Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Received-SPF: pass (athena.apache.org: domain of goudarzi@gmail.com designates
 209.85.212.44 as permitted sender)
MIME-Version: 1.0
Date: Thu, 23 Aug 2012 17:20:15 -0700
Message-ID: 
 <CAN1VBD-01aD7wT2w1eyY2KpHwcj+CoMjvE4=J5ZaSwYbmw_AyA@mail.gmail.com>
Subject: Steps to Manually Resolve Stuck Schema!! CASSANDRA-4561
From: Arya Goudarzi <goudarzi@gmail.com>
To: user@cassandra.apache.org
Cc: ilya@cardspring.com
Content-Type: multipart/alternative; boundary=20cf307abe7daee12704c7f7f08c

--20cf307abe7daee12704c7f7f08c
Content-Type: text/plain; charset=UTF-8

Just had a good conversation with rcoli in chat. Wanted to clarify the
steps for resolving this issue and see if there are any pitfalls I am
missing.

Issue: I upgraded from 1.1.2 to 1.1.3 a while ago and today I realized I
cannot make any schema changes since the fix in
https://issues.apache.org/jira/browse/CASSANDRA-4432.

Solution: Somehow I have to make Cassandra system's column family to forget
about those old schemas with nanosecond timestamps. I have to do this
either live or with a brief downtime. Please advice of any pitfalls or
incorrectness in my steps. I am planning to automate them so please advice.

Within a short downtime, I have to do this:

1. Take all nodes out of service;
2. Run nodetool flush on each;
3. Stop cassandra on each node;
4. Remove /var/lib/cassandra/data/system
5. Remove /var/lib/cassandra/saved_caches/system-*
6. Start all nodes;
7. cassandra-cli < schema_definition_file on one node only. (includes
create keyspace and create column familiy entries)
8. put the nodes back in service.
9. Done.

Please advice if I have got the steps correctly or if I am missing
something.

Thanks in advance for you help.

Cheers,
-Arya

--20cf307abe7daee12704c7f7f08c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Just had a good=C2=A0conversation=C2=A0with rcoli in chat. Wanted to clarif=
y the steps for resolving this issue and see if there are any pitfalls I am=
 missing.<div><br></div><div>Issue: I upgraded from 1.1.2 to 1.1.3 a while =
ago and today I realized I cannot make any schema changes since the fix in=
=C2=A0<a href=3D"https://issues.apache.org/jira/browse/CASSANDRA-4432">http=
s://issues.apache.org/jira/browse/CASSANDRA-4432</a>.=C2=A0</div>
<div><br></div><div>Solution: Somehow I have to make Cassandra system&#39;s=
 column family to forget about those old schemas with nanosecond timestamps=
. I have to do this either live or with a brief downtime. Please advice of =
any pitfalls or incorrectness in my steps. I am planning to automate them s=
o please advice.</div>
<div><br></div><div>Within a short downtime, I have to do this:</div><div><=
br></div><div>1. Take all nodes out of service;</div><div>2. Run nodetool f=
lush on each;</div><div>3. Stop cassandra on each node;</div><div>4. Remove=
 /var/lib/cassandra/data/system</div>
<div>5. Remove /var/lib/cassandra/saved_caches/system-*</div><div>6. Start =
all nodes;</div><div>7. cassandra-cli &lt; schema_definition_file on one no=
de only. (includes create keyspace and create column familiy=C2=A0entries)<=
/div>
<div>8. put the nodes back in service.</div><div>9. Done.</div><div><br></d=
iv><div>Please advice if I have got the steps correctly or if I am missing =
something.=C2=A0</div><div><br></div><div>Thanks in advance for you help.</=
div>
<div><br></div><div>Cheers,</div><div>-Arya</div><div>=C2=A0</div>

--20cf307abe7daee12704c7f7f08c--