aqc111
aqc111 copied to clipboard
bb-qq
DS1812+ report
This is more of a situation report than an issue report, just thought I'd share my results, as this NAS isn't listed in the README.
I gave this a try on my DS1812+ (cedarview package) with the QNAP dongle. Direct connection from NAS to PC (Aquantia AQtion 10GbE).
I didn't pay enough attention when initially plugging the dongle in and picked a USB2 port. This limited the dongle to 1gbps link speed and actual throughput to a lot less (obviously).
Interestingly, when connecting to a USB3(.0) port, the link speed would get stuck at 100mbps. Reboots, replugs, tinkering, nothing seemed to fix it. The only way to get out of it was forcing the link speed on the adapter in the PC.
In the end, I couldn't get any better performance out of it than 1gbps PC -> NAS and 750mbps NAS -> PC. Not sure why this is asymmetrical. Spent some time playing with both Windows and Linux TCP/IP tuning, to no avail. This performance was the same with both iperf3 (also parallel) and SMB file transfer.
Additional notes:
- Changing Ethernet configuration in DSM, or changing card settings on the PC, would often require a complete replug and reboot for things to work again. But once it works it keeps working as long as you don't touch anything, across PC and/or NAS reboots. No noteworthy messages in dmesg.
- Dongle was tested between PC and laptop and worked at the expected throughput.
- Multiple cat5e and cat6 cables were tried, no cat7 at hand, same results.
- Forced 2.5gbps link speed was also tried, same results.
- Different USB3 ports were tried, same results.
- NAS CPU usage never came above 15%.
- No other USB devices plugged in, lsusb reported 5gbps link to dongle.
If you have any bright ideas on improving the situation, I'd be happy to hear them. For now I've returned to using both built-in 1gbps in parallel.
Update: after somewhat regular usage on a Windows laptop I've found there as well the QNAP likes to regress to 100mbps when connected to a USB3 port, unless the link speed is forced. So that is not a Synology specific issue.
Have you tried Jumbo-Frame? It may help to increase throughput for platforms with poor CPU.
From the pictures on this page, the USB 2.0 ports are soldered on the main PCB directly, whereas USB 3.0 ports seem to be implemented on a different board.
I suspect this is the same signal quality issue reported about other models caused by that front ports are soldered on the main PCB, whereas back ports are not.
Hi,
Similar status report about experiences and problems related to DS1812+. I successfully installed the driver manually. The adapter is connected to a Netgear 10Gb switch using CAT6A cable.
root@mediacenter:~# uname -a
Linux mediacenter 3.10.105 #25426 SMP Wed Jul 8 03:16:31 CST 2020 x86_64 GNU/Linux synology_cedarview_1812+
root@mediacenter:~# lsusb
|__usb1 1d6b:0002:0310 09 2.00 480MBit/s 0mA 1IF (ehci_hcd 0000:00:1a.7) hub
|__usb2 1d6b:0002:0310 09 2.00 480MBit/s 0mA 1IF (ehci_hcd 0000:00:1d.7) hub
|__2-1 f400:f400:0100 00 2.00 480MBit/s 98mA 1IF (Synology Diskstation 46f375d5eeb70e)
|__usb3 1d6b:0001:0310 09 1.10 12MBit/s 0mA 1IF (uhci_hcd 0000:00:1a.0) hub
|__usb4 1d6b:0001:0310 09 1.10 12MBit/s 0mA 1IF (uhci_hcd 0000:00:1d.0) hub
|__usb5 1d6b:0001:0310 09 1.10 12MBit/s 0mA 1IF (uhci_hcd 0000:00:1d.1) hub
|__usb6 1d6b:0001:0310 09 1.10 12MBit/s 0mA 1IF (uhci_hcd 0000:00:1d.2) hub
|__usb7 1d6b:0002:0310 09 2.00 480MBit/s 0mA 1IF (xhci_hcd 0000:05:00.0) hub
|__usb8 1d6b:0003:0310 09 3.00 5000MBit/s 0mA 1IF (xhci_hcd 0000:05:00.0) hub
|__8-1 1c04:0015:0101 00 3.20 5000MBit/s 896mA 1IF (QNAP QNAP QNA-UC5G1T USB to 5GbE Adapter 04I22577)
root@mediacenter:~# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: sit0: <NOARP> mtu 1480 qdisc noop state DOWN
link/sit 0.0.0.0 brd 0.0.0.0
3: eth0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN qlen 1000
link/ether 00:11:32:11:12:63 brd ff:ff:ff:ff:ff:ff
inet 169.254.164.131/16 brd 169.254.255.255 scope global eth0
valid_lft forever preferred_lft forever
4: eth1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN qlen 1000
link/ether 00:11:32:11:12:64 brd ff:ff:ff:ff:ff:ff
inet 169.254.123.36/16 brd 169.254.255.255 scope global eth1
valid_lft forever preferred_lft forever
5: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether 24:5e:be:4f:7e:90 brd ff:ff:ff:ff:ff:ff
inet 192.168.0.50/16 brd 192.168.255.255 scope global eth2
valid_lft forever preferred_lft forever
