[Soekris] USB errors on net5501

[Greg Troxel]

I have a

  net5501 (0.5G ram)
  Soekris 1.5A power supply
  40G pata drive
  vpn1411 coprocessor
  NetBSD (netbsd-5 branch)

and

  WD Elements 2T external drive, USB connector to net5501

I've been having two mysterious problems.

One is that (with or without USB disk), booting takes about 3 minutes.
This was true with PXE boot as well as booting from the 40G drive.  With
PXE, the delay was after pxeboot_ia32.bin was loaded over tftp, and
before that program printed its banner.  (I had to set consdev in the
pxe loader so NetBSD didn't try to write to vga, but the delay happens
anyway.)  I'll write about this separately later, but wanted to mention
it in case someone thinks it's related to the USB problem.

The other problem is that the USB disk seems to work but I had a lockup
when rsyncing lots of data to it.  The disk has a single gpt partition,
and on dk0 (the partition) I have a single UFS2 filesystem that I mount
with journaling (WAPBL, o log).  This kind of disk works fine on a
regular computer, and I wrote and read this particular entire disk to
test it (before using newfs).

I get only 12 MB/s read, but I understand that the USB controller is
slow.  The problem is that after copying several GB, I got:

Mar  6 17:17:32 foo /netbsd: umass0 at uhub1 port 1 configuration 1 interface 0
Mar  6 17:17:32 foo /netbsd: umass0: Western Digital Ext HDD 1021, rev 
2.00/20.02, addr 2
Mar  6 17:17:32 foo /netbsd: umass0: using SCSI over Bulk-Only
Mar  6 17:17:32 foo /netbsd: scsibus0 at umass0: 2 targets, 1 lun per target
Mar  6 23:03:43 foo /netbsd: umass0: BBB reset failed, TIMEOUT
Mar  6 23:04:48 foo /netbsd: umass0: BBB bulk-in clear stall failed, TIMEOUT
Mar  6 23:05:53 foo /netbsd: umass0: BBB bulk-out clear stall failed, TIMEOUT

which I believe provoked a panic in the filesystem code (a NetBSD issue
that I'll debug - I have a backtrace).  Following this, detaching and
reattaching the USB cable led to:

Mar  7 17:10:41 foo /netbsd: uhub1: device problem, disabling port 1

but rebooting restored the drive.  The 40G disk was not particularly
loaded during the failure event.

I'm going to try a few more times, and then switch to powering the
net5501 directly off a West Mountain Radio Rigrunner with a 30 Ah
battery (plus about 3-4 4 Ah batteries) that are on float charge, to see
if this is due to the power supply fading on current peaks.

My collected net5501 notes are at:

http://www.lexort.com/blog/net5501.html

but they have fairly little additional relevant content at this time.

Greg


pgppuzUUwJX1O.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Power Supply responses

[Greg Troxel]

  I don't mean to point fingers specifically at Soekris devices for
  being hard on power supplies; I don't think they are.  Rather it
  sounds like the average low power DC supply on the market today just
  isn't that great.  For most uses, say occasional use powering a
  portable hard drive, that's no big deal.  However, it seems like the
  Soekris device -- based on its bullet-proof, no-moving-parts,
  just-let-it-run design -- puts a continuous rather than sporadic
  demand on the power supply.  In the average year I suspect the power
  supply for my portable USB drive is on approximately 100-200 hours.
  By comparison my Soekris boxes run continuously, or 59808 hours per
  year.

I don't think Soekris boxes are particularly hard on power supplies, but
it's not clear 1.5A is adequate (or if the labeled-1.5A supplies really
do source 1.5A - I should put one on my West Mountain Radio CBA-2 and
see how it copes.).

I have now hooked up my net5501 to several 12V lead-acide batteries
(30Ah, 3*4Ah) on float charge via:

  http://www.westmountainradio.com/product_info.php?products_id=rr_4008_c
  http://www.powerwerx.com/batteries-chargers/power-tender-plus-12v.html

So far I have had no recurrences of the USB lockups, but it's only been
syncing for 3h.

If the box is reliable with the solid supply, I would say the 1.5A
Soekris-provided supplies are only adequate for a net5501 with no disk
and no external USB devices.

