linuxptp-users

[Linuxptp-users] tx poll timeout

[Ledda W. E. <Wil...@it...>]

Hello,
Sometimes I see a "tx poll timeout" sending a delay requests. After about 20 seconds and then the synchronization restarts. What could be the root cause? It could be a problem of the driver or a problem of the network? What else?
...
...: master offset          5 s2 freq  +34434 path delay       278
...: poll tx timestamp timeout
...: port 1: send delay request failed
...: port 1: SLAVE to FAULTY on FAULT_DETECTED (FT_UNSPECIFIED)
...: driver changed our HWTSTAMP options
...: tx_type   1 not 1
...: rx_filter 1 not 12
...: port 1: FAULTY to LISTENING on FAULT_CLEARED
...: port 1: new foreign master 008063.fffe.db5e00-4
...: selected best master clock 00a069.fffe.01c094
...: port 1: LISTENING to UNCALIBRATED on RS_SLAVE
...: port 1: minimum delay request interval 2^3
...: master offset       -250 s2 freq  +34188 path delay       289
...: port 1: UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED
...: master offset         12 s2 freq  +34370 path delay       289
...: master offset         27 s2 freq  +34388 path delay       289
...

I'm using ptp4l version 1.3 with an Intel i350 ethernet interface with Red Hat RHEL 6.5.

$ uname -a
Linux xxxx 2.6.32-431.20.3.el6.x86_64 #1 SMP Fri Jun 6 18:30:54 EDT 2014 x86_64 x86_64 x86_64 GNU/Linux

$ ethtool -i eth1
driver: igb
version: 5.0.5-k
firmware-version: 1.61, 0x8000090f
bus-info: 0000:24:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: no

Thank you

William

Re: [Linuxptp-users] tx poll timeout

[Richard C. <ric...@gm...>]

On Wed, Jan 21, 2015 at 09:06:39AM +0000, Ledda William EXT wrote:
> Hello,
> Sometimes I see a "tx poll timeout" sending a delay requests. After about 20 seconds and then the synchronization restarts. What could be the root cause? It could be a problem of the driver or a problem of the network? What else?

This can be caused by:

1. link down
2. driver/stack latency delivering the time stamp
3. driver/hw dropping the time stamp

Cases 1 and 3 will always trigger the fault. If this happens too
often, you can reduce the fault duration with the configuration option
called "fault_reset_interval".

Case 2 can be mitigated by increasing the "tx_timestamp_timeout"
option. The default is 1 millisecond, and so you could try 2
milliseconds.

HTH
Richard

Re: [Linuxptp-users] tx poll timeout

[Ledda W. E. <Wil...@it...>]

Thank you Richard, 
I'll investigate further in this sense tuning the parameters.

William


William LEDDA
External Contractor
CODAC Section

ITER Organization, Building 72/4115, CHD, Control System Division
Route de Vinon-sur-Verdon - CS 90 046 - 13067 St Paul Lez Durance Cedex - France
Phone: +33 4 42 17 85 94
Get the latest ITER news on http://www.iter.org/whatsnew
-----Original Message-----
From: Richard Cochran [mailto:ric...@gm...] 
Sent: 21 January 2015 11:31
To: Ledda William EXT
Cc: lin...@li...
Subject: Re: [Linuxptp-users] tx poll timeout

On Wed, Jan 21, 2015 at 09:06:39AM +0000, Ledda William EXT wrote:
> Hello,
> Sometimes I see a "tx poll timeout" sending a delay requests. After about 20 seconds and then the synchronization restarts. What could be the root cause? It could be a problem of the driver or a problem of the network? What else?

This can be caused by:

1. link down
2. driver/stack latency delivering the time stamp 3. driver/hw dropping the time stamp

Cases 1 and 3 will always trigger the fault. If this happens too often, you can reduce the fault duration with the configuration option called "fault_reset_interval".

Case 2 can be mitigated by increasing the "tx_timestamp_timeout"
option. The default is 1 millisecond, and so you could try 2 milliseconds.

HTH
Richard