Limitations and known issues for Db2 Big SQL

The following limitations and known issues apply to Db2 Big SQL.

Unable to provision a Db2 Big SQL instance because the `c-bigsql-<instanceid>-db2u-restore-morph` job fails

Applies to: 4.6.4

When you provision a Db2 Big SQL service instance, the c-bigsql-<instanceid>-db2u-restore-morph job might fail repeatedly, and the Db2 Big SQL instance is not created.

Symptoms

The c-bigsql-<instanceid>-db2u-restore-morph job keeps failing during Db2 Big SQL provisioning and no Db2 Big SQL instance is created.

The standard output logs for the c-bigsql-<instanceid>-db2u-restore-morph pod show an error message similar to the following example:

+ echo 'Running db2start'
Running db2start
+ [[ 0 -le 1 ]]
+ RC=0
+ db2start
<timestamp>     0   0   SQL0901N  The SQL statement or command failed because of a database system error. (Reason "".)
SQL1032N  No start database manager command was issued.  SQLSTATE=57019
+ sleep 5
++ db2gcf -s
++ awk -F : '{print $2}'
++ sed 's/^ *//; s/ *$//; /^$/d'
+ db2state=Operable
+ [[ Operable == \A\v\a\i\l\a\b\l\e ]]
+ echo '* ERROR: Failed to start Db2 successfully'
* ERROR: Failed to start Db2 successfully
+ db2_kill -all
db2nkill: DB2 member 0 with PID 4294967295 does not exist.
+ rah ipclean -a
Application ipclean: Removing all IPC resources for db2inst1(500)
c-bigsql-<instanceid>-db2u-0.c-bigsql-<instanceid>-db2u-internal: ipclean -a completed ok
+ RC=1
+ (( n++ ))
+ [[ 1 -le 1 ]]
+ RC=0
+ db2start
<timestamp>     0   0   SQL0901N  The SQL statement or command failed because of a database system error. (Reason "".)
SQL1032N  No start database manager command was issued.  SQLSTATE=57019
+ sleep 5
++ db2gcf -s
++ awk -F : '{print $2}'
++ sed 's/^ *//; s/ *$//; /^$/d'
+ db2state=Operable
+ [[ Operable == \A\v\a\i\l\a\b\l\e ]]
+ echo '* ERROR: Failed to start Db2 successfully'
* ERROR: Failed to start Db2 successfully
+ db2_kill -all
db2nkill: DB2 member 0 with PID 4294967295 does not exist.
+ rah ipclean -a
Application ipclean: Removing all IPC resources for db2inst1(500)
c-bigsql-<instanceid>-db2u-0.c-bigsql-<instanceid>-db2u-internal: ipclean -a completed ok
+ RC=1
+ (( n++ ))
+ [[ 2 -le 1 ]]
+ return 1
+ echo 'Failed to start Db2'
Failed to start Db2
+ return 1
+ echo '* ERROR (Pre Restore Stage): Failed to successfully set registry variables. Failed at pre restore stage.'
* ERROR (Pre Restore Stage): Failed to successfully set registry variables. Failed at pre restore stage.
+ exit 1
+ return 1
+ rc=1
+ log '' 'Completed /db2u/db2u_restore_morph.sh  with RC code 1. Total Elapsed Time: 41'
+ fun=
+ msg='Completed /db2u/db2u_restore_morph.sh  with RC code 1. Total Elapsed Time: 41'
+ LINE=53
++ date +%F.%T.%N
+ printf '%.23s @ Line(%-5.5d) @ Fun %32.32s: ' 2023-04-03.21:48:24.303556217 53 ''
+ echo Completed /db2u/db2u_restore_morph.sh with RC code 1. Total Elapsed Time: 41
2023-04-03.21:48:24.303 @ Line(00053) @ Fun                                 : + exit 1
Completed /db2u/db2u_restore_morph.sh with RC code 1. Total Elapsed Time: 41
command terminated with exit code 1
+ exit 1

When you run the following command, you get the output

Global instance memory (% or 4KB)
(INSTANCE_MEMORY) = 80

Note: To get the Db2 Big SQL instance ID, run

oc get cm -l component=db2bigsql -o custom-columns="Instance Id:{.data.instance_id},Instance Name:{.data.instance_name},Created:{.metadata.creationTimestamp}"

oc exec -it c-bigsql-<instanceid>-db2u-0 -- bash -c "su - db2inst1 -c 'db2 get dbm cfg' | grep -i INSTANCE_MEMORY"

When you run the following command, only the Db2 Big SQL head and worker pods (c-bigsql-<instanceid>-db2u-x) and the failed c-bigsql-<instanceid>-restore-morph job pods are shown.
```
oc get pods | grep -i c-bigsql-<instanceid>-db2u
```

Workaround

To resolve the issue, decrease the Db2 instance memory.

