Database Reference
In-Depth Information
then a version 2.0 upgrade like the one mentioned previously: that is, the sequence
would be version 1.1.x to 1.2.9 to 2.0.7 and finally to 2.1.
A few of the configuration changes in version 2.0 release are
•
The property
index_interval
has been moved to the table level
and is no longer available in
cassandra.yaml
.
•
Virtual nodes (e.g.,
num_tokens
) are enabled by default with 2.0 and
later versions whereas with earlier versions it was disabled.
•
Java version 7 must be installed for 2.0 and later releases.
Now that we have discussed backward compatibility, let's see how to perform a
version upgrade on Cassandra nodes.
Performing an Upgrade with a Rolling Restart
Using the scenario mentioned in the preceding section, let's walk through the steps for
performing the sequential upgrade:
1.
Before we start the rolling upgrade, we should take care of the above-
mentioned changes and take a backup of all configurations belonging
to the previous Cassandra version and data. For data backup we can
create a snapshot as follows:
vivek@vivek-Vostro-3560:~$CASSANDRA_HOME$ bin/
nodetool -h localhost snapshot
twitter_keyspace
In the preceding command, please replace
$CASSANDRA_HOME$
with the currently installed Cassandra version (e.g.,
$apache-cas-
sandra-1.1.6$
).
In the preceding command, we are creating a snapshot for keyspace
twitter_keyspace
.
2.
Please make sure to remove all dead nodes from the running Cas-
sandra cluster before downloading the next version, 1.2.9 in this case.
3.
Then run the following command: