0

we have the following disks from lsblk , all disks are not lvm

sdc                8:32   0    80G  0 disk /var/hadoop1
sdd                8:48   0    80G  0 disk /var/hadoop2
sde                8:64   0    80G  0 disk
sdf                8:80   0    80G  0 disk

sdc and sdd disks are full ( 100% used )

the status is that sdc and sdd disks are full and we can use them

but we have a new disks sde and sdf , each disk with size of 20G

so

is it possible to add the sde disk to sdc in order to give another 20G for sdc ?

  • I think the disks are vmdk disk ( appears in vmware vcenter ) – yael Nov 25 '19 at 12:26
  • we cant - we not get approval for grow the disks , so we are stuck , but maybe we can used the other disks ? but how? – yael Nov 25 '19 at 12:35
  • yes , I know this is silly , but we cant for now , and cluster is with downtime – yael Nov 25 '19 at 12:40
  • lets say we get approval and they grow the disks to 120G then what are the steps from my side ( we have redhat 7.x ) – yael Nov 25 '19 at 12:44
  • 2
3

Instead of adding disks on the level of operation system you can do this directly in hadoop. You can add them to the dfs.datanode.data.dir property. The format is

<property>
  <name>dfs.datanode.data.dir</name>
    <value>file:///disk/c0t2,/disk/c0t3,/dev/sde,/dev/sdf</value>
</property>

I am not 100% sure hadoop can handle RAW disks. In such case you can create on each new disk one big partition, format it, mount it /var/hadoop3, /var/hadoop4 and use format:

<property>
  <name>dfs.datanode.data.dir</name>
    <value>file:///disk/c0t2,/disk/c0t3,/var/hadoop3,/var/hadoop4</value>
</property>

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.