search cancel

Pre-upgrade check if you have mitigated log4j in the Kafka section of the Data Repository in a multi node cluster

book

Article ID: 238208

calendar_today

Updated On:

Products

DX NetOps CA Performance Management - Usage and Administration

Issue/Introduction

To detect and avoid a potential upgrade issue if you have removed Kafka in a multi-node Data Repository cluster as noted in KB https://knowledge.broadcom.com/external/article?articleId=230262

Cause

If a node was down when Kafka was removed, it was only partially removed and will need to be cleaned up prior to an upgrade.  If it is not cleaned up, the upgrade will fail.

Environment

Multi-node DR installations that have removed Kafka to mitigate the log4j vulnerability

Resolution

1. Access adminTools and validate ALL nodes are UP

2. Still in the base menu of adminTools, choose 2 Connect to Database

3. Run the below call, if it returns results, proceed to step 4, if there are no results you are clear:

select * from V_MONITOR.USER_LIBRARIES where schema_name='public' and lib_name='KafkaLib';

Example:

dradmin=> select * from V_MONITOR.USER_LIBRARIES where schema_name='public' and lib_name='KafkaLib';
 schema_name | lib_name |   lib_oid   | author |   owner_id   |            lib_file_name             |       md5_sum       | sdk_version |          revision          |        lib_build_tag         | lib_version | lib_sdk_version |    source_url    |      description      | licenses_required |         signature          | dependencies | is_valid |         sal_storage_id
-------------+----------+-------------------+---------+-------------------+--------------------------------------------------------------+----------------------------------+-------------+----------------------------------------------+----------------------------------------------+-------------+-----------------+-------------------------+---------------------------------+-------------------+----------------------------------------------+--------------+----------+--------------------------------------------------
 public   | KafkaLib | 45035996273820854 | Vertica | 45035996273704962 | KafkaLib_02d313c660dad61c6e74732c4db7a71c00a000000001c4b6.so | ffb921497149b29b87f6e0b4cb2d3309 | 10.1.0   | releases/VER_10_1_RELEASE_BUILD_1_0_20210413 | releases/VER_10_1_RELEASE_BUILD_1_0_20210413 | v10.1.1-0  | 10.1.0     | http://www.vertica.com/ | Kafka streaming load and export |          | releases/VER_10_1_RELEASE_BUILD_1_0_20210413 |       | t    | 02d313c660dad61c6e74732c4db7a71c00a000000001c4b6

4. If step #3 does show the results in the example, ensure one last time that all nodes are UP:

select node_name, node_state from nodes;

Example:

dradmin=> select node_name, node_state from nodes;
     node_name     | node_state
-------------------+------------
 v_drdata_node0001 | UP
 v_drdata_node0002 | UP
 v_drdata_node0003 | UP
(3 rows)

5. If all nodes are still UP, drop the Kafka package:

DROP LIBRARY public.KafkaLib CASCADE;

6. Re-run step #3 to ensure there are no more results shown

Additional Information

This can be done in advance of the upgrade attempt.