Someone should hook up a current probe to a storage scope and get some
real data



pgpPLZEdE9neV.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Power Supply responses

[Greg Troxel]

Bob Gustafson bob...@rcn.com writes:

 http://www.westmountainradio.com/product_info.php?products_id=rr_4008_c

Note that this is not a power supply; it's a distribution device, the
equivalent of an outlet strip but for 12V DC.

In my case, I got another USB lockup even when on large batteries, so
the issue was not about power.
I'm trying again without journaling.


pgpmwsDbsqINx.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] RAID1

[Greg Troxel]

der Mouse mo...@rodents-montreal.org writes:

 [...] possibility of the inclusion of a simple RAID1 chip [...]
 There is no need for a RAID chip in order to be able to boot from a
 RAID1.  I have been doing so in several machines using Linux's md
 capabilities for several years.

 Well, not everybody runs Linux.  One of the few advantages of firmware
 (rarely actually hardware, though for RAID1-only that's relatively
 plausible) RAID is that it is - or at least can be if done right -
 completely transparent to software.

FWIW, I use raidframe in NetBSD for RAID1.   With raid
autoconfiguration, either disk will boot in pretty much any computer.
Other than partitioning and installation, it's nearly transparent.

That doesn't argue against including a RAID chip, but I wouldn't use it.

Given the hw availability schedule for the 6501, surely this ship has
sailed.


pgp9siNRjq2rp.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] FreeBSD - Quick help needed...

[Greg Troxel]

(My experience is on NetBSD mostly, but I expect it's the same.)
When I have a disk with read errors, I do

  dd if=/dev/rwd0d of=/dev/null bs=32k

to find the bad blocks, and then use the # from /var/log/messages and:

  dd if=/dev/rwd0d of=/dev/null iseek=N count=1

and then

  dd of=/dev/rwd0d if=/dev/zero seek=N count=1

Be extra careful to always include the count=1!

This writes to the bad block, and then fsck can do whatever it can do.
It's been effective at getting machines back with fairly little damage.


The other thing you can do is to boot to single user, and then to just
mount the filesystem without fsck, and then ^D, and I think it will skip
the fsck.   For root of course mount -u -o rw /.  Then you can edit
out fsck from the rc scripts if necessary.



pgpjb26vJl64G.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] vpn1411 for ipsec under linux

[Greg Troxel]

Gerhard Rieger gerhard.rie...@linbit.com writes:

 I need to improve IPSec throughput on net5501 boxes under Linux
 Ubuntu Lucid but before purchasing vpn1411 cards I'd like to hear if
 someone successfully uses this accelerator card in this way?

My experience under NetBSD 5 is:

  builtin geode acceleration is faster than 1411 (openssl speed, not
  IPsec)

  both geode and 1411 support leads to lockups (probably a NetBSD bug)


are you using the geode acceleration now?


pgp9RaZsUpogO.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] USB port on 6501, 2GB memory configuration?

[Greg Troxel]

Ralph Becker-Szendy ra...@lr.los-gatos.ca.us writes:

 Two quick questions about the 6501.  First: According to the 
 documentation, it has USB 2.0 ports.  But so does the 5501, and on my 
 5501 under OpenBSD, the USB ports run at USB 1.0 speed.  Which is a 
 problem, because I need to occasionally connect a USB disk drive for 
 backups, and at USB 1.0 speed, that takes ludicrously long.  So does 
 anyone know what chipset on the 6501 provides the USB?  And maybe even 
 whether OpenBSD supports that chipset at 2.0 speed?  I'm still running 
 OBSD 4.6, but new 6501's would be installed with version 4.9 (but I 
 think the USB stack hasn't changed significantly).

For what it's worth, this is what happens under NetBSD 5.1_STABLE on a
5501:

ohci0 at pci0 dev 21 function 0: vendor 0x1022 product 0x2094 (rev. 0x02)
ohci0: interrupting at irq 7
ohci0: OHCI version 1.0, legacy support
usb0 at ohci0: USB revision 1.0
gcscehci0 at pci0 dev 21 function 1: vendor 0x1022 product 0x2095 (rev. 0x02)
gcscehci0: interrupting at irq 7
gcscehci0: EHCI version 1.0
gcscehci0: companion controller, 4 ports each: ohci0
usb1 at gcscehci0: USB revision 2.0
uhub0 at usb0: vendor 0x1022 OHCI root hub, class 9/0, rev 1.00/1.00, addr 1
uhub0: 4 ports with 4 removable, self powered
uhub1 at usb1: vendor 0x1022 EHCI root hub, class 9/0, rev 2.00/1.00, addr 1
uhub1: 4 ports with 4 removable, self powered
umass0 at uhub1 port 1 configuration 1 interface 0
umass0: Western Digital Ext HDD 1021, rev 2.00/20.02, addr 2
umass0: using SCSI over Bulk-Only
scsibus0 at umass0: 2 targets, 1 lun per target
sd0 at scsibus0 target 0 lun 0: WD, Ext HDD 1021, 2002 disk fixed

