kudu-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dan Burkert <danburk...@apache.org>
Subject Re: Kudu Queries
Date Wed, 20 Dec 2017 19:30:09 GMT
Hi Ajay,

Have you looked at the documentation section on kudu.apache.org?  In
particular these sections may be helpful:

https://kudu.apache.org/docs/schema_design.html
https://kudu.apache.org/docs/administration.html#migrate_to_multi_master
https://kudu.apache.org/docs/administration.html#_adding_a_data_directory

Answers to specific questions inline:

On Wed, Dec 20, 2017 at 3:19 AM, ajay chaudhary <juniorajay@yahoo.com>
wrote:

we would want to know how would insert be handled when one of the node is
> 100% full for some reason.
>

As inserts are attempted, the tablets on that server will begin to fail and
be re-replicated to other tservers. This can cause a cascading scenario, so
you should avoid running with full disks.

How can we add a node to Kudu Cluster? what is the process.
>

Just stand up the tablet server process and point it to the existing master.


> Can we add a node to Kudu cluster and exisiting tables can make use of it
> automatically ? or is there any process around it?
>

Kudu won't rebalance automatically to the new server. If you create new
tables, or if tablets of existing tables failover and are re-replicated,
then the new tablet servers will be considered as destinations for the new
tablets.  The move-replica
<https://kudu.apache.org/docs/command_line_tools_reference.html#change_config-move_replica>
tool is available to manually move tablets around, but it's somewhat
difficult to use for large coordinated re-balancing, hopefully we'll grow
some better tools here soon.


> Can we add/delete more partition(hash) to exisiting kudu tables?
>

No, only range partitions can be added or removed.

Mime
View raw message