inet6 fe80::265e:beff:fe4f:7e90/64 scope link
valid_lft forever preferred_lft forever
root@mediacenter:~# ethtool eth2
Settings for eth2:
Supported ports: [ TP MII ]
Supported link modes: 100baseT/Full
1000baseT/Full
Supported pause frame use: No
Supports auto-negotiation: Yes
Advertised link modes: 100baseT/Full
1000baseT/Full
Advertised pause frame use: No
Advertised auto-negotiation: Yes
Speed: 5000Mb/s
Duplex: Full
Port: MII
PHYAD: 0
Transceiver: internal
Auto-negotiation: off
Supports Wake-on: g
Wake-on: d
Current message level: 0x00000007 (7)
drv probe link
Link detected: yes
My issue is that the driver is working in 1Gb mode and I do not see how to switch it to 5GB mode. Any guidance would be appreciated. Thanks,
Peter
Some further info:
root@mediacenter:~# iperf3 -s
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.0.100, port 51048
[ 5] local 192.168.0.50 port 5201 connected to 192.168.0.100 port 51049
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 109 MBytes 910 Mbits/sec
[ 5] 1.00-2.00 sec 68.7 MBytes 576 Mbits/sec
[ 5] 2.00-3.00 sec 97.8 MBytes 820 Mbits/sec
[ 5] 3.00-4.00 sec 120 MBytes 1.01 Gbits/sec
[ 5] 4.00-5.00 sec 85.9 MBytes 721 Mbits/sec
[ 5] 5.00-6.00 sec 117 MBytes 978 Mbits/sec
[ 5] 6.00-7.00 sec 121 MBytes 1.01 Gbits/sec
[ 5] 7.00-8.00 sec 120 MBytes 1.00 Gbits/sec
[ 5] 8.00-9.00 sec 87.8 MBytes 736 Mbits/sec
[ 5] 9.00-10.00 sec 92.8 MBytes 779 Mbits/sec
[ 5] 10.00-10.03 sec 3.31 MBytes 1.01 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.03 sec 1022 MBytes 855 Mbits/sec receiver
If I recall correctly, I for one was unable to force the link speed from the Linux end. I had to do on the other side (Windows box) but that was a direct connection. You can't do that from your switch unless it's a managed one, then you might.
I would rephrase my report after further investigations. Facts:
- the driver was installed and is working correctly
- the QNAP adapter shows 5G connection, Netgear switch shows the same
- the real-life speed of DS1812+ is about gigabit (not 5G)
- firmware update of the adapter (to FW 3.1.6) did not change anything
- use of 9K jumbo frame increased the speed 10-15% (from ~115MB/s to ~130MB/s)
My guesses about potential issue are the following:
- maybe DS1812+ hardware is a limiting factor (Intel Atom D2700 CPU or USB 3.0 interface)?
- maybe Linux kernel 3.10 TCP settings are limiting?
- maybe there is an issue in the driver?
- maybe my QNAP adapter is a faulty one?
- maybe Netgear XS508M switch has a compatibility issue with this QNAP adapter?
Any thoughts?
Currently, I am suspecting the cause of instability/performance issues is the driver of the USB host controller comes from the old kernel because this kind of issue is reported with older platforms comparatively.
And I am wondering that this issue might be resolved by changing some parameters of the kernel, but I could not find about it yet. Also if the next major update of DSM has a newer kernel for old platforms, this issue might be resolved automatically.
I would also like to confirm all of the results above on my 1812+ with both the Qnap and Sabrent 5Gb adapters. I did iperf3 tests from desktop(10GB) through Qnap QSW-M408-4C 10GbE Managed Switch to 1812+. Switch shows desktop at 10gb and aquantia chips at 5gb, but iperf maxes ~800Mbps, I also did jumbo frames 9k on both ends, etc.
The only thing else I might add was that when running iperf w/ "-w 2M" iperf failed w/ inability to set socket buffer size above 416kb. The initial searching I did indicated very conservative values in linux kernel or just old iperf error handling the window size: https://github.com/esnet/iperf/issues/757 links to => https://fasterdata.es.net/host-tuning/linux/
This indicates some TCP tuning (might help ???) Or it could be totally unrelated as I would need to dig further. Im surprised how well my DS1812+ has held up over 8 years (80 in technology years) but even 2x link aggregated 1GB adapaters just is not cutting it in 2020, Im ready to throw down for an 1820/1821 with single/dual 10Gbe if they ever get around to selling it, so this was my next best option, hope some of the info here might help get this working.
Im available to test any possible fixes. Thx
I moved on to an 1819+ with dual 10GbE card. It's a good upgrade to an 8 year old box.
Cross posting that it does not appear to be the nic at fault, I installed the 2.5GB RTL8156 to same results, essentially less than GB speed :'(