You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-user@hadoop.apache.org by Charley Newtonne <cn...@gmail.com> on 2014/06/26 21:30:15 UTC
Configuring Hadoop Client: Where is fail-over configured.
I have hadoop 2.4 installed in HA mode using QJM. I have verified the
cluster failover works as expected. The java clients are configured to
connect to the active NN by specifying the hdf://nn1.company.com:8020. If
this nn1 is down, how's the client know the location of the standby nn.
Where is the client spillover configured?
I have seen some references that this is configured in client's
core-site.xml, but this file only specifies the defaultFS (which has the
value of the logical cluster name) and ZK quorum nodes. None of these
appear to be related to client side failover.
Thanks in advance
Re: Configuring Hadoop Client: Where is fail-over configured.
Posted by Juan Carlos <jc...@redoop.org>.
Hi Charley,
in hdfs-site.xml you can find the property dfs.ha.namenodes, setting this
property every client will know which NN are elegibles to be active,
nothing else are required in client.
Regards.
2014-06-26 21:30 GMT+02:00 Charley Newtonne <cn...@gmail.com>:
> I have hadoop 2.4 installed in HA mode using QJM. I have verified the
> cluster failover works as expected. The java clients are configured to
> connect to the active NN by specifying the hdf://nn1.company.com:8020. If
> this nn1 is down, how's the client know the location of the standby nn.
> Where is the client spillover configured?
>
> I have seen some references that this is configured in client's
> core-site.xml, but this file only specifies the defaultFS (which has the
> value of the logical cluster name) and ZK quorum nodes. None of these
> appear to be related to client side failover.
>
> Thanks in advance
>
Re: Configuring Hadoop Client: Where is fail-over configured.
Posted by Juan Carlos <jc...@redoop.org>.
Hi Charley,
in hdfs-site.xml you can find the property dfs.ha.namenodes, setting this
property every client will know which NN are elegibles to be active,
nothing else are required in client.
Regards.
2014-06-26 21:30 GMT+02:00 Charley Newtonne <cn...@gmail.com>:
> I have hadoop 2.4 installed in HA mode using QJM. I have verified the
> cluster failover works as expected. The java clients are configured to
> connect to the active NN by specifying the hdf://nn1.company.com:8020. If
> this nn1 is down, how's the client know the location of the standby nn.
> Where is the client spillover configured?
>
> I have seen some references that this is configured in client's
> core-site.xml, but this file only specifies the defaultFS (which has the
> value of the logical cluster name) and ZK quorum nodes. None of these
> appear to be related to client side failover.
>
> Thanks in advance
>
Re: Configuring Hadoop Client: Where is fail-over configured.
Posted by Juan Carlos <jc...@redoop.org>.
Hi Charley,
in hdfs-site.xml you can find the property dfs.ha.namenodes, setting this
property every client will know which NN are elegibles to be active,
nothing else are required in client.
Regards.
2014-06-26 21:30 GMT+02:00 Charley Newtonne <cn...@gmail.com>:
> I have hadoop 2.4 installed in HA mode using QJM. I have verified the
> cluster failover works as expected. The java clients are configured to
> connect to the active NN by specifying the hdf://nn1.company.com:8020. If
> this nn1 is down, how's the client know the location of the standby nn.
> Where is the client spillover configured?
>
> I have seen some references that this is configured in client's
> core-site.xml, but this file only specifies the defaultFS (which has the
> value of the logical cluster name) and ZK quorum nodes. None of these
> appear to be related to client side failover.
>
> Thanks in advance
>
Re: Configuring Hadoop Client: Where is fail-over configured.
Posted by Juan Carlos <jc...@redoop.org>.
Hi Charley,
in hdfs-site.xml you can find the property dfs.ha.namenodes, setting this
property every client will know which NN are elegibles to be active,
nothing else are required in client.
Regards.
2014-06-26 21:30 GMT+02:00 Charley Newtonne <cn...@gmail.com>:
> I have hadoop 2.4 installed in HA mode using QJM. I have verified the
> cluster failover works as expected. The java clients are configured to
> connect to the active NN by specifying the hdf://nn1.company.com:8020. If
> this nn1 is down, how's the client know the location of the standby nn.
> Where is the client spillover configured?
>
> I have seen some references that this is configured in client's
> core-site.xml, but this file only specifies the defaultFS (which has the
> value of the logical cluster name) and ZK quorum nodes. None of these
> appear to be related to client side failover.
>
> Thanks in advance
>