Oracle ZFS Storage ZS5-4
 
Oracle ZFS Storage ZS3-BA
 
Oracle ZFS Storage ZS3-2
 
Oracle ZFS Storage ZS3-4
 
Oracle ZFS Storage ZS5-2
 
Sun ZFS Storage 7420
 
Oracle ZFS Storage ZS4-4
 

PLA-Support>Sun Systems>DISK>ZFS Storage>SN-DK: ZS
 

In this Document

Symptoms

Changes

Cause

Solution

References

Created from <SR 3-13156764911>

Applies to:

Symptoms

On certain ZFS Storage Appliances we have seen RaidZ2 filesystem filling up faster than anticipated.

This was observed where the customer was using a record size of 8k blocks for a NFS file system and the pool was using 8TB drives in a RaidZ2 pool.

One example was du -h reported 17K on an 8192 byte (8K) write from NFS.

Changes

Cause

The issue is with how 8TB drives allocate space.  A sector used to be 512 bytes but on the 8TB drives a sector is now 4KB.

In RaidZ2 when the space is allocated for parity, more space is allocated with the 8TB drives than with smaller drives because of this new sector format.

Solution

This issue should not be seen if the pool was mirrored.

Tests have shown that the use of larger record sizes, 128KB and above, will solve the issue as well.

Compression may also help.

Our recommendation is to use a mirrored pool, if it is being used for databases - or - to use 128KB recordsize or greater.

NOTE:  This information overrides some documentation on how to set up some Oracle databases/configurations.

Check for relevancy - 10-May-2018

References

This solution has no attachment