Decrease the Db2 instance memory setting in the Db2 Big SQL head pod.

oc rsh c-bigsql-<instanceid>-db2u-0 bash
su - db2inst1
db2 update dbm cfg using instance_memory 2000000

Save a copy of the c-bigsql-<instanceid>-restore-morph job YAML file.

oc get job c-bigsql-<instanceid>-restore-morph -o yaml | grep -v controller-uid > c-bigsql-<instanceid>-restore-morph.yaml

Delete the existing c-bigsql-<instanceid>-restore-morph job from the OpenShift® cluster.
```
oc delete job c-bigsql-<instanceid>-restore-morph
```
Recreate the c-bigsql-<instanceid>-restore-morph job based on the YAML file from step 2.
```
oc project ${PROJECT_CPD_INSTANCE}
oc apply -f  c-bigsql-<instanceid>-restore-morph.yaml
```
Important: Make sure that the project (namespace) that is stored in the c-bigsql-<instanceid>-restore-morph.yaml file is $PROJECT_CPD_INSTANCE.

After step 4, the c-bigsql-<instanceid>-restore-morph job pod is created.

If the issue is resolved, the c-bigsql-<instanceid>-restore-morph job pod finishes successfully, and more Db2 Big SQL pods are created.

Unable to query some tables after upgrading Cloud Pak for Data 4.0.x to 4.5.x or 4.6.0

Applies to: 4.6.0

Fixed in: 4.6.2

Queries on tables that contain certain data types can fail. You see an error in the bigsql.log file like the following example:

com.thirdparty.cimp.catalog.TableLoadingException: The table definition statement failed because some 
functionality was specified in the table definition that is not supported with the table type. Unsupported 
functionality: "\"timestamp(6)\" DATATYPE".

This problem occurs with tables that contain the following data types:

Columns where the data type is timestamp with a fixed precision. For example, timestamp(3).
Columns where the data type is decimal, but no precision or scale is specified. For example, decimal or decimal(10).
Columns that are a complex data type. For example, map, array, or struct.

This problem applies only to Db2 Big SQL instances in Cloud Pak for Data 4.0.x that are upgraded to Cloud Pak for Data 4.5.x or 4.6.0. Instances that are created in Cloud Pak for Data 4.5.x or upgraded to Cloud Pak for Data 4.6.2 or later are not affected.

For more information about the problem and how to resolve it, see the technote Unable to query some tables after upgrading Db2 Big SQL on Cloud Pak for Data 4.0.x to 4.5.x or 4.6.0.

INSERT operations on partitioned tables fail with an SQL5105N error

Applies to: 4.6.0

Fixed in: 4.6.2

INSERT operations on partitioned tables that contain a null value for a partitioning column might fail with an SQL5105N error. The detailed log entry shows an exception that is raised by the writer process.

To work around this problem, avoid inserting rows with null values in PARTITIONED BY columns.

Limitations

HBase tables are not supported.
Transactional tables are not supported.
Tables that are created in Db2 Big SQL are not propagated to Apache Atlas if Atlas is installed on the remote Cloudera Data Platform (CDP) Private Cloud Base cluster.
The LOAD command is not supported.

Limitations and known issues for Db2 Big SQL

Unable to provision a Db2 Big SQL instance because the c-bigsql-<instanceid>-db2u-restore-morph job fails

Unable to query some tables after upgrading Cloud Pak for Data 4.0.x to 4.5.x or 4.6.0

INSERT operations on partitioned tables fail with an SQL5105N error

Limitations

Unable to provision a Db2 Big SQL instance because the `c-bigsql-<instanceid>-db2u-restore-morph` job fails