. Check if endpoints are reachable

To validate please use the Validation page in the RLS installations and deployment section.

8. Modify the banner at interactive login

Update the file .bash_profile file with the correct message, setting:

- correct VO and alias

- "PRODUCTION MACHINE" instead of "STAND-BY MACHINE"

It should look something like:

STATUS: OS installed after 16.Oct.2003
iAS installed after 17.Oct.2003 (manjo)

installed RLS 2.0.2: alice, atlas, cms, dteam, lhcb
running endpoints: alice LRC, alice RMC
-> lxshare0217 - PRODUCTION MACHINE - rlsalice <-

9. Put machine in production

To change with a currently production machine, use DNS switch.

10. Modify the banner on the new spare machine

If you switch the machine, don' t forget to change the banner in .bash_profile in the old machine to SPARE MACHINE