Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Mon, 10 Apr 2017 14:28:41 +0000 (UTC)
From: "Kevin Rivait (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.13059159.1490460272000.248118.1491834521833@Atlassian.JIRA>
In-Reply-To: <JIRA.13059159.1490460272000@Atlassian.JIRA>
References: <JIRA.13059159.1490460272000@Atlassian.JIRA> <JIRA.13059159.1490460272048@jira-lw-us.apache.org>
Subject: [jira] [Commented] (CASSANDRA-13378) DS Cassandra3.0  Adding a
 datacenter to a cluster procedure not working for us
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Mon, 10 Apr 2017 14:28:46 -0000


    [ https://issues.apache.org/jira/browse/CASSANDRA-13378?page=3Dcom.atla=
ssian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=
=3D15962952#comment-15962952 ]=20

Kevin Rivait commented on CASSANDRA-13378:
------------------------------------------

Datastax has updated their procedure ( 1 small tweak).
https://docs.datastax.com/en/cassandra/3.0/cassandra/operations/opsAddDCToC=
luster.html
We updated our scripts to match and in that process, realized we had a prob=
lem with our DC name on the alter keyspace stmts. There was an extra space =
in front of the DC name.
Looks like the original procedure would have also worked.
We have run through the corrected procedure on EPG and have successfully st=
retched an existing cluster.
I will reschedule the activity for our current DEV physical  cluster. =20

> DS Cassandra3.0  Adding a datacenter to a cluster procedure not working f=
or us
> -------------------------------------------------------------------------=
-----
>
>                 Key: CASSANDRA-13378
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1337=
8
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Configuration
>         Environment: 5 node cluster ,=20
> =E2=80=A2=09Server Model: Dell R730xd
> =E2=80=A2=09Processor: 2 * Intel Xeon E5-2630 =E2=80=93 Total 16 cores=20
> =E2=80=A2=09Memory: 192 GB =20
> =E2=80=A2=09OS Drive: 2 * 120 GB=20
> =E2=80=A2=09Direct attached storage =E2=80=93 8 TB (4 * 2TB SSD) + 1 * 80=
0GB SSD
> =E2=80=A2=0910Gb Dual Port + 1Gb Dual Port Network Daughter Card=20
> OS - OEL 2.6.32-642.15.1.el6.x86_64
> java version "1.8.0_91"
> Java(TM) SE Runtime Environment (build 1.8.0_91-b14)
> Java HotSpot(TM) 64-Bit Server VM (build 25.91-b14, mixed mode)
> 6 nodes in other DC are identical
>            Reporter: Kevin Rivait
>            Priority: Blocker
>             Fix For: 3.0.9
>
>         Attachments: cassandra-rackdc.properties_from_dc1_node, cassandra=
-rackdc.properties_from_dc2_node, cassandra.yaml, DS-add-a-dc-C3.0.rtf
>
>
> i have replicated the issue on my personal cluster using VMs.
> we have many keyspaces and users developing on the Dev Cluster we are try=
ing to stretch.
> With my current VMs (10 total) i can create a 2 DC cluster  dc1 and dc2.
> i rebuild all nodes - clear data dirs and restart dc1.
> i also clear data dirs on dc2 nodes,  i donot restart yet,
> now i have single dc1 5 nodes cluster.
> snitch - GossipingPropertyFileSnitch
> i create keyspaces on DC1
> after i alter keyspaces with replication dc1 3   dc2 0    i can no longer=
 query tables - not enough replicas available for query at consistency ONE.
> same error with CQL using consistency local_one
> continue with procedure, startup dc2 nodes, =20
> alter replication on keyspaces to  dc1 3  dc2 2
> from dc2 nodes , nodetool rebuild -- dc1  fails
> i am attaching detailed steps with errors  and cassandra.yaml


--
This message was sent by Atlassian JIRA
(v6.3.15#6346)