New Alix board for 2013
-
I would like some feedback from anyone who has put the nanoBSD image (2.1.1 or 2.2) onto an SD card and run it. I guess you can put an SD card into a Windows device and write it with phydiskwrite in the same way as a CF card? And then put it in the APU, select the correct boot device, boot, set the interface from the console, and go?
I haven't put one into production, but I'm testing using nano (2.1.1-pre i386) off an SD card. Used physdiskwrite and a card reader, then put the card in the apu. You have to set the cam.boot.delay as mentioned in previous posts. The apu will boot automatically from any storage it finds, there should be no reason to get into the console.
-
Just to correct the pfSense and APU support :
- i was new with the APU and pfSense product and system
- the main trick is the serial console speed change between the APU and pfSense to know (115K vs 9600)
For the rest pfSense on the APU is working like a charm and i plan to purchase more APU for different locations…
APU + pfSense is a win-win solution : great work ! -
So you're not refering to this one:
http://pcengines.ch/msata16a.htmBecause from the image they are different brands.
The one from pcengines certainly works. (albeit the mentioned dma warnings when the board is overloaded).That PC Engines reference has a part S238 and so does the NetGate store. The PC Engines picture has 1 memory chip soldered on 1 side of the board, and the other on the reverse side. The NetGate picture has them both soldered on the same side - maybe it depends how the production supervisor is feeling on the day? :)
PC Engines warns "Not compatible with Sophos UTM", NetGate warns "Not compatible with Sophos UTM or pfSense".
-
- 50 concurrent iperf connections from PC1 to PC2
- 50 concurrent iperf connections from PC2 to PC1
Could you please show us your iperf command options so I could test it using a different brand of mSATA? I'm using Kingston SSDNow SMS200S3/120G and I never got any DMA errors using it even when I ran iperf.
I used to get the same DMA errors a lot on a x79 Chipset/LGA2011 CPU with Kingston SSDNow V300 Series SV300S37A/240G Drive connected to a SATA 6.0 Gbps port and running as squid AUFS cache. after changing it to diskd I rarely get them ever again. I also changed vfs.read_max=32 to vfs.read_max=128 and kern.ipc.nmbclusters to a value of 32768 but these two by themselves weren't enough.
I'm guessing it all has to do with drive & port transfer rate, but why would iperf saturate the drive in the first place? the msata16a is SATA II only (and performs lower than that) but the APU board ports should be using SATA III 6.0 Gbps according to AMD docs.
-
Don't have the exact command-history (i'm at home, test was done at office).
However it was something like:
server PC1: iperf -s -i 10 | grep "[SUM]"
client PC1: iperf -i 10 -t 99999 -P 50 -c 10.0.42.2 | grep "[SUM]"
server PC2: iperf -s
client PC2: iperf -t 99999 -P 50 -c 10.0.42.237 -
My APU.1C only shows 3 Gb/s
ad4: 15258MB <supersspeed s328="" 16gb="" v4629=""> at ata2-master UDMA100 SATA 3Gb/s</supersspeed>
Does anyone know what that ZFS Warning tells me in dmesg? "Expect unstable behavior" sounds not that good.
Trying to mount root from ufs:/dev/ad4s1a ZFS NOTICE: Prefetch is disabled by default if less than 4GB of RAM is present; to enable, add "vfs.zfs.prefetch_disable=0" to /boot/loader.conf. ZFS WARNING: Recommended minimum kmem_size is 512MB; expect unstable behavior. Consider tuning vm.kmem_size and vm.kmem_size_max in /boot/loader.conf.
I already set
vm.kmem_size="512000000" vm.kmem_size_max="768000000"
in /boot/loader.conf.local but get the warning anyway.
Greets, Jud3x
-
ZFS is completely off-topic, plus not used at all… someone perhaps might mute the stupid noise during boot.
-
Running iperf with GruensFroeschli's commands:
server PC1: iperf -s -i 10
client PC1: iperf -i 10 -t 99999 -P 50 -c 10.0.42.2[ 4] 0.0-10.0 sec 533 MBytes 448 Mbits/sec [ 4] 10.0-20.0 sec 557 MBytes 467 Mbits/sec [ 4] 20.0-30.0 sec 518 MBytes 435 Mbits/sec [ 4] 30.0-40.0 sec 525 MBytes 441 Mbits/sec [ 4] 40.0-50.0 sec 525 MBytes 440 Mbits/sec [ 4] 50.0-60.0 sec 525 MBytes 440 Mbits/sec [ 4] 60.0-70.0 sec 530 MBytes 445 Mbits/sec [ 4] 70.0-80.0 sec 527 MBytes 442 Mbits/sec [ 4] 80.0-90.0 sec 584 MBytes 490 Mbits/sec ...
top showing this while iperf is running (client is on re0):
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND 11 root -68 - 0K 304K RUN 0 3:26 66.99% [intr{irq256: re0}] 11 root -68 - 0K 304K CPU0 0 1:46 31.98% [intr{irq258: re2}]
no DMA errors/warnings at all.
