I have been trying random SQL work arounds, but only leads to minor shifts in the exception. See below:

java.lang.ArrayIndexOutOfBoundsException: 255
at org.mariadb.jdbc.client.impl.StandardReadableByteBuf.skipIdentifier(StandardReadableByteBuf.java:111)
at org.mariadb.jdbc.message.server.ColumnDefinitionPacket.<init>(ColumnDefinitionPacket.java:67)
at org.mariadb.jdbc.message.ClientMessage.readPacket(ClientMessage.java:198)
at org.mariadb.jdbc.client.impl.StandardClient.readPacket(StandardClient.java:815)
at org.mariadb.jdbc.client.impl.StandardClient.readResults(StandardClient.java:754)
at org.mariadb.jdbc.client.impl.StandardClient.readResponse(StandardClient.java:673)
at org.mariadb.jdbc.client.impl.StandardClient.execute(StandardClient.java:616)
at org.mariadb.jdbc.Statement.executeInternal(Statement.java:906)
at org.mariadb.jdbc.Statement.executeQuery(Statement.java:114)

Could you help giving some additional info, to try to reproduced the issue ? like:

• confirm this issue correspond to version 3.0.4
• indicate server type (mysql/mariadb/xpand/maxscale ... and version)
• could those queries concern specific DATATYPE like JSON/UUID ?

This is a query against MariaDB 10.5.7
The SQL is a Select Case, with two sub-Selects
It is replicated only within a large transaction (i.e. it cannot be replicated with a single SQL operation).
It does not occur using any MySQL Connector/J 5 or 8 versions.
We have not tested this with any other MariaDB Connector/J 3 version as 3.0.4 is the first to fix a LiquiBase issue that we have been waiting for.
I have asked internally to run a further test using the MariaDB Connector/J 2.7.5 driver and I should have a some result today.
The query does not involve JSON/UUID, here is sample SQL:
SELECT ifnull((select case data_string when 'S' then 1 when 'U' then 0 end from data_master where record_id = (select record_id from record_master where unit_id = 67010 and case_id = '2021ANG0410' and win_id = (select win_id from wins_master where schedule = 'DEC')) and field_id = (select field_id from field_master where field_name = 'approval_status')),0)

I would seem more a server issue than connectors.
Since it seems you can reproduce the problem, Is it possible to have wireshark log data that concern the moment where it fails ?

How can it be a server issue when the MySQL connector works, but the MariaDB connector does not?
I have no indication of an error in the MariaDB server log.
I can work on capturing a trace log.

Thank you, I appreciate your research. I have looking at Wireshark results and not seeing anything. Our test of the 2.7 driver was unsuccessful because it has other limitations that prevent us from using it.

Does that mean we could restructure our SQL and provide an alias to the sub-selects as a work around? Or is their a configuration on the MariaDB server to limit the size of server generated aliases?

Aliases are limited to 256 characters.
This occurs when column aliases length is > 250.

Aliases value are length encoded, and current implementation was expecting only 1 or 3 bytes length encoded prefix. Some queries might return a 9 byte length prefix even for a 251 length alias.

This will be corrected in next release

Correction is available using snapshot:

<repositories>

    <repository>

        <id>sonatype-nexus-snapshots</id>

        <name>Sonatype Nexus Snapshots</name>

        <url>https://oss.sonatype.org/content/repositories/snapshots</url>

    </repository>

</repositories>

<dependencies>

    <dependency>

        <groupId>org.mariadb.jdbc</groupId>

        <artifactId>mariadb-java-client</artifactId>

        <version>3.0.5-SNAPSHOT</version>

    </dependency>

</dependencies>