cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jeff Jirsa (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-13637) data updated to old value after flushing largest CFS
Date Sun, 27 Aug 2017 19:10:02 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-13637?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16143189#comment-16143189
] 

Jeff Jirsa commented on CASSANDRA-13637:
----------------------------------------

Do you know what consistency level you're using on the inserts and selects?


> data updated to old value after flushing largest CFS
> ----------------------------------------------------
>
>                 Key: CASSANDRA-13637
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-13637
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>         Environment: RHEL 6.6 64 bits
> Cassandra 2.2.6 with replication factor of 2
> 2 servers in the Cassandra ring. 
> One server is down for maintenance for more than 24 hours
>            Reporter: Hervé Toulan
>             Fix For: 2.2.6
>
>         Attachments: local.zip
>
>
> VitualMachine 1 : application server  + Cassandra 1
> VirtualMachine 2: application server  + Cassandra 2
> Cassandra1 and Cassandra2 are the only members of the Cassandra ring.
> I expect Cassandra to always be synchronized as soon as VM1 and Vm2 can communicate.
> Applicaiton server (AS1) of VM1  writes a value to Cassandra (a date).
> VM1 is shot down.
> VM2 reads the value written by AS1  as expected.
> Cassandra of VM2 perform a flush operation.
> VM2 reads the value older than the exp
> ected one, the last value written by AS1.
> I've lost the last inserted data, and I don't understand why.
> In attached log bcs.log the read in DB:
> 2.4.0.1-SNAPSHOT - LCL - 26 Jun 2017 07:15:47,959 INFO  [RMI TCP Connection(111)-172.25.153.163]
- Cassandra supervisor : check OK : last generation of CDR is : {color:red}Sun Jun 25 00:10:53
CEST 2017{color}
> 2.4.0.1-SNAPSHOT - LCL - 26 Jun 2017 07:16:47,963 INFO  [RMI TCP Connection(113)-172.25.153.163]
- Cassandra supervisor : check OK : last generation of CDR is : {color:red}Thu Jun 22 07:16:10
CEST 2017{color}
> Nothing has been inserted in database between the 2 logs but in debug.log from Cassandra
I can see a flush operation the 2017-06-24 at  07:16:37,056.
> What did we miss ?
> How data  can be overwritten  ?
> Regards,
> Hervé



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org


Mime
View raw message