[LU-12297] Add an option to disable mount.lustre automatical tune Created: 14/May/19  Updated: 26/Sep/20  Resolved: 14/Jul/20

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

Type: Improvement Priority: Minor
Reporter: Wang Shilong (Inactive) Assignee: Wang Shilong (Inactive)
Resolution: Duplicate Votes: 0
Labels: patch

Issue Links:
Duplicate
duplicates LU-12029 do not try to muck with max_sectors_k... Open
Related
is related to LU-9132 Tuning max_sectors_kb on mount Resolved
is related to LU-12387 l_tunedisk does not properly handle m... Resolved
Rank (Obsolete): 9223372036854775807

 Description   

Mount.lustre will try to tune max_sectors_kb, scheduler automatically for OSTs in default.
This could cause some unexpected behavior if there are external tools which try to do same
things using udev rules etc, as lustre could use raw block device or be on the top of logical (N logical) devices and as a result this tunings on fly will be a result of excellent butthearts
especially when tuning comes from storage.

In order to solve this problem, it makes sense to add a mount option like no_auto_tune
etc to disable automatically tune by mount.lustre util.

 Comments   
Comment by Gerrit Updater [ 14/May/19 ]

Wang Shilong (wshilong@ddn.com) uploaded a new patch: https://review.whamcloud.com/34859
Subject: LU-12297 utils: add an option to disable mount.lustre auto tune
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 73767e5dfa772f0a3d99f3b0aa35df37b5664c8d

Comment by Gu Zheng (Inactive) [ 15/May/19 ]

I think we'd better to control the tune from mount_lustre rather than only ldsikfs side.
The major goal of this option is let lustre free from backend storage tuning, and give a chance to manage-utils or admin (out of native lustre, e.g Exascalser from DDN) who are more professional and proficient in tuning field to take over the job.
Actually tune to zfs is more complex than ldiskfs, and this option is also very useful to osd-zfs sites.

Comment by Yang Sheng [ 15/May/19 ]

Hi, Shilong,

I think we have a similar way to do that. Please see LU-9132. Also you can refer to LU-9551 for further detail.

Thanks,
Yangsheng

Comment by Wang Shilong (Inactive) [ 15/May/19 ]

Hi Sheng,

Yup, that ticket solved some of problem that we hit currently, but isn't enough.
Further what want is to add a mount option disable any Lustre auto tune for ZFS/ldiskfs.

For now, we want to disable max_sector_kb and scheduler tune for ldiskfs, but newly
tuned approach might be added in the future.

Comment by Andreas Dilger [ 14/Jul/20 ]

Fix this via LU-12029.

Generated at Sat Feb 10 02:51:20 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.