Graceful shutdown

Provides detailed instructions to put Data Cataloging in idle state on an OpenShift® environment.

About this task

The procedure might be helpful in situations where Data Cataloging is behaving unexpectedly or there is a maintenance window of not more than one hour.

Procedure

Stop or wait for all running scans and policies to stop. Do not skip this step until everything looks stopped.
Do oc login to your OpenShift Container Platform cluster.
Do oc project <project> to select namespace where Data Cataloging service is installed.

Stop all pods and set replicas to zero for the following deployments. and respect the following order:

Note: Follow the suggested sequence.

Consumers

Run the following command to get the list of the deployments and filter by consumer.

$oc get deployments |grep consumer

Example output:

isd-consumer-ceph-le                    10/10   10           10          6d20h
isd-consumer-cos-le                     10/10   10           10          6d20h
isd-consumer-cos-scan                   10/10   10           10          6d20h
isd-consumer-file-scan                  10/10   10           10          6d20h
isd-consumer-protect-scan               10/10   10           10          6d20h
isd-consumer-scale-le                   10/10   10           10          6d20h
isd-consumer-scale-scan                 10/10   10           10          6d20h

Run the following to command to scale down replicas to 0 for every consumer deployment.

$oc scale --replicas=0 deployment <deployment_name>

Example output:

$oc scale --replicas=0 deployment isd-consumer-file-scan
deployment.apps/ isd-consumer-file-scan scaled

Ensure you scale down replicas to 0 for every consumer deployment.

Example output:

$oc get deployments|grep consumer-cos-scan
isd-consumer-file-scan                   0/0     0            0           6d20h

Producers

Run the following command to get the list of the deployments and filter by producer.

$oc get deployments |grep producer

Example output:

isd-producer-ceph-le                    1/1     1            1           6d20h
isd-producer-cos-le                     1/1     1            1           6d20h
isd-producer-cos-scan                   1/1     1            1           6d20h
isd-producer-file-scan                  1/1     1            1           6d20h
isd-producer-protect-scan               1/1     1            1           6d20h
isd-producer-scale-le                   1/1     1            1           6d20h
isd-producer-scale-scan                 1/1     1            1           6d20h

Run the following to command to scale down replicas to 0 for every producer deployment.
Important: Ensure that you scale down replicas to 0 for every producer deployment.
```
$oc scale --replicas=0 deployment <deployment_name>
```
Example output:
```
$oc scale --replicas=0 deployment isd-producer-file-scan
deployment.apps/ isd-producer-file-scan scaled
```

Ensure that you have scale down replicas to 0 for every producer deployment.

Example output:


$oc get deployments|grep producer-cos-scan
isd-producer-cos-scan                   0/0     0            0           6d20h

Connection manager
1. Run the following command to get the list of the statefulsets and filter by connection manager.
```
$oc get statefulsets |grep connmgr
```
  Example output:
  $oc get statefulsets |grep connmgr connmgr 0/0 0 0 6d20h
2. Run the following to command to scale down replicas to 0 for every connection manager statefulset.
  Important: Ensure that you have scale down replicas to 0 for every connection manager statefulset.
```
$oc scale --replicas=0 statefulset
```
  Example output:
  $oc scale --replicas=0 statefulset connmgr statefulset.apps/ connmgr scaled
3. Ensure that you have scale down replicas to 0 for every connection manager statefulset.
  Example output:
  $oc get statefulsets|grep connmgr connmgr 0/0 0 0 6d20h
Check whether db2whrest stopped inserting data properly

Ensure that no database activity is present before stopping db2whrest. It is necessary to check that specific csv files created are empty, which means there are no pending jobs being processed.
1. Run the following command to go to db2whrest.
```
oc rsh deployment/isd-db2whrest
```
  Now, list all CSV files related to merging that are empty.
  For example:
```
$for file in $(find /gpfs/gpfs0/db2wh/home/bluadmin -name *LOAD*.csv); do file $file; done |grep data
/gpfs/gpfs0/db2wh/home/bluadmin/ACOG_LOAD4.csv: data
/gpfs/gpfs0/db2wh/home/bluadmin/ACOG_LOAD1.csv: data
$
```
2. Wait for a few seconds and try again until you get an empty list.

DB2WHRest

Run the following command to get the list of the deployments and filter by db2whrest.

$oc get deployments |grep db2whrest

Example output:

$oc get deployments |grep db2whrest
isd-db2whrest                    1/1     1            1           6d20h

Run the following command to scale down replicas to 0 for every db2whrest deployment.

$oc scale --replicas=0 deployment <deployment_name>

Example output:

$oc scale --replicas=0 deployment db2whrest
deployment.apps/ db2whrest scaled

Ensure that all scale down replicas are 0 for every db2whrest deployment.

Example output:

$oc get deployments|grep db2whrest
isd-db2whrest                   0/0     0            0           6d20h

Stop Db2 instance. For procedure, see Stopping and starting a Db2 instance.
After checking that everything is down, Data Cataloging is successfully in an idle state. It is recommended to be in this state for a short period, not more than one hour. Warning messages on other Data Cataloging pods are expected.