I get about 11.7 MB/s (dd from raw disk at 32k blocksize, and 12.6 MB/s
at 256k), which is slow for USB2, but definitely the right order of
magnitude - I think that drive does about 23 MB/s when connected to an
Intel desktop.

I suspect it's not so much the USB stack as the chipset glue for ehci
attachment.

So you can probably either use NetBSD code as a guide or just merge some
of it into OpenBSD.


pgpnrBvHMLFD3.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Second serial on net4511

[Greg Troxel]

Fred C soekris-f...@hidzz.com writes:

 When booting my net4511 the OS report 2 serials. Can someone tell me
 where I can find the connector
 for the second serial port and the pinout schema?

on the net5501, there is a 10-pin header and soekris sells a
cable/pci-slot-cover-with-db9 piece that brings this second port out.

http://soekris.com/products/accessories/serial-connector-2nd-port-pci-bracket.html

But it seems the 4511 doesn't have such a header, and it requires work
to get at the pins.


pgpqrnkAqzrdi.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Building an OpenBSD router

[Greg Troxel]

I'm doing pretty much exactly what you are proposing, except:

  I'm running NetBSD instead.

  I have a 40G laptop drive with the IDE bracket.


Beware that SSDs are sometimes thicker than laptop drives.   I believe
that the 9.5mm drives are what fits in the case.

On the 5501 you can't boot from USB, and the transfer rate to external
disks is about 1/2 what it is on regular desktops.

I would get the higher-end 5501 with more memory; it seems little enough
more $ and you can't upgrade later.  (I expect the box to be still
working fine in 5-10 years, maybe much longer.)

The standard case is nice.







pgpdsJuWNNJNi.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] net5501 crashes ?

[Greg Troxel]

I have a 500 MHz unit, 512M ram, which I think is the 5501-70, new in
early 2011, updated to the latest bios mid spring (bios was old on
arrival from the factory).

I installed a vpn1411 and a new-old-stock 40G 2.5 PATA disk.

My experience has been with NetBSD 5.[01]_STABLE, really tip of netbsd-5
branch since about February.  It's been mostly good, with two issues -
one that is pretty clearly due to NetBSD issues, and one about which I'm
not sure.

* huge disk not stable with rump mount or WAPBL

I have 2T disk on the USB port (no hub).  With a regular (traditional
kernel) mount (of a GPT partition), without WAPBL, it has been
completely solid - up for probably 6 months solid except for a 24h power
failure recently.   With a rump mount, the system crashes after a while,
I think due to leaks and issues with multithreaded processes abending,
and with WAPBL I think the 0.5G of ram is overstressed by the 2T drive
during steady large writes (it's a backup drive, basically).

This is 99.999% clearly not a hardware issue.

* crypto accelerators

With opencrypto drivers for the vpn1411 or for the geode security
coprocessor, I got some kind of lockups/crash with heavy ssh.

I am guessing this is a driver bug, but I really don't know.

I am virtually certain this is not a power supply issue as the board is
wired to about 30 Ah (original, tired, maybe 15 Ah now) total of
lead-acid batteries ganged together, with a 6A-capable float charger.
Even spikes to 10A wouldn't faze this setup, relative to the 5V
regulator.



Unfortunately NetBSD-5's envstat(8) reports no drivers, so I don't know
the CPU temperature.


My notes-to-self while figuring this out are at:

  http://www.lexort.com/blog/net5501.html

(The disk stability issue was only half figured out last I updated the
page.  But my machine with kernel mount, no wapbl, no accelerators is
completely stable.)





pgpz7zOgzbT3F.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] net5501 power supply input specifications

[Greg Troxel]

Attila Kinali att...@kinali.ch writes:

 I finally got around to measure the internal supply pins of my
 net5501 to see what exactly is going on. What i see are very very
 short (1us) drops of the 12 down that are larger than 1V. I guess
 the drops of the 3.3V and 5V supply that i see at the same time
 come from the 12V drop. But with the measurement equipment i have
 here and the lack of schmeatics, i cannot be sure. (it could be the
 other way as well)

You could try putting a capacitor across the supply.  1us duration drops
of more than 1V sound pretty suspicious though that something in the
output filtering of your supply is messed up.

