| To: | xfs@xxxxxxxxxxx |
|---|---|
| Subject: | Re: block alignment |
| From: | Ric Wheeler <rwheeler@xxxxxxxxxx> |
| Date: | Mon, 07 Jan 2013 16:47:29 -0500 |
| In-reply-to: | <CAKSyJXdP0NSQrR+rfrCec+oCyVg-2SRmCmYf3V7kjC1-0x49bA@xxxxxxxxxxxxxx> |
| References: | <CAKSyJXdP0NSQrR+rfrCec+oCyVg-2SRmCmYf3V7kjC1-0x49bA@xxxxxxxxxxxxxx> |
| User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/17.0 Thunderbird/17.0 |
On 01/07/2013 04:33 PM, Bradley C. Kuszmaul wrote: What is the largest data block size that I'm likely to run into with xfs?I ask, since I'm doing hole punching, and I don't want to incur the I/O required to zero unaligned boundaries. So I want to keep the holes I punch sufficiently well aligned.-Bradley I think that you really need to worry about the block sizes below XFS - the storage block size specifically. Upstream (and modern distros like RHEL/SLES/etc) do work to export out alignment requirements and optimal IO sizes. You can find them in /sys/block/sda/. Martin Peterson wrote up a good overview of what we have in there: https://oss.oracle.com/~mkp/ Ric |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: block alignment, Ben Myers |
|---|---|
| Next by Date: | Re: [PATCH] xfs: serialize iclog write of xlog_cil_push, Dave Chinner |
| Previous by Thread: | Re: block alignment, Ben Myers |
| Next by Thread: | Reply to claim Funds from UN !!!, Sarah Horne |
| Indexes: | [Date] [Thread] [Top] [All Lists] |