1848206 Members
7389 Online
104022 Solutions
New Discussion

Re: LVM query

 
xcvzxvzxdv
Advisor

LVM query

I am finding that the file creation time in one of my volumes is high as com pared to other. IfI see the VXFS mount option from SAM they are 4

Choose one of the following VxFS mount configurations:

> No Guaranteed Data Integrity (Highest Performance)
> Full Integrity for Critical Metadata
> Full Integrity for All Metadata
> Full Integrity for All Metadata and User Data (Lowest Performance)

Currently it is set to 4 option, culd it be the cause that file creation time is more on this volume
4 REPLIES 4
A. Clay Stephenson
Acclaimed Contributor

Re: LVM query

First of all, UNIX has no file creation time so what you think is file creation time is something else (probably file change time (last chmod,chown,etc)) as opposed to modification time (last time data was changed)). In any event, those mount options don't affect any of the file modification times but instead affect the reliability that the filesystem metadata will be updated in an event of a system failure.
If it ain't broke, I can fix that.
xcvzxvzxdv
Advisor

Re: LVM query

let me express it. Oracle creates Archive logs during the processing, when we set the path to this volumes it takes longer time than when we set it to other volume.

I am basically finding the answer to this query
A. Clay Stephenson
Acclaimed Contributor

Re: LVM query

Okay. Now your question makes more sense and yes, your setting of "4" which corresponds to the vxfs mount option "log" does have an impact. Probably your best option for this kind of filesystem is "delaylog" which sacrifices some protection for performance. Man mount_vxfs for more details. You should also consider setting "nodatainlog" as well.
If it ain't broke, I can fix that.
Ninad_1
Honored Contributor

Re: LVM query

I am not aware of these options, but effectively what they must be doing is -
say for > Full Integrity for All Metadata
Instead of writing any changes/updates for file creation/modification metadata to be written to the kernel image or cache image whatever its called of the filesystem metadata (superblock) the information is updated to the actual disks before confirming to the process that the data has been updated - usually when the data is updated to cache/memory acknowledgement is given to the process ,and later the cache is flushed and actual data on disk is updated.
Similarly for > Full Integrity for All Metadata and User Data (Lowest Performance)
, I guess apart from the metadata - the actual data is also written to disk to guarantee integrity to highest order.

Regards,
Ninad