It could also be that something is drawing outrageous currents at 3.3V which
is stressing the regulator's current-handling ability and this is
appearing upstream.  But that seems less likely than a supply with
something wrong (a bad output filtering cap, perhaps).

 Hence i'd like to try another power supply but i don't have any
 12V with 3A here, only a 16V. Unfortunately, there is no specification
 on what the net5501 can take as 12V suplly and whether these 12V
 are connected to the harddisk or PCI connectors. 

(Someone else answered this already.)

I have a hard time believing that a supply that will provide 12V 3A
would have trouble.   But I can believe that a particular supply labeled
12V 3A would not actually provide that reliably.

 Hence i would like to ask the following questions:
 * What is the exact specification of the 12 supply input?
 * Where are those 12V connected to?
 * Is there any stabilization or voltage limiter between the supply input
   and the harddisk and/or PCI connector?


I'm not soekris, obviously, but my limited understanding is

  the net5501 only uses 5V and lower, actually

  the disk connector is a notebook disk, which is at most a 5V device

  same with PCI

  There is a regulator (perhaps a 7805) that brings 12V down to 5V.  One
  can jumper to not use the regulator, and then the input has to be
  exactly 5V.

I am using a net5501 with the input connected to a bunch of 12V lead acid
batteries (in parallel), more or less a 12Ah, 4 7Ah and 4 4Ah, all on
float charge and thus ranging from high 11s to 14.5V, typically 13.4V or
so.  This has worked fine.


pgp2enjDACRD8.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Bridging OpenBSD

[Greg Troxel]

  I've bridged vr0 - vr3 and they show up in the bridge.  I've assigned an
  IP number to vr0 and serve DHCP to that ip range.  If I connect to vr0,
  I can get an address via DHCP.  If I connect to the other ports, then I
  cannot.   If I understand correctly, connections from vr1 - vr3 will be
  bridged to vr0 and will get DHCP from vr0 via the bridge, but that is
  not happening.

I wonder if the issue is that dhcp is implemented by bpf, and bpf raw
frames are not bridged.



pgpFKCXfSh7eh.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] OpenBSD disk transfer speed - net5501/6501

[Greg Troxel]

On a net5501 with NetBSD 5.1_STABLE (not what you asked, but close
enough to be interesting :-), I get

  gdt 2 ~  dd if=/dev/rwd0d of=/dev/null bs=128k
  ^C1747+0 records in
  1747+0 records out
  228982784 bytes transferred in 8.162 secs (28054739 bytes/sec)

This is with a very old PATA drive, circa 2003 I think:

  viaide0 at pci0 dev 20 function 2
  viaide0: Advanced Micro Devices CS5536 IDE Controller (rev. 0x01)
  viaide0: bus-master DMA support present
  viaide0: primary channel wired to compatibility mode
  viaide0: primary channel interrupting at irq 14
  atabus0 at viaide0 channel 0
  viaide0: secondary channel wired to compatibility mode
  viaide0: secondary channel ignored (disabled)
  wd0 at atabus0 drive 0: FUJITSU MHT2040AT
  wd0: drive supports 16-sector PIO transfers, LBA addressing
  wd0: 38154 MB, 77520 cyl, 16 head, 63 sec, 512 bytes/sect x 78140160 sectors
  wd0: 32-bit data port
  wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 5 (Ultra/100)
  wd0(viaide0:0:0): using PIO mode 4, Ultra-DMA mode 5 (Ultra/100) (using DMA)

but still 28 MB/s seems slow, even for an old notebook drive.  So I
wonder if the controller is maxing out before the drive.

(I'm still not sure if you are using dd to a file in the filesystem, or
if you're talking about using the raw device as I did above.)

What are other people seeing with PATA and SATA drives?

Can you post your dmesg?


pgpHkFJe1x64i.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] NetBSD with a net6501 and the SSD drive anyone?

[Greg Troxel]

Ben Greenfield b...@cogs.com writes:

 I'm looking for a new board and I have settled on the net6501. The
 features that sold me are the GPIO pins and the gigabit ethernet. I
 now see it also has an SSD drive. Has anyone used the Transend mSATA
 ssd with NetBSD?

