Our Community is getting an upgrade! To get everything ready for the relaunch, we’ll be placing the site in read-only mode starting September 21st.
We really appreciate your understanding while we get things set up behind the scenes. Catch up on all the exciting details about the move here.
Need help or have questions? Drop us a line at [email protected]
Created 12-21-2016 02:19 AM
want to add a few nodes to a running cluster. does it require any down time and will the running jobs get affected?
Created 12-21-2016 10:03 PM
Basically, adding nodes (hosts) requires no downtime and will not affect running jobs at all.
Btw, you may try to do this by Ambari Add Hosts function. Scenario:
1. Go to Ambari Hosts page.
2. Click on Actions button located on the top left of this page.
3. Select "Add New Hosts " option.
You will see a wizard which will guide you through the whole process, you can add hosts with any slaves (DataNode, NodeManager, clients .etc) installed on them.
Hope this helps 🙂
Created 12-21-2016 03:17 AM
@ARUN :
No downtime is required for simple adding nodes for just DN, NM and other components. Even for adding client node(Edge node) doesn't require any downtime.
Created 12-21-2016 03:42 AM
Created 12-21-2016 10:03 PM
Basically, adding nodes (hosts) requires no downtime and will not affect running jobs at all.
Btw, you may try to do this by Ambari Add Hosts function. Scenario:
1. Go to Ambari Hosts page.
2. Click on Actions button located on the top left of this page.
3. Select "Add New Hosts " option.
You will see a wizard which will guide you through the whole process, you can add hosts with any slaves (DataNode, NodeManager, clients .etc) installed on them.
Hope this helps 🙂
Created 12-22-2016 03:00 AM
@ARUN
As @Xi Wang responded and suggested. I assess her response as appropriate. Please see also doc link (assuming you use Ambari 2.1.1.0, change to the proper link if you use an earlier version):
Also, keep in mind that only new data will use the new data nodes, unless you execute rebalance hdfs command, then you have a distribution of all the data across all data nodes. Default threshold is 10, but you could change it to your desired threshold. You may want to execute it during off-peak hours.