[resolved] TAC High Availability + MariaDB failover JDBC connection

[Anonymous]

Hello,
We have TAC (v6.1.1) up and running on dual servers, shared storage, commandline/jobservers on both, and we are running a master-master setup on MariaDB. When initially deploying TAC we simply pointed TAC to mariahost01 to ensure everything was working. Current connection URL:

jdbc:mariadb://mariahost01:3306/talend_administrator

The challenge we're facing now is that if mariahost01 goes down, there is nothing in place to reconnect to mariahost02. According to the MariaDB docs, the JDBC driver is capable of accepting failover parameters ( https://mariadb.com/kb/en/mariadb/about-mariadb-connector-j/#failover-parameters) but when using something like this

jdbc:mariadb:sequential://mariahost01:3306,mariahost02:3306/talend_administrator

in the URL string we get "Driver cannot understand url" error and fails the check. 
Questions:

has anyone successfully got this working? Or is there a better solution for this?
what version of the MariaDB JDBC driver ships with TAC 6.1.1?

Appreciate the help,
T.

[Anonymous]

Found the answer to the second question (which is always the way after you post in the forums, am I right?):

./tac/apache-tomcat/webapps/org.talend.administrator/WEB-INF/lib/mariadb-java-client-1.1.7.jar

MariaDB docs: sequential ... since 1.3.0
MariaDB current stable release: 1.5.2
*sigh*
@Talend - any chance this has been updated in an upcoming release or am I on my own in terms of trial & error? The HA documentation for TAC is terrible at best so any guidance on how to actually make it HA would be most welcome.

[Anonymous · Accepted solution]

Solution: download more current client and deployed to:

./tac/apache-tomcat/webapps/org.talend.administrator/WEB-INF/lib/mariadb-java-client-1.1.7.jar

Restarted TAC servers, checked job execution, then updated TAC database URL with:

jdbc:mariadb:sequential://mariahost01:3306,mariahost02:3306/talend_administrator

Passed check, saved it, checked job execution, and things appear to be working as expected. Next step will be to fail the first database host, but will let things run for a day or so to make sure we are stable with the new client.
@Talend - please consider shipping with a more recent MariaDB driver so that HA users can take advantage of the failover capabilities.