 Is there any reason to not to expect it to work. NetBSD shouldn't need
 special support for a mSATA device should it?

I have a 6501 and the 32G ssd sitting at work, awaiting time to set it
up.  I expect it's going to be ok, and I'll send a note here with my
results.



pgpjz1vJNaN6L.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Using a net6501 as a home server

[Greg Troxel]

William Ahern will...@25thandclement.com writes:

 On Mon, Oct 15, 2012 at 11:44:30AM -0700, Wesley PA4WDH wrote:

 The host will be Gentoo, i haven't decided about the guest yet. It seems a
 bit overkill to use gentoo there too. The workload would be email with an
 MTA, Secure IMAP, Webmail and maybe some spam filtering. The number of
 emails would be somewhere in the 10s per day or so, so i guess that won't
 be a problem. My main question would be if it's responsive enough to be
 usable. Does anyone have any experience with virtualisation on an atom ?

 Excluding virtualization, it should be more than enough. I'm preparing a
 small mail and web server for hosting which will take on much more traffic
 than that. The only bottleneck I ran into was mutt loading my 200MB+ mbox
 files, but compared to an opteron it's only a few seconds longer. (It's
 actually mutt's pokey parsing; using my own custom mbox and MIME parser, I
 can load and parse the file as fast as the disk can send it--less than 2s.)

 FWIW, I'm using 2 20GB Intel SLC SSDs, one for /home and the other for /var.

 Throwing virtualization into the mix, though, gives me pause. Even with
 hardware vx instructions, virtual hosts still have a huge stack of software
 to go through for simple things like disk I/O, and these paths execute a
 lot. Like with mutt, there's not much overhead with the Atom to make up for
 suboptimal code paths. If you're worried about security, I'd use OpenBSD or
 Ubuntu (the former because unmaintained and unpatched it's still a hard
 target to break, and the latter because Debian Apt takes most of the burden
 out of maintenance.)

With xen and a pv guest, the virtualization overhead is low.  I have
measured about a 10% loss (not on a soekris) from raw disk in dom0 to
file in dom0, and also from file in dom0 to raw 'disk' in domU.  So
basically I concluded that disk IO suffers about 10% (with no activity
in dom0 or other domUs).

I have not tried xen on a 6501.



pgpxZwYz5K9zi.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Clock losing time on Net5501

[Greg Troxel]

Lars Noodén lars.noo...@gmail.com writes:

 Yes, I figured out that hardclock(9) was not a command.  What the
 solution looks like, is that OpenBSD sets the hardware clock during a
 proper shutdown or reboot.  That was not something I did often and over
 the months the clock drifted.  However, I was not able to find anything
 about the hardware clock in boot(8), reboot(8), or shutdown(8).

I'm not sure why OpenBSD only sets the hw clock during shutdown.  I get
it that in theory a clean shutdown will happen, but it would seem
sensible to set it often enough while up.  I'm not sure how disruptive
that is, though.  So you might read the kernel code, possibly implement
a callout to do it every 24h.  But really this should be brought up on
the OpenBSD kernel lists.

I use NetBSD, and I am actually not sure of how the hwclock is handled.
But I have not noticed the large drift issue.

Assuming this has not diverged from NetBSD, look for resettodr() in the
kernel.  NetBSD calls this when setting time.  It also calls it on
shutdown (on i386; this is MD).  But it doesn't seem to call it
periodically.



pgpP_NerACQna.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Clock losing time on Net5501

[Greg Troxel]

Stuart Henderson s...@spacehopper.org writes:

 So, in theory, this should force it to be written:

 # date $(date +%Y%m%d%H%M.%S)

Yes, but that will set fractional seconds to zero.  You probably want a
C program that does gettimeofday()/settimeofday()


pgp5eB7CUHWfC.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Problem on net5501 upgrading to FreeBSD 9.1 from 8.3

[Greg Troxel]

These lines are absent from the 9.1 boot:

  ad1: DMA limited to UDMA33, device found non-ATA66 cable
  ad1: 305245MB WDC WD3200BEVT-00ZCT0 11.01A11 at ata0-slave UDMA33 

So something about 9.1 is not finding the disk at the driver level.  But
since the bootloader read the kernel from the disk, it seems clear that
the BIOS found the disk ok.

