I have two CentOS 6.5 systems running ptp 2.3.1. The slave system does not update the system time, despite apparently syncing with the master and acquiring clock control. I installed ptpd by building RPMs using the rpmbuild.sh script, and changed the following options from their defaults in the config file:
How do you know it's not updating the system time? When you run the slave with -V, are you getting statistics? What is the status file showing - /var/run/ptpd2.status?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Host info : artemis, PID 5399
Local time : Tue Jul 07 12:47:43 CDT 2015
Kernel time : Tue Jul 07 17:47:43 GMT 2015
Interface : eth2
Preset : slaveonly
Transport : ipv4 + libpcap, multicast
Delay mechanism : E2E
Sync mode : TWO_STEP
PTP domain : 0
Port state : PTP_SLAVE
Local port ID : 002590fffe6a2838(unknown)/1
Best master ID : 002590fffe6a2750(unknown)/1
Best master IP : 172.16.0.2
GM priority : Priority1 128, Priority2 128, clockClass 13
Time properties : ARB timescale, tracbl: time N, freq N, src: INTERNAL_OSCILLATOR(0xa0)
UTC properties : UTC valid: N, UTC offset: 0
Offset from Master : 0.000000000 s, mean 0.000000000 s, dev 0.000000000 s
Mean Path Delay : 0.000000000 s, mean 0.000000000 s, dev 0.000000000 s
Clock status : calibrated, in control
Clock correction : 0.000 ppm, mean 0.000 ppm, dev 0.000 ppm
Message rates : 1/s sync, 1/s delay, 1/2s announce
TimingService : current PTP0, best PTP0, pref PTP0
TimingServices : total 1, avail 1, oper 1, idle 0, in_ctrl 1
Performance : Message RX 3/s, TX 0/s
Announce received : 42
Sync received : 83
Follow-up received : 0
DelayReq sent : 75
DelayResp received : 75
State transitions : 3
PTP Engine resets : 1
Thanks,
Jeff
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Slave is showing zero follow-up messages received. It's receiving sync but not follow-up, without this it cannot compute the offset. What is the master showing, is it sending followUp? Is firewall disabled on both? What about tcpdump/tshark, how does the traffic look like? What's the master config - can you enable libpcap and see if it makes a difference? Finally, what's the NIC type and driver and what does "ethtool -T <ethX>" show?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I ran WireShark on an external system, and I see the PTPv2 follow-up messages going across the wire. Source is the master port 320, destination is 224.0.1.129 port 320.
The NIC is an Intel Pro/1000, using the e1000e driver.
Thanks for you help, it's working now! I had disabled the firewall on the master, but apparently not on the slave. Completely disabling iptables and turning on pcap in ptpd2 seems to have been the magic combination.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have two CentOS 6.5 systems running ptp 2.3.1. The slave system does not update the system time, despite apparently syncing with the master and acquiring clock control. I installed ptpd by building RPMs using the rpmbuild.sh script, and changed the following options from their defaults in the config file:
Below is the log file from the slave:
Thanks,
Jeff
Hi,
How do you know it's not updating the system time? When you run the slave with -V, are you getting statistics? What is the status file showing - /var/run/ptpd2.status?
Hi,
The two machines are about 3 seconds apart, so I can see the clock offset easily. Neither is running ntpd.
The statistics I'm getting are apparently all zeros:
Below are the contents of /var/run/ptpd2.status:
Thanks,
Jeff
Slave is showing zero follow-up messages received. It's receiving sync but not follow-up, without this it cannot compute the offset. What is the master showing, is it sending followUp? Is firewall disabled on both? What about tcpdump/tshark, how does the traffic look like? What's the master config - can you enable libpcap and see if it makes a difference? Finally, what's the NIC type and driver and what does "ethtool -T <ethX>" show?
I ran WireShark on an external system, and I see the PTPv2 follow-up messages going across the wire. Source is the master port 320, destination is 224.0.1.129 port 320.
The NIC is an Intel Pro/1000, using the e1000e driver.
ethtool -T output is:
Thanks for you help!
Jeff
The remaining questions are:
Please send ptpd a SIGUSR2 and check the logs - it should dump all counters, check for errors.
Thanks,
Wojciech
Thanks for you help, it's working now! I had disabled the firewall on the master, but apparently not on the slave. Completely disabling iptables and turning on pcap in ptpd2 seems to have been the magic combination.
Hi - good to hear it worked - however it should still work without libpcap. Can you confirm?