[etherlab-dev] Subject: Re: Bus scan takes long time at low rates

Thu May 2 10:41:08 CEST 2013

Hi,

 Martin wrote that the internal thread is handling scan in the idle phase,
and the application is driving this in the cyclic phase.

That explains why the scan time increases when the application is in
control.

The application in the example runs at 10Hz i.e. call the cyclic functions
10 times pr. second.

Having scan time at 3 minutes shows that is requires 3*60*10 = 1800 calls
to perform a bus scan.

Is it any way of reducing 1800 calls downto a smaller number?

Best regards
Steffen Dalgard
SINTEF ICT
Norway

-----Original Message-----
From: Martin Troxler [mailto:martin.troxler at komaxgroup.com]
Sent: 30. april 2013 12:48
To: steffen.sint at gmail.com
Cc: kunz at sim.tu-darmstadt.de; etherlab-dev at etherlab.org
Subject: Bus scan takes long time at low rates

Hi Steffen

In the idle phase, the transmitting of the frame is done by a internal
thread of the master which runs as fast as the bus allows.

The transmitting of the frame in cyclic phase is done within the realtime
thread (i.e. within the calls to ecrt_master_send/receive) to ensure proper
cyclic transmission of process data.

Why do you need to do a rescan in cyclic mode? You should leave cyclic mode
if the number of slaves on the changes. The master will then scan the bus
automatically.

Regards

Martin

Note:

This e-mail is for the named person's use only. It may contain confidential
and/or privileged information. If you have received this e-mail in error,
please notify the sender immediately and delete the material from any
system. Any unauthorized copying, disclosure, distribution or other use of
this information by persons or entities other than the intended recipient
is prohibited.

Thank You.
*From:* Jürgen Kunz [mailto:kunz at sim.tu-darmstadt.de]
*Sent:* 30. april 2013 12:01
*To:* Steffen Dalgard
*Cc:* etherlab-dev at etherlab.org
*Subject:* Re: [etherlab-dev] Bus scan takes long time at low rates

Hi Steffen,

what host-OS do you use? I tried the same with Linux-VM (with RT-Preempt
kernel) and Windows-host, got the same result. You may try following:
* Use a RT-Preempt kernel in the VM
* Use ethercat-capable network-drivers in the VM (e1000 or maybe e1000e
should work with Oracle VM Virtualbox)
* Use a Linux Host with RT-Preempt kernel and set the VM to realtime
priority (98)
I've never tried this before, but this might reduce issues.

Regards
Jürgen

 Am 30.04.2013 10:22, schrieb Steffen Dalgard:

Hi,

I am setting up a environment for development an test.

In order to make it easy to replicate at different sites it is running in a
virtual machine inside Oracle VM VirtualBox Manager.

The EtherCAT slaves are connected to a dedicated bridged interface using
the generic Ethernet driver.

As you understand this is not a high performance system.

Before activating the master it takes below 1 sec for scanning the bus with
5 slaves.

When starting the userspace application and activating the master for
cyclic operation the rescan take very long time. It seems that the rescan
gets slower when cycle time are long. In my application I am using 10HZ
cycle time.

In the attached logs the time for scanning is about 3 minutes.

During the rescan period the processor load increase significantly (* 20)
but the packets on the bus stays constant at 10 packets pr. sec.

Anyone having experience how to reduce time for scanning?

I can reduce the cycle time downto 100HZ but the scan time is still long.

A downside is that I get more timeout / unmatched packets.

Anyone having ideas what causing packet loss?

The virtual machine is running

Linux 10.04 LTS

gcc (Ubuntu 4.4.3-4ubuntu5.1) 4.4.3 is used for compilation

I have attached three logs showing syslog output

More details and examples can be added if needed.

Any comments welcome.

Best regards
Steffen Dalgard
SINTEF ICT
Norway

"SYSLOG messages when starting up Master"

Apr 30 09:29:09 ecmaster-vm kernel: [86526.177110] EtherCAT: Master driver
1.5.1 41dc9a4a0f76
Apr 30 09:29:09 ecmaster-vm kernel: [86526.178626] EtherCAT: 1 master
waiting for devices.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.184362] ec_generic: EtherCAT
master generic Ethernet device module 1.5.1 41dc9a4a0f76
Apr 30 09:29:09 ecmaster-vm kernel: [86526.184375] EtherCAT: Accepting
08:00:27:BF:57:AC as main device for master 0.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.184391] EtherCAT 0: Starting
EtherCAT-IDLE thread.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.184423] EtherCAT 0: Link state
of ecm0 changed to UP.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.188311] EtherCAT 0: 5 slave(s)
responding on main device.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.188315] EtherCAT 0: Slave states
on main device: PREOP.
Apr 30 09:29:09 ecmaster-vm kernel: [86526.193124] EtherCAT 0: Scanning bus.
Apr 30 09:29:10 ecmaster-vm kernel: [86526.774874] EtherCAT 0: Bus scanning
completed in 580 ms.
Apr 30 09:29:10 ecmaster-vm kernel: [86526.774879] EtherCAT 0: Using slave
0 as DC reference clock.