Is 9.1 known to be fussier about ata interfaces?  Note the non-ATA66
cable on a UDMA100 controller.


pgp9Z42Mgpka2.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] RESOLVED: Problem on net5501 upgrading to FreeBSD 9.1 from 8.3

[Greg Troxel]

Corey Halpin chal...@cs.wisc.edu writes:

 On 2013-02-17, Corey Halpin wrote:
   I have a net5501-70 which is running FreeBSD 8.3.  It's got one internal 
 SATA HD from which it boots, and also has a couple of external drives 
 connected via a USB hub.  No expansions cards are installed. 
 
   I attempted to update it to 9.1-RELEASE following the instructions here:
   http://www.freebsd.org/releases/9.1R/installation.html#AEN36
 
   When I reboot into the 9.1 kernel, the boot process chokes with:
 
   Mounting from ufs:/dev/ufsid/49fabde14312a481 failed with error 19.

   I've discovered that if I set FLASH=Secondary, such that the hard drive 
 appears as the primary, then the machine boots properly.

That sounds like a potential bug in 9.1, where it doesn't find a
secondary drive if the primary one is missing.


pgp8M4UO_l7we.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Soekris] net6501, booting from USB, NetBSD install

[Greg Troxel]

I have a net6501, and updated the firmware to 1.41c.  I have an mSATA
SSD (bought from Soekris), which seems to be recogized.  I am trying to
boot off USB, and have tried

  698G drive - hangs (apparent known issue)

  IDE USB adapator with 111G drive - not recognized by BIOS

  8G thumbdrive - shows up, but booting pauses for a while and just says
  failed:

8G failure:

  81: USB 01  Xlt -2-32   Mbyte

   boot 81

  No Boot device available, enter monitor.


  comBIOS Monitor.   Press ? for help.

It takes about 14s from 'boot 81' to the error.  Presumably 80 is the
(emtpy) mSATA drive.

My flash drive contained:

  hand-constructed bootable image, with NetBSD mbr, ffvs1 with
  bootxx_ffsv1, and bits trom installcd filesystem

and then just to make sure, I used

ftp://ftp.netbsd.org/pub/NetBSD/NetBSD-6.1/images/NetBSD-6.1-i386-install.img.gz

written uncompressed instead.  Same behavior.

I am using the Soekris-provided 3A power supply.

I will keep trying, and probably also try PXE booting.

Does the boot-from-USB bios code have any particular requirements?
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] net6501, booting from USB, NetBSD install

[Greg Troxel]

Following up to myself, I then tried a different USB/IDE adaptor:

  A newer CablesToGo IDE/SATA adaptor seemed to work, and resulted in the
  second line below.

  #+BEGIN_EXAMPLE
  Soekris USB Expansion ROM ver. 1.01  20111203

  81: USB 01  Xlt -2-32   Mbyte
  81: USB 01  ST3120026A  Xlt 1024-255-63  117 Gbyte0 11
  #+END_EXAMPLE

Booting was fine; I just had to drop to the boot prompt and 'consdev
com0'.  Installation of NetBSD 6.1 on the mSATA SSD (as wd0) was then
uneventful.

Curiously, after booting NetBSD from wd0, the pesky Verbatim thumbdrive
was recognized and worked just fine.

In case anyone is curious, my collected notes are at:

http://www.lexort.com/blog/net5501.html



Greg Troxel g...@work.lexort.com
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Redundant Power Supply or battery backup for net6501

[Greg Troxel]

Chris Boot bo...@bootc.net writes:

 On 01/08/2014 17:54, Andreas Steinel wrote:
 On Fri, Aug 1, 2014 at 12:49 PM, Greg Troxel g...@lexort.com wrote:
 I have a net5501, and measured that with a 2.5 disk (not ssd) it draws
 about 500 mA at 12V.  I have it connected to several sealed lead-acid
 batteries in parallel, typically a 12Ah and two 7Ah.  These batteries
 are on float charge to keep them at 13.5V when there is AC power from
 the utility.  When the utility power fails, the net5501 continues to
 run, and should last for many hours.
 
 Thank you for sharing your solution. How is the (complete) schematic
 for your setup (I saw the fuses and bypass capacitor missing schematic
 of phk which uses a similar setup as yours)? Is there also a charge
 protecting circuit involved or is the battery charging restricted
 (e.g. delta-V)?

