Support Forums : DataStax

DataStax Support Forums » DataStax Enterprise

Building Brisk from source to update Cassandra to 0.8.2

(10 posts) (5 voices)

Started 1 year ago by t0mas
Latest reply from gibrown
This topic is not resolved

Tags:

t0mas
Member

I have a few problems with Brisk-1.0-beta2 because it uses Cassandra 0.8.1 which has some bugs. These bugs are fixed in the 0.8.2 release of Cassandra so I'm looking for a way to "upgrade" my Brisk setup to the new Cassandra.

The issue that's currently bothering me the most is CASSANDRA-2846: Changing replication_factor does not work.

I looked into the source archive and found out that all files needed for creating a Debian package are in there. What would be the steps of upgrading the Cassandra version? Is there an overview of all Brisk specific patches that would need to be re-applied? Or could I just change the dependency version in the build-properties?

Posted 1 year ago #
joaquin
Administrator

Hello,

I just spoke with our Brisk team and you should be fine with just replacing the jar file. If you were to be rebuilding it from source, then changing the dependency version should be all that is needed as well. At this point, the additional Brisk specific patches on top of Cassandra 0.8.1 are already apart of Cassandra core.

Thanks,
Joaquin

Posted 1 year ago #
t0mas
Member

Thanks, I'll do so in our test environment and report how it worked out. Also a quick suggestion for the forum: You should add the "email me when a reply is received" option so people know when something is answered :-)

Posted 1 year ago #
t0mas
Member

The update unfortunately causes issues, because of the changes in CASSANDRA-1777. Those changes caused a new problem where Cassandra accidentally adds a "/" character to the returned address, this is fixed in CASSANDRA-3044 (for 0.8.5).

The combination however still doesn't work, because when the node has its rpc_address set to "0.0.0.0" (required for Brisk to allow access from localhost) then it also returns "0.0.0.0" in TokenRange.endpoints which Brisk then tries to connect to during setup of the job, but connecting to "0.0.0.0" obviously doesn't work.

Posted 1 year ago #
anthony.ikeda
Member

Any progress on running Brisk with 0.8.5? I tried replacing the jars and although brisk cassandra -t works, when I try and query with hive I get:

hive> select * from Registry;
OK
Failed with exception java.io.IOException:java.io.IOException: Could not get input splits
Time taken: 0.133 seconds

Posted 1 year ago #
cathy@datastax.com
Member

Disregard the comment in the brisk cassandra.yaml file, and set rpc_address from '0.0.0.0' to the same value used for listen_address. I believe that should fix you.

I encountered other issues running terasort which required that /etc/hosts was configured on all nodes, and once that was in place I needed to specify both listen_address and rpc_address to hostname. I am not sure if this last step will be required for you but thought that I would mention it.

Posted 1 year ago #
anthony.ikeda
Member

Okay, that didn't help at all. I'm guessing there are more configuration settings required than just that - I get the same result:

# Address to bind to and tell other Cassandra nodes to connect to. You
# _must_ change this if you want multiple nodes to be able to
# communicate!
#
# Leaving it blank leaves it up to InetAddress.getLocalHost(). This
# will always do the Right Thing *if* the node is properly configured
# (hostname, name resolution, etc), and the Right Thing is to use the
# address associated with the hostname (it might not be).
#
# Setting this to 0.0.0.0 is always wrong.
listen_address: localhost

# The address to bind the Thrift RPC service to -- clients connect
# here. Unlike ListenAddress above, you *can* specify 0.0.0.0 here if
# you want Thrift to listen on all interfaces.
#
# Leaving this blank has the same effect it does for ListenAddress,
# (i.e. it will be based on the configured hostname of the node).
# Brisk nodes should leave this as 0.0.0.0
# rpc_address: 0.0.0.0
rpc_address: localhost

hive> create external table Registry(registry_id string, external_id string, name string)
> stored by 'org.apache.hadoop.hive.cassandra.CassandraStorageHandler'
> with serdeproperties ("cassandra.columns.mapping" = "c-externalId,c-name")
> tblproperties ("cassandra.range.size" = "100",
> "cassandra.ks.name" = "RegistryFoundation","cassandra.slice.predicate.size" = "100" );
OK
Time taken: 0.092 seconds
hive> select * from Registry;
OK
Failed with exception java.io.IOException:java.io.IOException: Could not get input splits
Time taken: 0.045 seconds
hive>

Posted 1 year ago #
cathy@datastax.com
Member

Can you try to use the IP address of the machines instead of localhost?
I am trying to chase down some host resolution bugs when I run brisk with 0.8.5 and 1.0 but it seems if that each cluster I work with needs a slightly different specification: my mac works with localhost, my debian cluster requires hostname (not localhost or IP address). I'll run some tests today and see if it helps.

Posted 1 year ago #
cathy@datastax.com
Member

We have confirmed that Brisk is currently only compatible with Cassandra 0.8.3 and soon to be released 0.8.6.

Posted 1 year ago #
gibrown
Member

I am running into similar problems due to trying to use CounterColumns which are only supported through Hadoop with cassandra 0.8.3+. I have tried replacing the jars:

resources/cassandra/lib/apache-cassandra-0.8.1-brisk1-beta2.jar
resources/cassandra/lib/apache-cassandra-thrift-0.8.1-brisk1-beta2.jar

Tested by replacing with both 0.8.6 and 0.8.3. In both cases I did get the Pig demo (003) to run, but was unable to access my CounterColumns via Pig.

Most relevant error seems to be: Failed to read data from "cassandra://SiteStats/views"

Any ideas on what else I might need to upgrade. The next thing I am thinking of trying is rebuilding Brisk from source rather than just copying in the jars. Any tips?

Thanks for any help you can provide.
-Greg

More background info:

>describe keyspace SiteStats
Keyspace: SiteStats:
Replication Strategy: org.apache.cassandra.locator.SimpleStrategy
Durable Writes: true
Options: [replication_factor:1]
ColumnFamily: views
Key Validation Class: org.apache.cassandra.db.marshal.BytesType
Default column value validator: org.apache.cassandra.db.marshal.CounterColumnType
Columns sorted by: org.apache.cassandra.db.marshal.BytesType
Row cache size / save period in seconds: 0.0/0
Key cache size / save period in seconds: 200000.0/14400
Memtable thresholds: 0.0703125/1440/15 (millions of ops/minutes/MB)
GC grace seconds: 864000
Compaction min/max thresholds: 4/32
Read repair chance: 1.0
Replicate on write: false
Built indexes: []

Pig script:
views = LOAD 'cassandra://SiteStats/views' USING CassandraStorage() AS (name, columns: bag {T: tuple(cname, cnt)});

dump views;

Posted 1 year ago #

RSS feed for this topic

Reply

You must log in to post.