Craig,
Thank you for digging deep into the issue. Interesting info about JMC controller. We had no time to debug this issue on our own, so your information is very valuable.
I wonder whether USB-to-1GbE will work well for you. It's not only a limitation of throughput, but also USB as higher latency. GSM has quite relxed requirements and USN alone works fine, but stacking of Ethernet and USB is something I'm suspicious about. Let us know your testing results. Worst case you'll have to get another Atom board from a more reliable manufacturer.
"what(): RuntimeError: no control response" error is because UHD sends a command to UmTRX, but doesn't get anything back. Seems like link is going down in the middle of communiction.
Transceiver starts fine without OpenBTS, because it basically does nothing. The real work starts when OpenBTS sends "POWER ON" command to the transceiver.
On Sat, Mar 16, 2013 at 6:45 PM, Craig Reading craig.reading1@gmail.com wrote:
All,
Thanks for you help I have sort of resolved this issue:
UmTRX now hanging off TP-Link TL-SG1005D Gigabit switch and Shuttle XS35-703 V2 running OpenBTS hanging off Netgear FS605 10/100Mbps switch.
Root cause: JMicron JMC250 connecting to a Gigabit switch and the link partner enabled the IEEE 802.3az Energy Efficient Ethernet feature
War and peace can be found here:
http://genesysguru.com/blog/blog/2013/03/16/umtrx-networking-issues/
Regards Craig
On 16 March 2013 12:33, Alexander Chemeris alexander.chemeris@gmail.com wrote:
We temporarily solved issues with this JMC chip by putting a switch between it and UmTRX, but not all models of switches worked and even after that stability was not perfect.
Looking forward to know more about your experience.
On Sat, Mar 16, 2013 at 1:39 AM, Craig Reading craig.reading1@gmail.com wrote:
Yes - shuttle with JMicron JMC250 (off top of my head). Depends on chip revision - complete pain!
I'll post my experiences - basically no joy on Pre revision 4 chips.
Regards Craig On 15 Mar 2013, at 21:24, Alexander Chemeris alexander.chemeris@gmail.com wrote:
Craig,
Which NIC is that? We had the same experience (no Gbit connection) on Shuttle computers. I don't have this information near my hands, but IIRC they had JMicron NICs. We never had issues with Intel motherboards.
Please excuse typos. Written with a touchscreen keyboard.
-- Regards, Alexander Chemeris CEO/Founder Fairwaves LLC http://fairwaves.ru
On Mar 15, 2013 11:26 PM, "Craig Reading" craig.reading1@gmail.com wrote:
Hi Alexander
I'll email the wireshark trace to you shortly. Yes I have been playing with network setup but can't remember the configuration when successful!
The openBTS server has a 1Gbit NIC but can't get it to negotiate anything other than 100Mbit even after updating drivers and much other 'playing' with cables hubs and switches!
Mr Amazon is bringing some new network components tomorrow which should help further my investigations.
This might well be one of these weird network issues you have seen that we might get to the bottom off.
Regards Craig
On 15 Mar 2013, at 19:04, Alexander Chemeris alexander.chemeris@gmail.com wrote:
Craig,
Could you please send a Wireshark capture of the start attempt? Send it to me directly if it's more then 100k.
This looks very much like a networking issue. Have you changed your network setup since it was working? May be connected it through a different switch? We found that some models of switches does not work well and introduce weird issues.
Is your network card 100Mbit or 1Gbit?
If it's an UmTRX issue indeed, we'll replace it. Our beta testers is our most precious asset.
Please excuse typos. Written with a touchscreen keyboard.
-- Regards, Alexander Chemeris CEO/Founder Fairwaves LLC http://fairwaves.ru
On Mar 15, 2013 10:07 PM, "Craig Reading" craig.reading1@gmail.com wrote:
Hi All,
After my initial success early in the week things seems to have gone pear shaped!
I cannot get OpenBTS to start reliably as per traces below. After a power cycle of the UmTRX I can normally get OpenBTS to start for a few seconds before the transceiver terminates again. ./transceiver starts and runs on its own OK but there is a lot less traffic in a wireshark trace compared to being started from OpenBTS.
When the transceiver starts from OpenBTS a Wireshark trace shows a lot of UDP packets.
Approx 460 x 36 byte UDP packets for the first 0.5 seconds Then some variable length UDP packets all ACKed back with 36 bytes After 0.8 from startup the UmTRX sends a 28 byte message:
14d0000700000005000000140000000000a6d5180000000000100100
After approx 1.0s from startup I then only get a few packets back from UmTRX After 5s I get nothing back from UmTRX After 6s I get an ICMP Port Unreachable
If I had more time I would look into the transceiver source code to figure out where things are stopping.
I suspect this might be a networking issue and have tried changing some Linux kernel buffer settings without success. I have also reset my OpenBTS database just in case I had an option in there causing this. I'll carry on checking out the network side of things but I am starting so suspect there might be an issue with the UmTRX.
Help!
Regards Craig
> Traces
ALERT 46937531359488 TRXManager.cpp:408:powerOn: POWERON failed with status -1 transceiver: no process killed linux; GNU C++ version 4.1.2 20080704 (Red Hat 4.1.2-52); Boost_104100; UHD_003. 004.000-93a49d0
terminate called after throwing an instance of 'uhd::runtime_error' what(): RuntimeError: no control response EMERG 1101519168 OpenBTS.cpp:134:startTransceiver: Transceiver quit with status 6. Exiting.
or:
ALERT 47613775102208 TRXManager.cpp:408:powerOn: POWERON failed with status -1 transceiver: no process killed linux; GNU C++ version 4.1.2 20080704 (Red Hat 4.1.2-52); Boost_104100; UHD_003. 004.000-93a49d0
ALERT 47832442212256 UHDDevice.cpp:469:open: UHD make failed, device type=umtrx, addr=192.168.1.10,name=UmTRX,serial=13 ALERT 47832442212256 runTransceiver.cpp:95:main: Transceiver exiting...
EMERG 1104513344 OpenBTS.cpp:134:startTransceiver: Transceiver quit with status 256. Exiting.
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru