Debian on Dell Kace M300

thank you for all the help

Ray,

> I recently purchased some Lexar 128GB SSD from
> Amazon
> for $10.49 that are a perfect fit after removal
> from case to mount in replacement of the stock
> 16GB device. Pictures attached.
> I suspect the 256GB, 512GB and 1TB versions would
> fit as well.

So this SSD case has 2 pieces glued togother, and you'd just pry them open, i.e no screws?

-bodhi
===========================
Forum Wiki
bodhi's corner

bodhi Wrote:
-------------------------------------------------------
> Ray,
>
> > I recently purchased some Lexar 128GB SSD from
> >
> Amazon
> > for $10.49 that are a perfect fit after removal
> > from case to mount in replacement of the stock
> > 16GB device. Pictures attached.
> > I suspect the 256GB, 512GB and 1TB versions
> would
> > fit as well.
>
> So this SSD case has 2 pieces glued togother, and
> you'd just pry them open, i.e no screws?

No screws, just pops apart with a little pressure around the seams.

Ray

> I recently purchased some Lexar 128GB SSD from
> Amazon
> for $10.49 that are a perfect fit after removal
> from case to mount in replacement of the stock
> 16GB device. Pictures attached.
> I suspect the 256GB, 512GB and 1TB versions would
> fit as well.

I tried this Lexar SSD in the M300 SATA slot. Nice form factor, but damn it is too slow! as slow as my USB flash drive. I guess we'd get what we pay for.

-bodhi
===========================
Forum Wiki
bodhi's corner

> I tried this Lexar SSD in the M300 SATA slot. Nice
> form factor, but damn it is too slow! as slow as
> my USB flash drive. I guess we'd get what we pay
> for.

However, perhaps it should be tried on another box so see if something about this board that has slowed it down.

-bodhi
===========================
Forum Wiki
bodhi's corner

bodhi Wrote:
-------------------------------------------------------
> > I tried this Lexar SSD in the M300 SATA slot.
> Nice
> > form factor, but damn it is too slow! as slow
> as
> > my USB flash drive. I guess we'd get what we
> pay
> > for.
>
> However, perhaps it should be tried on another box
> so see if something about this board that has
> slowed it down.

i bought 3 different ssd drives, and did some bench marks on them. the 256 gig is really slow as well, like usb stick snail speed. if your interested i can do the other 2 again, which were really fast and give you the results from each and models. i only paid like 10-13 dollars each.

@mrc333777,


> if your
> interested i can do the other 2 again, which were
> really fast and give you the results from each and
> models. i only paid like 10-13 dollars each.

That would be great ! please do.

-bodhi
===========================
Forum Wiki
bodhi's corner

bodhi Wrote:
-------------------------------------------------------
> @mrc333777,
>
>
> > if your
> > interested i can do the other 2 again, which
> were
> > really fast and give you the results from each
> and
> > models. i only paid like 10-13 dollars each.
>
> That would be great ! please do.


dont know how much this will help but this is the best i could come up with.
i used ontrack easy recovery disk diagnostics to get these numbers.

first set of numbers is from a usb 3.0 port, second set up numbers is from
usb 2.0 port.

old usb 3.0 Seagate FreeAgent GoFlex controller from an external drive
240 MB/s , 40 mb/s - inland professional SSD 120GB P/N: 120GBM
240 MB/s , 40 mb/s - CORE ortial.com , OC-150-128 128GB 6Gb sata ssd tlc
128 MB/s , 40 mb/s - oem ssd 256GB sata III ssd s3105 series

super speed usb 3 controller from some generic hd enclosure got the numbers from all 3 drives
usb 3.0 port 400 mb/s , usb 2.0 port 48 mb/s

old usb 3.0 Seagate FreeAgent GoFlex controller from an external drive

240 MB/s , 40 mb/s - inland professional SSD 120GB P/N: 120GBM
240 MB/s , 40 mb/s - CORE ortial.com , OC-150-128 128GB 6Gb sata ssd tlc
128 MB/s , 40 mb/s - oem ssd 256GB sata III ssd s3105 series

What is exactly the 240 MB/s and 40 mb/s in this benchmark? Are they sequential read and random read speeds?

Quote

i used ontrack easy recovery disk diagnostics to get these numbers.

I think it would be best if you plug in the SSD drives to the SATA slot.

Assuming the SDD is mount at /media/sda1.

Test write

cd /media/sda1
dd if=/dev/zero of=test_write.img bs=1G count=1 oflag=dsync

Test read

hdparm -Tt /dev/sda

-bodhi
===========================
Forum Wiki
bodhi's corner

The Lexar 128G SSD read speed in this M300 box is slow, but not too bad.

