incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tim Dunphy <bluethu...@gmail.com>
Subject Re: decommission dead node without token
Date Sun, 01 Jun 2014 04:17:35 GMT
sorry for the barrage of questions.... but I'm having trouble with the
nodetool remove phase.

[root@cassandra01 ~]# nodetool removetoken -h 10.10.1.68
-9208584805646615844

Gives me a unix usage message. As in:

[root@cassandra01 ~]# nodetool removetoken -h 107.170.178.68
-9208584805646615844 | head -5
Unrecognized option: -9208584805646615844
usage: java org.apache.cassandra.tools.NodeCmd --host <arg> <command>

 -a,--include-all-sstables   includes sstables that are already on the
                             most recent version during upgradesstables
 -c,--compact                print histograms in a more compact format

I think the problem is that the bash environment is getting confused by the
minus sign in front of the token. But that is the token and the minus sign
is integral to it. I've even tried quoting it with both single quotes and
double quotes to no avail. I still get the same usage message.

Can someone please help with the right syntax here?

Thanks
Tim


On Sun, Jun 1, 2014 at 12:01 AM, Tim Dunphy <bluethundr@gmail.com> wrote:

> Just wondering.. might I be able to retried the 'lost' intital_token of
> the node I want to remove by going:
>
> [root@beta-new:~] #nodetool ring | grep 10.10.1.102 | head -1
> 107.170.175.102  rack1       Down   Normal  123.41 KB       32.19%
>      -9208584805646615844
>
> Then
>
> 1) place that token in the cassandra.yaml of a new node with a new IP as
>  -9208584805646615844 -1
> 2)  set auto_bootstrap in the cassandra.yaml
> 3) once the node boots run 'nodetool removetoken -9208584805646615844'
> 4) run nodetool cleanup
>
> Would this be the way to get out of this bind? Just want to be sure I
> understand the process.
>
> Thanks
> Tim
>
>
> On Sat, May 31, 2014 at 10:38 PM, Tim Dunphy <bluethundr@gmail.com> wrote:
>
>> hey guys,
>>
>>  I have two dead nodes in my ring.
>>
>> Status=Up/Down
>>
>> |/ State=Normal/Leaving/Joining/Moving
>>
>> --  Address          Load       Tokens  Owns   Host ID
>>             Rack
>>
>> DN  10.10.1.102  123.41 KB  256     32.2%
>> ddcafc75-24ed-4c6a-99bb-afe3dd551a9c  rack1
>>
>> UN  10.10.1.94   170.81 KB  256     35.4%
>> fd2f76ae-8dcf-4e93-a37f-bf1e9088696e  rack1
>>
>> DN  10..10.10.64   ?          256     32.5%
>> f2a48fc7-a362-43f5-9061-4bb3739fdeaf  rack1
>>
>>
>> Unfortunately I am early in my learning curve and failed to preserve the
>> initial_token for each of the nodes that were taken out of service. As
>> things stand now I cannot do a nodetool repair on the main keyspace because
>> of the dead nodes.
>>
>> Is there any way to decommission the dead nodes if you no longer have
>> their initial_token values?
>>
>>
>> Thanks
>>
>> Tim
>>
>> --
>> GPG me!!
>>
>> gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B
>>
>>
>
>
> --
> GPG me!!
>
> gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B
>
>


-- 
GPG me!!

gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B

Mime
View raw message