Data Repository vsql migration_status.sql script returns errors.

Products

• Network Observability
• CA Performance Management

Issue/Introduction

Adding nodes to a single node DR DB to build a cluster. The single node system was running for a long time back to the 22.2.x releases.

The final documented step to Add a Node to the Data Repository Cluster involves segmenting the database. The process is in the Segment Database Tables documentation topic.

Segmentation spreads the disk space usage load for DB tables around the cluster so it's shared among the cluster nodes. When running through that process an error is seen.

The following steps are taken, in the order listed.

 

1. Followed the add a node documentation up to the "Create the Buddy Projection and Mark the Cluster K-safety of 1" section.
2. Ran the "select export_objects..." command.
3. Edited the resulting sql file.
4. Ran the segment command with --zerotables flag.
5. Ran the edited vsql migration_status.sql command following the segement tables documentation. It shows this error.0

 

$ vsql -U dradmin -w drpass -f /home/dradmin/migration_status.sql
vsql:/home/dradmin/migration_status.sql:21: WARNING 4468:  Projection <scheme>.migration_status_b1_super> is not available for query processing.
Execute the select start_refresh() function to copy data into this projection.
         The projection must have a sufficient number of buddy projections and all nodes must be up before starting a refresh
CREATE PROJECTION
 MARK_DESIGN_KSAFE
----------------------
Marked design 1-safe
(1 row)

Environment

All supported DX NetOps Data Repository releases 24.3.7 and earlier.

Cause

The vsql run of migration_status.sql calls a segment.py script. That script was not updated to account for the changes in the latest Vertica release in use.

Resolution

DE165743 will resolve this with an updated segment.py script. It's scheduled for inclusion in the 24.3.8 release, which is an ETA and is subject to change.

Watch the Resolved Issues lists for the DE165743 entry to confirm the solution is included in 24.3.8 when it is released.

Additional Information

What impact is there running the now multi-node DR DB cluster without segmentation being completed?

• Until segmentation is completed, new DB tables will be segmented. Existing tables will remain unsegmented.
• Report Views will not be as fast as they would be using segmented tables.
• Unsegmented tables will use 1/3 more data per node than it would if it were segmented.

As long as the DB is properly set to KSafety=1 the cluster will behave normally. There will be a likely minor disk space consumption impact until the segmentation process is completed.