| To: | Dave Chinner <david@xxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: XFS mount failuer on RAID5 |
| From: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
| Date: | Sun, 18 Oct 2009 23:55:38 -0400 |
| Cc: | Michael Monnerie <michael.monnerie@xxxxxxxxxxxxxxxxxxx>, xfs@xxxxxxxxxxx |
| In-reply-to: | <20091019013018.GQ9464@xxxxxxxxxxxxxxxx> |
| References: | <389deec70910152009i7656aaf2s9d929b8b68e4e740@xxxxxxxxxxxxxx> <200910161019.10286@xxxxxx> <4AD890D1.4010208@xxxxxxxxxxx> <200910172327.05122@xxxxxx> <20091019013018.GQ9464@xxxxxxxxxxxxxxxx> |
| User-agent: | Mutt/1.5.19 (2009-01-05) |
On Mon, Oct 19, 2009 at 12:30:18PM +1100, Dave Chinner wrote: > > IIRC, the main reason for 4k sectors on MD RAID5/6 is that changing > the IO alignment from 4k to 512 byte IOs (i.e. sub-page sized) > causes MD to flush and invalidate the stripe cache. Hence every > time XFS writes a super block, AGF, AGFL or AGI, things go much > slower because of this flush/invalidate. By setting the sector size > to 4k, the SB/AGF/AGFL/AGI are all 4k in size and hence IO alignment This should not happen anymore with 2.6 series kernels. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: 2.6.31 xfs_fs_destroy_inode: cannot reclaim, Christoph Hellwig |
|---|---|
| Next by Date: | [PATCH] xfs: I/O completion handlers must use NOFS allocations, Christoph Hellwig |
| Previous by Thread: | Re: XFS mount failuer on RAID5, Dave Chinner |
| Next by Thread: | Re: XFS mount failuer on RAID5, Justin Piszcz |
| Indexes: | [Date] [Thread] [Top] [All Lists] |