linuxptp-users

[Linuxptp-users] use of linuxptp as a g.8275.2 Telecom slave?

[Dan G. <da...@po...>]

Greetings, all. I've been using ptp for a couple years now, almost exclusively with hardware-based appliances doing G.8275.2 v4/v6 unicast in a telecom environment. I'd like to be able to extend ptp to some of our chipset-capable enterprise NTP infrastructure but can't find a good reference for how to do so. Based on some of the historic threads, I know the mode is supposed to be supported, but all the OS distribution howtos focus on multicast discovery/delivery and very little on unicast or end2end.

Has anyone successfully gotten a unicast PTP slave (v4 or v6) running in a deployed environment and would you mind sharing some configuration snippets?

Thanks.
Dan 

-- 
Dan Geist

Re: [Linuxptp-users] use of linuxptp as a g.8275.2 Telecom slave?

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

On Wed, Jan 02, 2019 at 03:20:58PM +0000, Dan Geist wrote:
> Greetings, all. I've been using ptp for a couple years now, almost
> exclusively with hardware-based appliances doing G.8275.2 v4/v6
> unicast in a telecom environment. I'd like to be able to extend ptp
> to some of our chipset-capable enterprise NTP infrastructure but
> can't find a good reference for how to do so. Based on some of the
> historic threads, I know the mode is supposed to be supported, but
> all the OS distribution howtos focus on multicast discovery/delivery
> and very little on unicast or end2end.

G.8275.2 is fully supported as of v2.0.  See the example configuration
file in configs/G.8275.2.cfg in the source tree.

Cheers,
Richard

Re: [Linuxptp-users] use of linuxptp as a g.8275.2 Telecom slave?

[Dan G. <da...@po...>]

Thanks Richard. I THINK I have things up and running correctly, but am still not sure why syncing system clock (and status utilities) aren't working. Here's my relevant config:

----------------------------------------
# 
# default stuff up here
#

#
# G.8275.2 Options
#
dataset_comparison              G.8275.x
G.8275.defaultDS.localPriority  128
masterOnly                      0
G.8275.portDS.localPriority     128
domainNumber                    44
hybrid_e2e                      1
inhibit_multicast_service       1
unicast_listen                  1
unicast_req_duration            60
summary_interval                4

[unicast_master_table]
table_id                       1
logQueryInterval               2
UDPv4                          XX.191.49.132
UDPv4                          XX.191.49.133

[em2.98]
unicast_master_table           1
--------------------------------------

The two indicated sources are GPS-based grandmasters and are reporting equal delay_resp and delay_req message rates. On the linuxptp host, ptp4l seems to be working but the phc2sys is waiting indefinitely:

root@psp6esxd04:/etc# systemctl restart ptp4l && tail -f /var/log/syslog | grep ptp
Jan  3 11:56:46 psp6esxd04 ptp4l: [6006.750] selected /dev/ptp1 as PTP clock
Jan  3 11:56:46 psp6esxd04 ptp4l: [6006.750] port 0: hybrid_e2e only works with E2E
Jan  3 11:56:46 psp6esxd04 ptp4l: [6006.751] port 1: INITIALIZING to LISTENING on INIT_COMPLETE
Jan  3 11:56:46 psp6esxd04 ptp4l: [6006.751] port 0: INITIALIZING to LISTENING on INIT_COMPLETE
Jan  3 11:56:46 psp6esxd04 ptp4l: [6007.077] port 1: new foreign master 0080ea.fffe.8c2430-1
Jan  3 11:56:47 psp6esxd04 phc2sys: [6007.753] Waiting for ptp4l...
Jan  3 11:56:48 psp6esxd04 ptp4l: [6008.144] port 1: new foreign master 0080ea.fffe.8c2100-1
Jan  3 11:56:48 psp6esxd04 phc2sys: [6008.754] Waiting for ptp4l...
Jan  3 11:56:49 psp6esxd04 phc2sys: [6009.755] Waiting for ptp4l...
Jan  3 11:56:50 psp6esxd04 phc2sys: [6010.756] Waiting for ptp4l...
Jan  3 11:56:50 psp6esxd04 ptp4l: [6011.076] selected best master clock 0080ea.fffe.8c2430
Jan  3 11:56:50 psp6esxd04 ptp4l: [6011.077] updating UTC offset to 37
Jan  3 11:56:50 psp6esxd04 ptp4l: [6011.077] port 1: LISTENING to UNCALIBRATED on RS_SLAVE
Jan  3 11:56:51 psp6esxd04 phc2sys: [6011.757] Waiting for ptp4l...
Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] updating UTC offset to 37
Jan  3 11:56:52 psp6esxd04 phc2sys: [6012.759] Waiting for ptp4l...
Jan  3 11:56:53 psp6esxd04 phc2sys: [6013.760] Waiting for ptp4l...
Jan  3 11:56:54 psp6esxd04 ptp4l: [6014.399] port 1: UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED
Jan  3 11:56:54 psp6esxd04 phc2sys: [6014.761] Waiting for ptp4l...
Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.077] updating UTC offset to 37
Jan  3 11:56:55 psp6esxd04 phc2sys: [6015.762] Waiting for ptp4l...
Jan  3 11:56:56 psp6esxd04 phc2sys: [6016.763] Waiting for ptp4l...
Jan  3 11:56:57 psp6esxd04 phc2sys: [6017.764] Waiting for ptp4l...
Jan  3 11:56:58 psp6esxd04 phc2sys: [6018.766] Waiting for ptp4l...
Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.077] updating UTC offset to 37
Jan  3 11:56:59 psp6esxd04 phc2sys: [6019.767] Waiting for ptp4l...
Jan  3 11:57:00 psp6esxd04 phc2sys: [6020.768] Waiting for ptp4l...
Jan  3 11:57:01 psp6esxd04 phc2sys: [6021.769] Waiting for ptp4l...
Jan  3 11:57:02 psp6esxd04 phc2sys: [6022.770] Waiting for ptp4l...
Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.077] updating UTC offset to 37


and pmc just seems to do nothing:

root@psp6esxd04:/etc# /usr/local/sbin/pmc -u -b 0 'GET TIME_STATUS_NP'
sending: GET TIME_STATUS_NP



Hopefully, I'm just missing something obvious due to unfamiliarity. Any thoughts here?


Thanks
Dan





----- On Jan 3, 2019, at 1:42 AM, Richard Cochran ric...@gm... wrote:

> 
> G.8275.2 is fully supported as of v2.0.  See the example configuration
> file in configs/G.8275.2.cfg in the source tree.
> 
> Cheers,
> Richard

-- 
Dan Geist

Re: [Linuxptp-users] use of linuxptp as a g.8275.2 Telecom slave?

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

On Thu, Jan 03, 2019 at 04:58:35PM +0000, Dan Geist wrote:
> summary_interval                4

Leave this at zero for debugging, so you can see the synchronization messages.

Looking at the ptp4l logs:

Jan  3 11:56:50 psp6esxd04 ptp4l: [6011.077] port 1: LISTENING to UNCALIBRATED on RS_SLAVE
Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] updating UTC offset to 37
Jan  3 11:56:54 psp6esxd04 ptp4l: [6014.399] port 1: UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED
Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.077] updating UTC offset to 37
Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.077] updating UTC offset to 37
Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.076] selected best master clock 0080ea.fffe.8c2100
Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.077] updating UTC offset to 37

Something is causing the BMC to trigger every four seconds.  That
seems wrong.  Maybe you have a logAnnounceInterval mismatch?

Also, what is the phc2sys command line?

Thanks,
Richard

Re: [Linuxptp-users] use of linuxptp as a g.8275.2 Telecom slave?