My APU.1C only shows 3 Gb/s
ad4: 15258MB <supersspeed s328="" 16gb="" v4629=""> at ata2-master UDMA100 SATA 3Gb/s</supersspeed>
This is what I'm seeing on mine:
$ dmesg | grep SATA atapci0: <ati 800="" ixp700="" sata300="" controller="">port 0x4020-0x4027,0x4040-0x4043,0x4028-0x402f,0x4044-0x4047,0x4000-0x400f mem 0xfeb04000-0xfeb043ff at device 17.0 on pci0 ad4: 114473MB <kingston sms200s3120g="" 507abbf0="">at ata2-master UDMA100 SATA 6Gb/s</kingston></ati>
Here's the confusing part:
SATA300 is 300 MB/s (3Gbps)
UDMA100 is 100 MB/s
but then it states SATA 6Gbps at the end…difference between AHCI and legacy IDE should give you around 15% performance boost not much more so not sure what is wrong here.
Is there a way to benchmark drives under pfSense?
Thanks.
-
ZFS is completely off-topic, plus not used at all… someone perhaps might mute the stupid noise during boot.
I've actually considered making it more, rather than less present in pfSense.
But go ahead, keep on keepin' on about things.
-
Is there a way to benchmark drives under pfSense?
Well if nothing else you could just dd to null to get an idea of sequential transfer speed.
-
@gonzopancho:
ZFS is completely off-topic, plus not used at all… someone perhaps might mute the stupid noise during boot.
I've actually considered making it more, rather than less present in pfSense.
But go ahead, keep on keepin' on about things.
Merely pointing out there's a noisy, scary, confusing, misleading warning making people do completely unnecessary "tweaks" .
As for making it more present - not interested in "cool factor" stuff. This is a firewall/router and the thing eats RAM for lunch. Geez. ::)
-
Not that I would need blazing speeds on a pfSense router, but today I tested SSD throughput on the APU for the above mentioned
ad4: 15258MB <supersspeed s328="" 16gb="" v4629=""> at ata2-master UDMA100 SATA 3Gb/s</supersspeed>
with
dd if=/dev/zero of=/tmp/delme bs=12M count=100; rm /tmp/delme
and I got:
100+0 records in 100+0 records out 1258291200 bytes transferred in 47.450329 secs (26518071 bytes/sec)
which is really bad.
-
What SSD is that?
-
That is the one sold by PC Engines
http://www.pcengines.ch/msata16a.htm
I read a review of the APU and this SSD and the guy at ipfire,org got the same bad results. He also states, that it is a cheap SSD and would buy another. http://planet.ipfire.org/post/pc-engines-apu1c-a-review
Greets, Jud3x
-
Yeah that's just down the the SSD itself rather than the board then, and it's pretty much what I'd expect for an SSD in that price range.
Read speed is far better than write speed though, and TBH I'm not sure what you'd want to be constantly writing at high speed to a small SSD like this for anyway, not least because of the limited write cycles of NAND.
-
Anyone else having issues bridging the LAN interfaces? I get intermittent connectivity when I try and I can see the physical connection dropping. It should be noted that as soon as I remove the bridge I no longer get connection drops.