 Personally, I use Mini-Box OpenUPS devices:

 http://www.mini-box.com/OpenUPS
 http://www.mini-box.com/OpenUPS2

 They have proper charge controllers on board for all kinds of battery
 chemistries, and work really nicely.

What I do is to provide 12V via the 5.5x2.1 barrel connector to the
net5501 in a case, just as if the wire were coming from a wall wart.

Each battery has a fuse in the positive lead, and then Anderson
Powerpole connectors, and is plugged into a West Mountain Radio
RigRunner.   I then have a 12V wall wart (that produces more) via a
cheap solar charge controller hooked up to the RigRunner as well.

This sounds a little kludgy, and it sort of is, but I'm coming at this
from having parts for ham radio emergency use of 12V batteries and a
very small off-grid solar system, so it works for me.

Compared to all of this, if you don't already have it and aren't already
familiar, the OpenUPS looks really nice.  And, it seems to have
low-voltage shutdown, monitoring, etc.  Plus, you can set the output for
6V so that you lose less in the regulator.

The picoUPS looks interesting too, but needs to be a in 12V world, it seems.

Putting this inside the rackmount case may or may not work.  Where my
net5501 is I have enough other space that I haven't had to worry about
this.

You can also just plug the 12V power supply into a regular 120VAC UPS
(or rather your local normal line voltage).


___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Limitations of Net6501 as a network bridge

[Greg Troxel]

Jed Clear jcl...@speakeasy.net writes:

 I haven't explored iptables in bridge/layer 2 mode, but there is no
 fundamental reason you can't packet sniff or firewall traffic in
 bridge mode. The traffic has to pass through your kernel.

I don't know about iptables, but ipfilter in NetBSD can do filtering by
matching on IP headers for bridges.

With snort, you should be able to pick any of the bridged interfaces,
but once you start firewalling you probably want the WAN-facing one.


pgpuE7a9dOpiM.pgp
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] I see development of the net6801 has been dropped.

[Greg Troxel]

Bob Bishop <r...@gid.co.uk> writes:

> HI,
>
>> On 22 Oct 2015, at 01:45, Greg Troxel <g...@lexort.com> wrote:
>> 
>> I have a net5501 (with 40G PATA spinning disk) that I run on 12V, from
>> several AGM batteries and a float supply.  I find that it draws 0.5A at
>> 12-13V, which is about 6W.  However, if I fed it 6V instead, it would
>> draw the same 0.5A and be about 3W.
>
> Nope. The switching regulator isn’t very lossy so it would draw about
> 1A at 6V. Try it.

Interesting; I will have to try that.  IIRC (long ago, so unreliable)
From reading the schematic it was a 7805, which is not switching, and I
don't remember seeing anything big enough to be a buck downconverter,
but I could well be wrong.


signature.asc
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] I see development of the net6801 has been dropped.

[Greg Troxel]

Christopher Sean Hilton <ch...@vindaloo.com> writes:

> On Wed, Oct 21, 2015 at 08:45:47PM -0400, Greg Troxel wrote:
>> Presumably you are using a 110V input?  (and I wonder if a Kill-A-Watt?)
>> 
>> I have a net5501 (with 40G PATA spinning disk) that I run on 12V, from
>> several AGM batteries and a float supply.  I find that it draws 0.5A at
>> 12-13V, which is about 6W.  However, if I fed it 6V instead, it would
>> draw the same 0.5A and be about 3W.
>> 
>> I know the net6501 runs on 12V (or really enough about 5V to make the
>> regulator work) as well, and that's attractive.  Are there options for
>> the supermicro, or anything else, for 12V, other than a regular
>> inverter?
>
> I'm on 110AC and yes the meter is a Kill-a-watt. 
>
> The SuperMicro machine would certainly not work on 12V. 
>
> I not saying that the SuperMicro machine can replace the Soekris in
> every use case. For my usage though, and that of anyone who's gonna
> plug it into the wall for power, I'm saying it's not a bad deal.

Thanks for the clarification.  Indeed, 12V is not normal and your point
that it doesn't use much wall power is totally fair.


signature.asc
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] I see development of the net6801 has been dropped.