# hdparm -Tt /dev/sda

/dev/sda:
 Timing cached reads:   672 MB in  2.00 seconds = 335.35 MB/sec
 Timing buffered disk reads: 486 MB in  3.01 seconds = 161.65 MB/sec

The write speed is about half as slow as my spinning rust HDD on another box.

# dd if=/dev/zero of=test_write.img bs=500M count=1 oflag=dsync

1+0 records in
1+0 records out
524288000 bytes (524 MB, 500 MiB) copied, 11.7481 s, 44.6 MB/s

Note that the "Timing cached reads" is also half as slow as of my spinning rust HDD on another box.

All and all, it's a great improvement over USB drive. But not up to the performance we usually see with regular HDDs.

-bodhi
===========================
Forum Wiki
bodhi's corner

mrc333777 Wrote:
-------------------------------------------------------
> bodhi Wrote:
> -------------------------------------------------------
> > @mrc333777,
> >
> >
> > > if your
> > > interested i can do the other 2 again, which
> > were
> > > really fast and give you the results from
> each
> > and
> > > models. i only paid like 10-13 dollars each.
> >
> > That would be great ! please do.
>
>
> dont know how much this will help but this is the
> best i could come up with.
> i used ontrack easy recovery disk diagnostics to
> get these numbers.
>
> first set of numbers is from a usb 3.0 port,
> second set up numbers is from
> usb 2.0 port.
>
> old usb 3.0 Seagate FreeAgent GoFlex controller
> from an external drive
> 240 MB/s , 40 mb/s - inland professional SSD 120GB
> P/N: 120GBM
> 240 MB/s , 40 mb/s - CORE ortial.com , OC-150-128
> 128GB 6Gb sata ssd tlc
> 128 MB/s , 40 mb/s - oem ssd 256GB sata III ssd
> s3105 series
>
> super speed usb 3 controller from some generic hd
> enclosure got the numbers from all 3 drives
> usb 3.0 port 400 mb/s , usb 2.0 port 48 mb/s


Yeah that doesn't work, that controller may be USB 3.0 but the numbers I'm seeing there shows it's maxing out aganist a SATA II bottleneck. it's limited to 3Gbit and not 6Gbit. For an external spinner this is fine, as not many can go over 100MB/s, and no 2.5s on a consumer level can even reach that yet.

From the numbers I've seen, the lexar drives are fine, but lack DRAM cache. they'll be okay so long as you're not manipulating lots of small files which will create I/O overhead.

bodhi Wrote:
-------------------------------------------------------
> The Lexar 128G SSD read speed in this M300 box is
> slow, but not too bad.
Yes the Lexar NS100 SSD is not really meant for speed, it was designed as a cheap replacement for spinning disks. The controller is an embedded DRAM-less design which means it was designed for cost not speed. The Lexar NQ100 SSD is only slightly more expensive and should provide a little better performance, but also was not designed for speed.

Wanted to give a PSA to anyone looking to try the Cockpit web management package from main or backports on your Kace, it forcibly changes your init system to systemd upon install. On the Kace this isn't an issue because there's plenty of RAM, and no extra RAM seems to be used much, but on other systems with 128MB or 256MB of RAM this may pose a small issue later on. I didn't need to change any init scripts or anything in the uboot, it just worked as normal, which is weird to me. but it works, and cockpit works.

So far I didn't even notice it did this until like last week, and the same box just had a bookworm update with no issues to report aside from the ssh key being regenerated during update. As this is the test machine and the other Kace is doing things, I'm going to wait until bookworm goes full-release to upgrade everything else, but it seems like everything went smoothly so far using the 6.2.7 kernel.

In the future as more things get incorporated into systemd, as much as I'm against its existence, it's finally mature enough I think it'd be an okay thing to start transitioning over the rootfs images to systemd init completely, and have a different rootfs variant available for anyone that wants to stick with sysvinit (i.e. 128/256M RAM devices, 512+ shouldn't have an issue with systemd). So many packages are starting to rely on it that it is becoming quite a problem for anyone using sysvinit, to such an extent that any init scripts orphaned by the package maintainers that have wholly moved over to systemd are now in a separate package that likely won't last. Specifically Intel's IWD wireless daemon, which aims to replace the likes of packages like wicd, has such an init script that has been orphaned and is now included in such a script. I learned this the hard way many months ago utilizing it on my pogo v4 when wicd wouldn't start the network connection automatically and came to find out that it was super bugged.

@sudos,

https://forum.doozan.com/read.php?2,35581,135216#msg-135216

-bodhi
===========================
Forum Wiki
bodhi's corner