How do retries work for data connectors?

[Lawlor-Bagcal, Kyle]

Hello,

I want to better understand how retries of a data connector work in
shibboleth.

*Example code:*

>From the attributes resolver file:

    <!--      Primary    -->
    <DataConnector id="myDatabase" noRetryDelay="PT30S"
xsi:type="RelationalDatabase"
validatorRef="shibboleth.NonFailFastValidator">
        <BeanManagedConnection>OracleDBdataSource</BeanManagedConnection>
        <FailoverDataConnector ref="FailoverAWSmysql" />
        <QueryTemplate>
            .... redacted SQL query for getting some attributes...
        </QueryTemplate>

        <Column columnName="...." attributeID="...." />
        <ResultCache elementTimeToLive="PT60S"/>
    </DataConnector>

    <!--      Failover  -->
    <DataConnector id="FailoverAWSmysql" xsi:type="RelationalDatabase"
validatorRef="shibboleth.NonFailFastValidator">
        <BeanManagedConnection>MySQLdataSource</BeanManagedConnection>
        <QueryTemplate>
            .... redacted
        </QueryTemplate>

        <Column columnName="...." attributeID="...." />
        .... redacted ....
        <ResultCache elementTimeToLive="PT60S"/>
    </DataConnector>

from conf/global.xml:

    <bean id="hikariConfig" class="com.zaxxer.hikari.HikariConfig">
        <property name="poolName" value="springHikariCP" />
        <property name="connectionTimeout" value="2000" />
        <property name="validationTimeout" value="1000" />
        <property name="maximumPoolSize" value="20" />
        <property name="dataSourceClassName"
value="oracle.jdbc.pool.OracleDataSource" />
        <property name="dataSourceProperties">
            <props>
                .... redacted ....
            </props>
        </property>
    </bean>

    <bean id="hikariConfigMySQL" class="com.zaxxer.hikari.HikariConfig">
        <property name="poolName" value="springHikariCPmysql" />
        <property name="connectionTimeout" value="10000" />
        <property name="maximumPoolSize" value="20" />
        <property name="dataSourceClassName"
value="com.mysql.jdbc.jdbc2.optional.MysqlDataSource" />
        <property name="dataSourceProperties">
            <props>
                .... redacted ....
            </props>
        </property>
    </bean>

*Scenario:*

1. Let's say the shibboleth app is trying to get attributes from the
primary data connector.
2. Assume a thread grabs a connection from the primary HikariCP but upon
execution the ResolutionException is raised.

*Questions:*

   - After #2 occurs, what happens next? Does the thread pause for 30
   seconds (noRetryDelay) to retry the primary connector? Or does it
   immediately go to the failover data connector and then pending the result
   of that lookup, it may or may not retry the entire path?
   - How many times will the data connector retry? Is there a limit? Where
   is the limit set? Is it hard coded?
   - Is it true that the only exception that the retry or failover logic
   responds to is the ResolutionException
   <https://git.shibboleth.net/view/?p=java-identity-provider.git;a=blob;f=idp-attribute-resolver-impl/src/main/java/net/shibboleth/idp/attribute/resolver/dc/rdbms/impl/RDBMSDataConnector.java;hb=28bccab21a065a57c4f03cd34da4011b5b9dca9f#l154>
   (for data connectors)?

If we can keep the answers scoped to my example code above that would be
helpful for me to understand it better.

Thanks,
-- 

Kyle Lawlor-Bagcal, Web and Mobile Team Lead

Teachers College Information Technology (TCIT)

Teachers College | Columbia University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shibboleth.net/pipermail/users/attachments/20210330/8a7e681f/attachment.htm>