Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.os.linux.hardware > #3178 > unrolled thread

Momentus XT sshd slowness?

Started byPeter Chant <pete@petezilla.co.uk>
First post2016-08-13 17:22 +0100
Last post2016-08-14 13:25 +0100
Articles 4 — 3 participants

Back to article view | Back to comp.os.linux.hardware


Contents

  Momentus XT sshd slowness? Peter Chant <pete@petezilla.co.uk> - 2016-08-13 17:22 +0100
    Re: Momentus XT sshd slowness? "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2016-08-14 05:21 -0400
      Re: Momentus XT sshd slowness? Richard Kettlewell <invalid@invalid.invalid> - 2016-08-14 10:38 +0100
      Re: Momentus XT sshd slowness? Peter Chant <pete@petezilla.co.uk> - 2016-08-14 13:25 +0100

#3178 — Momentus XT sshd slowness?

FromPeter Chant <pete@petezilla.co.uk>
Date2016-08-13 17:22 +0100
SubjectMomentus XT sshd slowness?
Message-ID<e18vpiFth4oU1@mid.individual.net>
Hi all.

Running a laptop with a Seagate Momentus XT 1TB hard drive that is a
couple of years old (the drive, the laptop is probably somewhat older).
Using btrfs on top of luks.  Think-pad T400 with 8GB of ram so quite
elderly.

Booting up and initial loading of applications (Firefox, Thunderbird
etc) is very slow.  It might have got slower lately but I have no
subjective measurements.

Looking at atop the machine is IO bound.  The disk IO is maxed but the
the CPUs are definitely not.  I wonder whether btrfs on top of luks is
causing problems for the flash cache in the sshd drive, whether the has
run out of cache due to wear or whether I'm trying to do to much with it
(which I don't think would be the case).

As a quick test I put an old ssd in it using ext4, without luks.  It
went like a rocket.  However, it is not like for like and doing a like
for like test would be a fair amount of hassle.

I suspect I should just splash out on a ssd (the old one I have is too
small for what I want).  The ssd will be worth more than the rest of the
laptop, but if it is IO bound only, the rest of the machine is OK then
it seems a reasonable course of action.  If I update the laptop the ssd
could go into the newer machine so the ssd would not be wasted on an update.

It looks like I've come to a conclusion, get a ssd.  However, I wonder
if there are any other views or similar experiences out there?


Pete

[toc] | [next] | [standalone]


#3179

From"David W. Hodgins" <dwhodgins@nomail.afraid.org>
Date2016-08-14 05:21 -0400
Message-ID<op.yl6hmhmta3w0dxdave@hodgins.homeip.net>
In reply to#3178
On Sat, 13 Aug 2016 12:22:09 -0400, Peter Chant <pete@petezilla.co.uk> wrote:

> Running a laptop with a Seagate Momentus XT 1TB hard drive that is a
> couple of years old (the drive, the laptop is probably somewhat older).
> Using btrfs on top of luks.  Think-pad T400 with 8GB of ram so quite
> elderly.

See the comments in
http://clients.teksavvy.com/~davidwhodgins/autofstrim

I run the script from cron.daily. Note that the filesystem must be
mounted when the script is run.

Most of the script is redundant now, as fstrim now supports the
option "-a", instead of having to manually find the file systems
on an ssd drive, but it works, so I haven't bothered changing it.

For a quick fix, just run (as root) "fstrim -a", with the file
systems mounted, after rebooting with the discard option(s) added.

Regards, Dave Hodgins

-- 
Change dwhodgins@nomail.afraid.org to davidwhodgins@teksavvy.com for
email replies.

[toc] | [prev] | [next] | [standalone]


#3180

FromRichard Kettlewell <invalid@invalid.invalid>
Date2016-08-14 10:38 +0100
Message-ID<87fuq7667z.fsf@LkoBDZeT.terraraq.uk>
In reply to#3179
"David W. Hodgins" <dwhodgins@nomail.afraid.org> writes:
> Peter Chant <pete@petezilla.co.uk> wrote:
>> Running a laptop with a Seagate Momentus XT 1TB hard drive that is a
>> couple of years old (the drive, the laptop is probably somewhat older).
>> Using btrfs on top of luks.  Think-pad T400 with 8GB of ram so quite
>> elderly.
>
> See the comments in
> http://clients.teksavvy.com/~davidwhodgins/autofstrim
>
> I run the script from cron.daily. Note that the filesystem must be
> mounted when the script is run.
>
> Most of the script is redundant now, as fstrim now supports the
> option "-a", instead of having to manually find the file systems
> on an ssd drive, but it works, so I haven't bothered changing it.
>
> For a quick fix, just run (as root) "fstrim -a", with the file
> systems mounted, after rebooting with the discard option(s) added.

What makes you think trimming is meaningful for a hybrid drive?

-- 
http://www.greenend.org.uk/rjk/

[toc] | [prev] | [next] | [standalone]


#3181

FromPeter Chant <pete@petezilla.co.uk>
Date2016-08-14 13:25 +0100
Message-ID<e1b69lFept6U1@mid.individual.net>
In reply to#3179
On 08/14/2016 10:21 AM, David W. Hodgins wrote:
> On Sat, 13 Aug 2016 12:22:09 -0400, Peter Chant <pete@petezilla.co.uk>
> wrote:
> 
>> Running a laptop with a Seagate Momentus XT 1TB hard drive that is a
>> couple of years old (the drive, the laptop is probably somewhat older).
>> Using btrfs on top of luks.  Think-pad T400 with 8GB of ram so quite
>> elderly.
> 
> See the comments in
> http://clients.teksavvy.com/~davidwhodgins/autofstrim

Added --allow-discards to cryptsetup

fstrim -v /mnt/system-pool

gave the following message FITRIM ioctl failed: Operation not supported.


> 
> I run the script from cron.daily. Note that the filesystem must be
> mounted when the script is run.
> 
> Most of the script is redundant now, as fstrim now supports the
> option "-a", instead of having to manually find the file systems
> on an ssd drive, but it works, so I haven't bothered changing it.
> 
> For a quick fix, just run (as root) "fstrim -a", with the file
> systems mounted, after rebooting with the discard option(s) added.
> 
> Regards, Dave Hodgins
> 

[toc] | [prev] | [standalone]


Back to top | Article view | comp.os.linux.hardware


csiph-web