Apache Ignite cache empty, table has values

问题

I have a basic Ignite persistence setup working. For now, it's a single node running in server mode, from which I connect two clients. When I insert data via SQL from one of these clients ("Client 1" below), I can SELECT it again with SQL and get results back. From "Client 2" below, when I try to grab the cache which ought to have been created to represent this table, it is null.

The server runs the apacheignite image using the following Dockerfile:

server/Dockerfile:

FROM apacheignite/ignite

# for jdbc connection
EXPOSE 10800

EXPOSE 47100-47109
EXPOSE 47500-47509

# for rest api
EXPOSE 8080

WORKDIR /app

COPY ./config /app/config

RUN mv /opt/ignite/apache-ignite/libs/optional/ignite-aws /opt/ignite/apache-ignite/libs/

RUN mv /opt/ignite/apache-ignite/libs/optional/ignite-rest-http /opt/ignite/apache-ignite/libs/

CMD /opt/ignite/apache-ignite/bin/ignite.sh /app/config/ignite-config.xml -v

The relevant parts of the ignite-config.xml referenced here is as follows:

/app/config/ignite-config.xml

<bean class="org.apache.ignite.configuration.IgniteConfiguration">
        <property name="dataStorageConfiguration">
            <bean class="org.apache.ignite.configuration.DataStorageConfiguration">
                <property name="defaultDataRegionConfiguration">
                    <bean class="org.apache.ignite.configuration.DataRegionConfiguration">
                        <property name="persistenceEnabled" value="true"/>
                    </bean>
                </property>
            </bean>
        </property>
        <property name="discoverySpi">
            <bean class="org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi">
                <property name="ipFinder">
                    <bean class="org.apache.ignite.spi.discovery.tcp.ipfinder.vm.TcpDiscoveryVmIpFinder">
                        <property name="addresses">
                            <list>
                                <value>127.0.0.1:47500..47509</value>
                            </list>
                        </property>
                    </bean>
                </property>
            </bean>
        </property>
    </bean>

I've also made sure to activate the cluster on the Server container using since Persistence sets the cluster to inactive by default.

/opt/ignite/apache-ignite/bin/control.sh --activate

Client 1

Client 1 connects over the Thin JDBC driver and does some CREATE TABLEs and INSERTs. When I later SELECT in something like DBeaver, I can see this data.

This client also runs in docker, alongside the other in Compose. All it really does is call the following in a loop, where $file is a SQL file containing CREATE TABLE, INSERTs, etc.

/opt/ignite/apache-ignite/bin/sqlline.sh -u jdbc:ignite:thin://server:10800 -f $file;

One such SQL file looks like:

DROP TABLE IF EXISTS attributes;
DROP INDEX IF EXISTS idx_attributes_token;

CREATE TABLE attributes (
    token VARCHAR,
    attributeId LONG,
    name VARCHAR,
    PRIMARY KEY(token, attributeId)
) WITH "affinityKey=token";

CREATE INDEX idx_attributes_token ON attributes (token);

INSERT INTO attributes (token, attributeId, name) VALUES ('abc123', 123, 'some name');

Client 2

Client 2 connects to Ignite more or less as follows:

IgniteConfiguration igniteConfiguration = new IgniteConfiguration();
TcpDiscoverySpi discoverySpi = new TcpDiscoverySpi();
TcpDiscoveryMulticastIpFinder ipFinder = new TcpDiscoveryMulticastIpFinder();

ipFinder.setAddresses(igniteConfig.getAddresses()); // will be "localhost:10800" -- I map 10800:10800 when the "Server" container runs

discoverySpi.setIpFinder(ipFinder);
igniteConfiguration.setDiscoverySpi(discoverySpi);

Ignite ignite = Ignition.start(igniteConfiguration);

IgniteCache<Object, String> attributesCache = ignite.cache("SQL_PUBLIC_ATTRIBUTES"); // null

//  if i use ignite.getOrCreateCache("SQL_PUBLIC_ATTRIBUTES") instead then of course it's a cache of size 0.

Ultimately, I want to be able to grab hold of this cache and use the StreamVisitor described in the Ignite docs to perform operations on each. That would allegedly look like:

... 
IgniteDataStreamer<Object, AttributeWithGroup> attributeStreamer = ignite.dataStreamer(attributesCache.getName());

attributeStreamer.receiver(StreamVisitor.from((cache, entity) -> {
    Object key = entity.getKey();
    Attribute attribute = entity.getValue();
    // do some stuff
}));

...but the lambda never runs, because the cache is null.

Why is there this disconnect between what I can see in SQL and what I can see in the Cache?

回答1:

They way you are starting your "Client 2" it will start as a single server node with no caches.

You should also be using 47500 for discovery port and not 10800. 10800 is for thin clients.

Please also note that you have mismatching IpFinder's, Vm versus Multicast.

来源：https://stackoverflow.com/questions/55728025/apache-ignite-cache-empty-table-has-values

标签

Docker

ignite