LAN: bridge0
MGMT: re1
WIFI: ath0
WIRED: re2BRIDGE0 Members: WIFI, WIRED
System Log:
Mar 30 19:13:03 syslogd: kernel boot file is /boot/kernel/kernel Mar 30 19:13:05 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:05 check_reload_status: rc.newwanip starting re2 Mar 30 19:13:08 php: rc.newwanip: rc.newwanip: Informational is starting re2. Mar 30 19:13:08 php: rc.newwanip: rc.newwanip: on (IP address: ) (interface: opt3) (real interface: re2). Mar 30 19:13:08 php: rc.newwanip: rc.newwanip: Failed to update opt3 IP, restarting... Mar 30 19:13:08 check_reload_status: Configuring interface opt3 Mar 30 19:13:11 php: rc.interfaces_wan_configure: The command '/sbin/ifconfig 're2' inet delete' returned exit code '1', the output was 'ifconfig: ioctl (SIOCDIFADDR): Can't assign requested address' Mar 30 19:13:11 check_reload_status: Linkup starting re2 Mar 30 19:13:11 kernel: re2: link state changed to DOWN Mar 30 19:13:13 check_reload_status: Linkup starting re2 Mar 30 19:13:13 kernel: re2: link state changed to UP Mar 30 19:13:14 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:16 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:16 check_reload_status: rc.newwanip starting re2 Mar 30 19:13:18 php: rc.newwanip: rc.newwanip: Informational is starting re2. Mar 30 19:13:18 php: rc.newwanip: rc.newwanip: on (IP address: ) (interface: opt3) (real interface: re2). Mar 30 19:13:18 php: rc.newwanip: rc.newwanip: Failed to update opt3 IP, restarting... Mar 30 19:13:24 php: rc.interfaces_wan_configure: The command '/sbin/ifconfig 're2' inet delete' returned exit code '1', the output was 'ifconfig: ioctl (SIOCDIFADDR): Can't assign requested address' Mar 30 19:13:24 check_reload_status: Linkup starting re2 Mar 30 19:13:24 kernel: re2: link state changed to DOWN Mar 30 19:13:26 check_reload_status: Linkup starting re2 Mar 30 19:13:26 kernel: re2: link state changed to UP Mar 30 19:13:26 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:29 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:29 check_reload_status: rc.newwanip starting re2 Mar 30 19:13:32 php: rc.newwanip: rc.newwanip: Informational is starting re2. Mar 30 19:13:32 php: rc.newwanip: rc.newwanip: on (IP address: ) (interface: opt3) (real interface: re2). Mar 30 19:13:32 php: rc.newwanip: rc.newwanip: Failed to update opt3 IP, restarting... Mar 30 19:13:32 check_reload_status: Configuring interface opt3 Mar 30 19:13:34 php: rc.interfaces_wan_configure: The command '/sbin/ifconfig 're2' inet delete' returned exit code '1', the output was 'ifconfig: ioctl (SIOCDIFADDR): Can't assign requested address' Mar 30 19:13:35 check_reload_status: Linkup starting re2 Mar 30 19:13:35 kernel: re2: link state changed to DOWN Mar 30 19:13:37 kernel: re2: link state changed to UP Mar 30 19:13:37 check_reload_status: Linkup starting re2 Mar 30 19:13:37 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:39 php: rc.linkup: Hotplug event detected for WIRED(opt3) but ignoring since interface is configured with static IP ( ) Mar 30 19:13:39 check_reload_status: rc.newwanip starting re2 Mar 30 19:13:42 php: rc.newwanip: rc.newwanip: Informational is starting re2. Mar 30 19:13:42 php: rc.newwanip: rc.newwanip: on (IP address: ) (interface: opt3) (real interface: re2). Mar 30 19:13:42 php: rc.newwanip: rc.newwanip: Failed to update opt3 IP, restarting... Mar 30 19:13:47 php: rc.interfaces_wan_configure: The command '/sbin/ifconfig 're2' inet delete' returned exit code '1', the output was 'ifconfig: ioctl (SIOCDIFADDR): Can't assign requested address' Mar 30 19:13:47 check_reload_status: Linkup starting re2 Mar 30 19:13:47 kernel: re2: link state changed to DOWN
-
SpankIt, are you using 2.1?
This has been fixed in 2.1.1. There is a long thread about it somewhere.Steve
-
Hi there!
I have an issue with APU1C and pfsense 2.1, the LiveCD is not booting. Here are more details:
Booting stops after "Starting the BTX loader" message.http://pcengines.info/forums/?page=post&id=2ADF62BD-420D-447D-A387-A9B047776417&fid=40251191-FF24-48A8-BB0E-995B04812ADE
Anyone has an idea what's wrong?
Bye
-
Marama,
I didn't try to install from CD, instead I'd recommend to install via USB-Stick. Choose the amd64 memstick-image and follow the instructions to write the image to the stick. Mount it with a running freebsd and append 'console="comconsole" to /boot/loader.conf (don't forget to mount the stick and change the filepath - this is not copy & paste):
$ echo 'console="comconsole"' >> /$USB_MOUNT_POINT/boot/loader.conf
Now you're good to go. Boot from the USB- Stick and pass the following line to the kernel during startup:
$ set kern.cam.boot_delay=10000
Now you can follow the install-scripts as described in various How-To's.
If you want to use an SD-Card, it's even simpler - just follow the instructions from the CF-Card install to the letter, but during first boot pass the
$ set kern.cam.boot_delay=10000
line to the kernel, and everything works as expected. Once booted it's best to add the "kern.cam.boot_delay" statement to /boot/loader.conf. Don't forget to remount the partition before editing, it's readonly per default.
Have fun!
-
Interrupt the boot loader.
type-
set kern.cam.boot_delay=10000
boot
After booting add kern.cam.boot_delay=1000 to /boot/loader.conf.local via edit or from console.sir could you please explain this further? how to do it etc..thanks