Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
Date: Mon, 25 Aug 2014 01:11:05 -0700 (PDT)
From: tsi <thorsten.sinn@t-systems.com>
To: cassandra-user@incubator.apache.org
Message-ID: <1408954265263-7596454.post@n2.nabble.com>
In-Reply-To: <1407935222509-7596278.post@n2.nabble.com>
References: <1407843186116-7596245.post@n2.nabble.com>
 <CAEDUwd2ESEJUVcsZK8=R_aXcO=Tnv7yh7Za3o0cTb+tpbJyZgw@mail.gmail.com>
 <1407935222509-7596278.post@n2.nabble.com>
Subject: Re: Replacing a dead node in Cassandra 2.0.8
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

I read some additional threads concerning outdated snapshots and deleted da=
ta
so if the VM snapshot is older than  gc_grace_seconds I suppose that data
deleted meanwhile for which the tombstones have been removed would come bac=
k
to life again if I just restore the VM from a snapshot being older than
gc_grace_seconds.
So after restoring the VM snapshot I should make sure Cassandra doesn=E2=80=
=99t
start up automatically with the outdated snapshot data.
Then I could=20

1)either startup the node with the replace_address option (do I still need
to clear data directories first to avoid that deleted data comes back to
life again?)

2) restore a Cassandra snapshot being not older than gc_grace_seconds by
physically replacing SSTables in the data directory (did I get it right tha=
t
the node must be up for running StorageService or sstableloader?)  and run
nodetool repair afterwards.

Is this correct?=20


tsi wrote
> OK, now supposing Cassandra is run in a VM that crashes and I restore it
> from a snapshot done some time ago. Data is stored redundantly
> (replication factor 3) and I'm using consistency level QUORUM for reads
> and writes. That means no data should be lost as the latest data will at
> least be stored on another node. Now what do I have to do to sync the
> "dead node" again after restoring the VM from the snapshot? Will a
> nodetool repair command be sufficient?


--
View this message in context: http://cassandra-user-incubator-apache-org.30=
65146.n2.nabble.com/Replacing-a-dead-node-in-Cassandra-2-0-8-tp7596245p7596=
454.html
Sent from the cassandra-user@incubator.apache.org mailing list archive at N=
abble.com.