Removed reiserfs list from cc - getting a bit xfs specific here.
>
> ----- Original Message -----
> From: Steve Lord <lord@xxxxxxx>
>
> > p.s. For the filesystem mkfs and mount part of mongo.pl I would recommend
> > this:
> >
> > if ( $FILESYSTEM eq "xfs" ) {
> > system("mkfs -t xfs -f -l size=16384b $DEVICE") ;
> > system("mount -o logbufs=4,osyncisdsync $DEVICE $TESTDIR") ;
> > }
>
>
> Could these mount options can be also applied on production server ?
> (i.e. mail, web and sql db server ?)
Yes, they apply anywhere, look at this message for a bit more of
an explaination:
http://linux-xfs.sgi.com/projects/xfs/mail_archive/0105/msg00663.html
>
> Is there possibility to limit memory used for disk/file cacheing ?
Not really - this is a linux thing not an xfs thing.
>
> How mkfs.xfs calculates values for agcount and agsize ?
> (I have spent some time trying to find apropriate "-l SIZE"
> for each created fs)
If you take the latest cvs tree it has just had some code merged from Irix
which adds new mkfs options which make controlling the agsize simpler.
>
> On 15GB filesystem, did it makes big difference that it was created
> with "-l" value twice as you suggest ?
>
> Am I wrong thinking that noatime, nodiratime mount options also are
> tweaking XFS performance (as for ext2fs) ? (I dont need these info at
> all).
Also only in the cvs tree (as of yesterday) noatime was not getting passed
on to xfs correctly. nodiratime does not mean anything to xfs.
>
> Regards,
> Blizbor
Steve
|