cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Arya Goudarzi (JIRA)" <j...@apache.org>
Subject [jira] Updated: (CASSANDRA-1341) Duplicate nodes showing on the ring when starting 3 nodes on empty cluster
Date Thu, 29 Jul 2010 23:23:15 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-1341?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Arya Goudarzi updated CASSANDRA-1341:
-------------------------------------

    Description: 
When bringing a cluster of 3 nodes up, if user does not boot up the nodes serially and wait
for bootstrap to finish on each before restarting the next one, the cluster may end up having
duplicate nodes showing on the ring.

Steps to Reproduce:
1. Configure a cluster of 3 from clean slate of trunc;
2. Start the first node;
3. Start the second node; 
4. As soon as the second node starts ans says Finished Hinted Handoff to node 1, start node
3;
5. Load the schema from YAML using JMX or use Thrift to define some schema so that we could
use nodetool (CASSANDRA-1286)

I ended up with a strange ring configuration:

[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.132 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 
[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.133 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 
[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.134 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 


In my case

10.50.26.132 = node 1
10.50.26.133 = node 2
10.50.26.134 = node 3 

As you see node2 is repeated Twice. Node 3 is running however does not show up in the ring!

Now, if in step 4 if I wait for bootstrap of node 2 to kick in and finish, I won't have this
problem.

  was:
When bringing a cluster of 3 nodes up, if user does not boot up the nodes serially and wait
for bootstrap to finish on each before restarting the next one, the cluster may end of having
duplication nodes showing on the ring.

Steps to Reproduce:
1. Configure a cluster of 3 from clean slate of trunc;
2. Start the first node;
3. Start the second node; 
4. As soon as the second node starts ans says Finished Hinted Handoff to node 1, start node
3;
5. Load the schema from YAML using JMX or use Thrift to define some schema so that we could
use nodetool (CASSANDRA-1286)

I ended up with a strange ring configuration:

[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.132 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 
[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.133 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 
[agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.134 ring
Address         Status State   Load            Token                                     
 
                                       162149197822681929203296559575110871861    
10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565    
 
10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997    
 
10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861   
 


In my case

10.50.26.132 = node 1
10.50.26.133 = node 2
10.50.26.134 = node 3 

As you see node2 is repeated Twice. Node 3 is running however does not show up in the ring!

Now, if in step 4 if I wait for bootstrap of node 2 to kick in and finish, I won't have this
problem.


> Duplicate nodes showing on the ring when starting 3 nodes on empty cluster
> --------------------------------------------------------------------------
>
>                 Key: CASSANDRA-1341
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1341
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.7 beta 1
>         Environment: CentOS 5.2
> Trunk
>            Reporter: Arya Goudarzi
>
> When bringing a cluster of 3 nodes up, if user does not boot up the nodes serially and
wait for bootstrap to finish on each before restarting the next one, the cluster may end up
having duplicate nodes showing on the ring.
> Steps to Reproduce:
> 1. Configure a cluster of 3 from clean slate of trunc;
> 2. Start the first node;
> 3. Start the second node; 
> 4. As soon as the second node starts ans says Finished Hinted Handoff to node 1, start
node 3;
> 5. Load the schema from YAML using JMX or use Thrift to define some schema so that we
could use nodetool (CASSANDRA-1286)
> I ended up with a strange ring configuration:
> [agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.132 ring
> Address         Status State   Load            Token                                
      
>                                        162149197822681929203296559575110871861    
> 10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565
     
> 10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997
     
> 10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861
    
> [agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.133 ring
> Address         Status State   Load            Token                                
      
>                                        162149197822681929203296559575110871861    
> 10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565
     
> 10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997
     
> 10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861
    
> [agoudarzi@cas-test1 ~]$ nodetool --host=10.50.26.134 ring
> Address         Status State   Load            Token                                
      
>                                        162149197822681929203296559575110871861    
> 10.50.26.133    Up     Normal  41.61 KB        34543310227330005404531081788197792565
     
> 10.50.26.132    Up     Normal  41.61 KB        77078606092447313337452907717168818997
     
> 10.50.26.132    Up     Normal  41.61 KB        162149197822681929203296559575110871861
    
> In my case
> 10.50.26.132 = node 1
> 10.50.26.133 = node 2
> 10.50.26.134 = node 3 
> As you see node2 is repeated Twice. Node 3 is running however does not show up in the
ring!
> Now, if in step 4 if I wait for bootstrap of node 2 to kick in and finish, I won't have
this problem.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message