[Greg Troxel]

Christopher Sean Hilton  writes:

> On Wed, Oct 21, 2015 at 11:28:27PM +0200, Iustin Pop wrote:
>
>> Given that just the Atom processor on the Supermicro has a TDP of 20W, I
>> doubt that the whole system is not significantly more. Otherwise, why
>> would they ship with a *200W* PSU?
>
> That's the same processor that was in the proposed Net6801-70 (Atom
> C2758) Note well that the Net6801-70 is listed as system power between
> 9W Idle and 31W active: http://beta.soekris.net/net6801.html. I'm not
> sure that TDP is a good indicator of the power that the CPU uses. 
>
> In my case I took the precaution of running my SuperMicro box on a
> Watt Meter for a couple of weeks. It averaged 25W. The processor TDP
> is 13W and the power supply is a 200W unit. My box has 4GB of RAM and
> 2.5" form factor 64GB SSD. My Soekris Net6501-50 with the same hard
> drive averages 18W on the same watt meter. 
>
> I suspect the 200W on the power supply is the maximum it can supply
> and that it's the smallest power supply that SuperMicro has.

Presumably you are using a 110V input?  (and I wonder if a Kill-A-Watt?)

I have a net5501 (with 40G PATA spinning disk) that I run on 12V, from
several AGM batteries and a float supply.  I find that it draws 0.5A at
12-13V, which is about 6W.  However, if I fed it 6V instead, it would
draw the same 0.5A and be about 3W.

I know the net6501 runs on 12V (or really enough about 5V to make the
regulator work) as well, and that's attractive.  Are there options for
the supermicro, or anything else, for 12V, other than a regular
inverter?


signature.asc
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Dying net6501 servers

[Greg Troxel]

I have a few net6501 that I am not yet using (have been sitting powered
off).  I am wondering how much is understood about the failure mode, and
if setting the clock frequency much lower (half or quarter of max) would
reduce heat and the likelihood of failure.  For most of my uses, I need
a box to do something simple (NTP, low-speed router), and really a
net5501 is adequate.


signature.asc
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] Dying net6501 servers

[Greg Troxel]

Brook Milligan <br...@nmsu.edu> writes:

> On Sep 2, 2016, at 9:49 AM, Greg Troxel <g...@lexort.com> wrote:
>> I have a few net6501 that I am not yet using (have been sitting powered
>> off).  I am wondering how much is understood about the failure mode, and
>> if setting the clock frequency much lower (half or quarter of max) would
>> reduce heat and the likelihood of failure.  For most of my uses, I need
>> a box to do something simple (NTP, low-speed router), and really a
>> net5501 is adequate.
>
> Interesting idea.  I have a box I would be willing to try this on, as
> I have it doing similar (generally low-power) things.  How does one
> control the clock frequency?

On NetBSD, see est(4), and machdep.est.frequency.*.

I have used this on notebooks.  It doesn't appear on my net5501 under
netbsd-6.  An i5-2310 (desktop) shows multiple frequencies but I have
never tried to turn that down.



signature.asc
Description: PGP signature
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] RIP net5501

[Greg Troxel]

Nicolas  writes:

> Alleluïa, it's alive !!!
>
> Thank you all for your advice.
>
> The PSU was indeed faulty. I quickly tested the sokris box with a
> similar PSU, and LEDs turned on !!
>
> I just need to find the right input jack (2.1x5.5 mm), and I'll have a
> working net5501.

You have a broken power supply with a cord with the right connector, so
you should be all set.   I've cut those cords and installed anderson
powerpoles (amateur radio standard 12V connenctor), letting me use
different supplies/equipment.
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech

Re: [Soekris] So long and thanks for all the fish!

[Greg Troxel]

Good luck and I have also appreciated the advice and camraderie of this
list.

I am running NetBSD, and my net5501 bought in 2011 is still working just
fine.  I am eyeing an apu4c4, which feels like it might be like a
net8501 if such a thing existed and were reliable.   But the edgerouter
lite is interesting too.  I know people have run NetBSD on the ERL3:

  https://blog.netbsd.org/tnf/entry/hands_on_experience_with_edgerouter

and haven't been following the newer ones.

It sounds like there is OpenBSD support for that (yay) and I wonder how
different the 6P vs 3 is hardware wise.

Perhaps one of the BSDs can host a mailinglist for running BSD/Linux (or
maybe Linux people have a home?) on headless reliable multi-interface
boxes, which I think these days is pc engines and edgerouters.  Perhaps
some other-fruit Pi with multiple interfaces and enough ram.

Greg

___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech