Write failures, 0.80.2 on cluster using ec2 testing scripts.

[Ben Hardy]

Hi folks,

I'm using Hadoop to write into Voldemort, everything running on EC2 but am seeing the following exception pop up:

voldemort.VoldemortException: 3 metadata refresh attempts failed.

at voldemort.client.DefaultStoreClient.put(DefaultStoreClient.java:214)

at voldemort.client.DefaultStoreClient.put(DefaultStoreClient.java:193)

I am using the contrib/ec2-testing scripts to create the instances and configure the cluster. The storage being used is memory-only. There are 10 nodes in the cluster.

My config is as follows. I am guessing that I am getting something wrong here.

server.properties:

node.id=0

voldemort.home=/root/voldemort

enable.bdb.engine=false

enable.memory.engine=true

enable.cache.engine=false

max.threads=1500

stores.xml:

<stores>

    <store>

        <name>user_joined</name>

        <replication-factor>3</replication-factor>

        <preferred-reads>2</preferred-reads>

        <required-reads>1</required-reads>

        <preferred-writes>3</preferred-writes>

        <required-writes>1</required-writes>

        <persistence>memory</persistence>

        <routing>client</routing>

        <key-serializer>

            <type>string</type>

            <schema-info>utf8</schema-info>

        </key-serializer>

        <value-serializer>

            <type>identity</type>

        </value-serializer>

    </store>

</stores>

I'm using the contrib/ec2-testing script for cluster.xml generation in the following way:

$VOLDY_EC2_BIN/voldemort-clustergenerator.sh \

--clustername "Voldemort_User_Data_Cache" \

--hostnames $VOLDEMORT_HOME/config/voldy-ec2-instances \

--partitions 15  >$VOLDEMORT_HOME/config/cluster.xml

Any clues would be appreciated. There's not much Googleable info about that exception and I didn't understand the source code well enough to figure out what the exception meant.

Thanks

Ben

[Ben Hardy]

I have seen the error of my ways. This is course was a total newbie
configuration error. My node ids I had put in my server.properties on
each node did not correspond to the node ids in the cluster.xml.