Apr 30 09:31:20 ecmaster-vm kernel: [86656.548051] EtherCAT WARNING 0: 1
datagram TIMED OUT!
Apr 30 09:35:17 ecmaster-vm kernel: [86894.412791] EtherCAT WARNING 0: 1
datagram TIMED OUT!
Apr 30 09:35:17 ecmaster-vm kernel: [86894.412811] EtherCAT WARNING 0: 1
datagram UNMATCHED!
Apr 30 09:36:04 ecmaster-vm kernel: [86940.780824] EtherCAT WARNING 0: 2
datagrams TIMED OUT!
Apr 30 09:36:04 ecmaster-vm kernel: [86940.780842] EtherCAT WARNING 0: 2
datagrams UNMATCHED!

"SYSLOG messages when starting application"

Apr 30 09:38:49 ecmaster-vm kernel: [87106.284019] EtherCAT: Requesting
master 0...
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284231] EtherCAT: Successfully
requested master 0.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284279] EtherCAT 0: Domain0:
Logical address 0x00000000, 2 byte, expected working counter 3.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284284] EtherCAT 0:   Datagram
domain0-0-main: Logical offset 0x00000000, 2 byte, type LRW.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284287] EtherCAT 0: Domain1:
Logical address 0x00000002, 0 byte, expected working counter 0.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284668] EtherCAT 0: Master
thread exited.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.284671] EtherCAT 0: Starting
EtherCAT-OP thread.
Apr 30 09:38:49 ecmaster-vm kernel: [87106.285073] EtherCAT WARNING 0: 1
datagram UNMATCHED!
Apr 30 09:38:56 ecmaster-vm kernel: [87113.384968] EtherCAT 0: Domain 0:
Working counter changed to 2/3 (2+0).
Apr 30 09:38:59 ecmaster-vm kernel: [87116.386563] EtherCAT 0: Domain 0:
Working counter changed to 3/3 (3+0).
Apr 30 09:39:00 ecmaster-vm kernel: [87117.288055] EtherCAT 0: Slave states
on main device: PREOP, OP.

"SYSLOG messages when forcing bus rescan by extracting and inserting one of
the slaves"

Apr 30 09:41:25 ecmaster-vm kernel: [87261.585150] EtherCAT 0: Domain 0:
Working counter changed to 2/3 (2+0).
Apr 30 09:41:25 ecmaster-vm kernel: [87261.585644] EtherCAT 0: 4 slave(s)
responding on main device.
Apr 30 09:41:25 ecmaster-vm kernel: [87262.085167] EtherCAT 0: Scanning bus.
Apr 30 09:44:02 ecmaster-vm kernel: [87418.885938] EtherCAT 0: Bus scanning
completed in 157196 ms.
Apr 30 09:44:02 ecmaster-vm kernel: [87418.885953] EtherCAT 0: Using slave
0 as DC reference clock.
Apr 30 09:44:04 ecmaster-vm kernel: [87420.685695] EtherCAT 0: 5 slave(s)
responding on main device.
Apr 30 09:44:04 ecmaster-vm kernel: [87420.685710] EtherCAT 0: Slave states
on main device: INIT, PREOP, OP.
Apr 30 09:44:04 ecmaster-vm kernel: [87421.185114] EtherCAT 0: Scanning bus.
Apr 30 09:47:25 ecmaster-vm kernel: [87621.985808] EtherCAT 0: Bus scanning
completed in 201196 ms.
Apr 30 09:47:25 ecmaster-vm kernel: [87621.985825] EtherCAT 0: Using slave
0 as DC reference clock.
Apr 30 09:47:30 ecmaster-vm kernel: [87627.184968] EtherCAT 0: Domain 0:
Working counter changed to 3/3 (3+0).
Apr 30 09:47:31 ecmaster-vm kernel: [87628.088202] EtherCAT 0: Slave states
on main device: PREOP, OP.

 _______________________________________________

etherlab-dev mailing list

etherlab-dev at etherlab.org

http://lists.etherlab.org/mailman/listinfo/etherlab-dev

-- 
Dipl.-Inform. Jürgen Kunz

Technische Universität Darmstadt <http://www.tu-darmstadt.de/>
FG Simulation, Systemoptimierung und Robotik<http://www.sim.tu-darmstadt.de/>
Hochschulstr. 10
64289 Darmstadt

Tel.: ++49 (0) 6151-16-70383
Fax: ++49 (0) 6151-16-6648
E-Mail: kunz(at)sim.tu-darmstadt.de
Homepage: http://www.sim.tu-darmstadt.de
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-dev/attachments/20130502/709ad9e1/attachment-0001.htm>