[LU-8042] mkfs.lustre should set ashift=12 recordsize=1024k compression=lz4" by default for new ZFS OSTs Created: 19/Apr/16  Updated: 30/Jun/16  Resolved: 30/Jun/16

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Improvement Priority: Minor
Reporter: Andreas Dilger Assignee: WC Triage
Resolution: Duplicate Votes: 0
Labels: None

Issue Links:
Related
is related to LU-8342 ZFS dnodesize and recordsize should b... Resolved
Rank (Obsolete): 9223372036854775807

 Description   

It would be useful for mkfs.lustre to set ashift=12, recordsize=1024k, and compression=lz4 for new ZFS OSTs. It doesn't make sense to do this for MDTs.



 Comments   
Comment by Jinshan Xiong (Inactive) [ 19/Apr/16 ]

Are you proposing to turn on data compression on OSTs by default?

Comment by Andreas Dilger [ 19/Apr/16 ]

Definitely ashift and recordsize should be set. Compression maybe should be in a separate patch, but both Keith and Zhiqi reported that compression helped performance.

Comment by Jinshan Xiong (Inactive) [ 19/Apr/16 ]

I have a concern that the data generated by test program tend to be compressible. But I agree that shift and record size should be made default for sure.

Comment by Gerrit Updater [ 29/Apr/16 ]

Andreas Dilger (andreas.dilger@intel.com) uploaded a new patch: http://review.whamcloud.com/19892
Subject: LU-8042 osd-zfs: set ashift=12 recordsize=1024k by default
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: e6057e5ce5e11dd8be6317c96056498c9b4f09e9

Comment by Andreas Dilger [ 30/Jun/16 ]

Close this as a duplicate of LU-8342.

Generated at Sat Feb 10 02:14:07 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.