Reply
Highlighted
New Contributor
Posts: 1
Registered: ‎07-02-2015
Accepted Solution

Scaling a Cloudera Manager File System

I did a test installation of Cloudera Manager 5.4 on a cluster of 4 machines (VM's). The services and everything installed fine. However, I had configured my VM's with only 20 GB of Disk space to begin with (Dynamically Allocated). As soon as the cluster started, the health checks ran and returned critical issues with Log Directory Free Space for all the services. I did a df -h and returned this.

 

 

[hadoop@node1 ~]$ df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_node1-lv_root
                       15G   13G  1.6G  90% /
tmpfs                 5.0G  232K  5.0G   1% /dev/shm
/dev/sda1             477M   59M  393M  13% /boot
cm_processes          5.0G  7.3M  5.0G   1% /var/run/cloudera-scm-agent/process

 

I see that it has mounted a new file system (/dev/mapper/vg_node1-lv_root) on the root directory. 

 

Now, I have attached another disk (/dev/sdb1) and mounted it to the OS. How do I configure it to be used by Cloudera Manager? 

 

df -h after attaching the disk returns:

Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_node1-lv_root
                       15G   13G  1.6G  90% /
tmpfs                 5.0G  232K  5.0G   1% /dev/shm
/dev/sda1             477M   59M  393M  13% /boot
cm_processes          5.0G  7.3M  5.0G   1% /var/run/cloudera-scm-agent/process
/dev/sdb1              50G  180M   47G   1% /disk1

Thank You!

Cloudera Employee
Posts: 578
Registered: ‎01-20-2014

Re: Scaling a Cloudera Manager File System

The filesystem /var/run/cloudera-scm-agent/process is RAM-only. If you want
Cloudera Manager and CDH to use /disk1 now, you should edit the service
properties and switch the properties that write to /opt and /var to use
/disk1. There's no clean way to do this unfortunately, has to be a
per-service basis.

Regards,
Gautam Gopalakrishnan