[Dan G. <da...@po...>]

----- On Jan 3, 2019, at 12:24 PM, Richard Cochran ric...@gm... wrote:

> On Thu, Jan 03, 2019 at 04:58:35PM +0000, Dan Geist wrote:
>> summary_interval                4
> 
> Leave this at zero for debugging, so you can see the synchronization messages.

Done. Lots of verbose output now.
 
> Looking at the ptp4l logs:
> 
> Jan  3 11:56:50 psp6esxd04 ptp4l: [6011.077] port 1: LISTENING to UNCALIBRATED
> on RS_SLAVE
> Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] selected best master clock
> 0080ea.fffe.8c2100
> Jan  3 11:56:52 psp6esxd04 ptp4l: [6012.144] updating UTC offset to 37
> Jan  3 11:56:54 psp6esxd04 ptp4l: [6014.399] port 1: UNCALIBRATED to SLAVE on
> MASTER_CLOCK_SELECTED
> Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.076] selected best master clock
> 0080ea.fffe.8c2100
> Jan  3 11:56:54 psp6esxd04 ptp4l: [6015.077] updating UTC offset to 37
> Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.076] selected best master clock
> 0080ea.fffe.8c2100
> Jan  3 11:56:58 psp6esxd04 ptp4l: [6019.077] updating UTC offset to 37
> Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.076] selected best master clock
> 0080ea.fffe.8c2100
> Jan  3 11:57:02 psp6esxd04 ptp4l: [6023.077] updating UTC offset to 37
> 
> Something is causing the BMC to trigger every four seconds.  That
> seems wrong.  Maybe you have a logAnnounceInterval mismatch?

Perhaps... but I have it set to once a second, which matches my GM:

PTP Domain Number	: 	44
Clock Class	: 	6
Master Clock Type	: 	One Step
Clock Class Profile	: 	G.8275.2
Master Delay Mechanism	: 	End To End
Max Slaves Supported	: 	100
Max Static Slaves Supported	: 	0
Max Announce Message Rate	: 	1 Packet per 1 Second
Max Sync Message Rate	: 	32 Packets per Second
Max Delay Response Message Rate	: 	32 Packets per Second
Max Lease Duration	: 	300 sec
Announce Extension TLV	: 	N/A

The BMC messages do settle down after a while. I commented out one of the GMs and it seems to only iterate on BMC 2 or 3 times, now.

> Also, what is the phc2sys command line?

It's being run via systemd, but:
# ps -Aef | grep phc2sys
root      5041     1  0 16:27 ?        00:00:00 /usr/local/sbin/phc2sys -w -s em2.98

and for reference:
# ethtool -T em2.98
Time stamping parameters for em2.98:
Capabilities:
        hardware-transmit     (SOF_TIMESTAMPING_TX_HARDWARE)
        software-transmit     (SOF_TIMESTAMPING_TX_SOFTWARE)
        hardware-receive      (SOF_TIMESTAMPING_RX_HARDWARE)
        software-receive      (SOF_TIMESTAMPING_RX_SOFTWARE)
        software-system-clock (SOF_TIMESTAMPING_SOFTWARE)
        hardware-raw-clock    (SOF_TIMESTAMPING_RAW_HARDWARE)
PTP Hardware Clock: 1
Hardware Transmit Timestamp Modes:
        off                   (HWTSTAMP_TX_OFF)
        on                    (HWTSTAMP_TX_ON)
Hardware Receive Filter Modes:
        none                  (HWTSTAMP_FILTER_NONE)
        ptpv1-l4-event        (HWTSTAMP_FILTER_PTP_V1_L4_EVENT)
        ptpv2-l4-event        (HWTSTAMP_FILTER_PTP_V2_L4_EVENT)
        ptpv2-l2-event        (HWTSTAMP_FILTER_PTP_V2_L2_EVENT)

Thanks
Dan

-- 
Dan Geist