Error from JDBC - COPY: Wrong Header size

Morning,

Looking at vertica log:

2023-07-06 08:37:26.039 Init Session:0x7ff4ef7db700-a0000000632a5a @v_xxx_node0001: 22V04/2767: COPY: Wrong Header size 501. Expected 505
LOCATION: parseHeader, /data/jenkins/workspace/RE-ReleaseBuilds/RE-Knuckleboom_2/server/vertica/EE/UDL/ParseFormats.h:729

My personal opinion is that data for loading are not formed correctly. Parser is not reporting error properly.
Would be very nice if developer from client driver team comment, what is the source of this error. How to find what is wrong with data.

Thank you
Sergey

Hi, I've asked client driver team to investigate. However, I know that JDBC converts certain batch loads (preparedStatement, or SQL COPY) into something like a COPY LOCAL. It would help if you could share the syntax of the command (COPY? PreparedStatement? something else?) that causes this and any sample data - particularly if there are UTF-8 or Unicode strings, or NUMERIC/DECIMAL (e.g. BigInteger, BigDecimal) types in the Java. Open a support case if necessary, or just email me if you don't want to post here.

Hi,

Thank you very much for looking.

Here is SQL that causes error:

COPY schema.table FROM local STDIN UNCOMPRESSED NATIVE EXCEPTIONS '/tmp/230705BA_0_RejectedReason' REJECTED DATA '/tmp/230705BA_0_RejectedData' AUTO

Table has 125 columns of types Boolean, Char, Integer, Timestamp, Varchar datatypes. No NUMERIC/DECIMAL.

I am very unsure about Unicode strings, that needs investigating.

There is not sample data readily available - app getting data from network, do some internal ETL, and load directly into Vertica. To get sample data, app has to be modified to produce it.

I will look on app code, how it loads - using prepared statement/bulk insert or using stream.

FYI app display interesting behaviour. It works in many instances in parallel without problems. If app throw JDBC error that we are discussing, it never recover and continue to throw same error. App is discarding connection that get error and make new connection, and this does not help - same error show up again. Restarting app - i.e. restarting JVM - fixes problem. Another thing - this JDBC error appear on loading to only one table, loads to many other tables never gets problem.

Sergey