| To: | weber@xxxxxxxxxx, Xfs <xfs@xxxxxxxxxxx> |
|---|---|
| Subject: | Re: gentoo linux, kernel 3.10.31 mount options bug? |
| From: | Stan Hoeppner <stan@xxxxxxxxxxxxxxxxx> |
| Date: | Sun, 23 Feb 2014 16:10:43 -0600 |
| Delivered-to: | xfs@xxxxxxxxxxx |
| In-reply-to: | <9963a76fbd006355c9ffd79f341c9971@xxxxxxxxxx> |
| References: | <9963a76fbd006355c9ffd79f341c9971@xxxxxxxxxx> |
| Reply-to: | stan@xxxxxxxxxxxxxxxxx |
| User-agent: | Mozilla/5.0 (Windows NT 5.1; rv:24.0) Gecko/20100101 Thunderbird/24.3.0 |
On 2/23/2014 3:37 AM, Marko Weber|8000 wrote:
...
> linux /bzImage-3.10.31 root=/dev/md2 elevator=cfq clocksource=hpet
^^^^^^^^^^^^
cfq tends to defeat much of the parallelism in XFS, decreasing
throughput substantially. This is documented in the XFS FAQ and has
been discussed here many times in the past. It has been recommended for
a few years now that XFS not be used with the cfq elevator. Use
deadline with md arrays on plain HBAs and noop on SSDs or any device
with [F|B]BWC, i.e. RAID HBA or SAN controller.
If you're using cfq to allow shaping of per process IO with control
groups, simply using cfq alone may slow down XFS throughput to the point
that you don't need to bother with control group optimizations.
--
Stan
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH v6] ext4: Add support FALLOC_FL_COLLAPSE_RANGE for fallocate, Dave Chinner |
|---|---|
| Next by Date: | [PATCH] [RFC] xfs: byte range buffer dirty region tracking, Dave Chinner |
| Previous by Thread: | Re: gentoo linux, kernel 3.10.31 mount options bug?, Eric Sandeen |
| Next by Thread: | Re: gentoo linux, kernel 3.10.31 mount options bug?, Marko Weber|8000 |
| Indexes: | [Date] [Thread] [Top] [All Lists] |