Menu
▾
▴
linuxptp-users
|
From: Aman S. <ama...@gm...> - 2013-07-03 08:44:12
|
Hi All, When running two machines in Master and slave mode then if change the time in runtime (for both master and slave) ,then it will not update automatically to the corresponding slave.Can anyone please Help me in this. Thanks Aman |
|
From: Richard C. <ric...@gm...> - 2013-07-03 11:41:53
|
On Wed, Jul 03, 2013 at 02:14:05PM +0530, Aman Sharma wrote: > > When running two machines in Master and slave mode then if change the time > in runtime (for both master and slave) ,then it will not update > automatically to the corresponding slave.Can anyone please Help me in this. You have not told us what the problem is. Can you show us exactly what commands you are running? Thanks, Richard |
|
From: Richard C. <ric...@gm...> - 2013-07-03 18:39:53
|
On Wed, Jul 03, 2013 at 05:46:04PM +0530, Aman Sharma wrote:
>
> When we start the PTP on both machine, for first time slave time syncs to
> Master time.
> But during runtime if we are changing time(manually changing System time)
> of Master or Slave both are not syncing to each other.
The master will not synchronize to the slave!
Also, it does not make sense to change the time on the slave. That is
the job of the ptp4l program.
> We have to restart the PTP(i.e have to run command again) to make them sync
> to each other.
>
> Please let us know the command or process, so that Master and Slave can
> sync there time during run time
As Jacob wrote, there is an option for jumping the clock.
>From the man page:
pi_offset_const
The maximum offset the PI controller will correct by changing
the clock frequency instead of stepping the clock. When set to
0.0, the controller will never step the clock except on start.
The default is 0.0.
If you set this to, say 1.0, then the slave will jump the clock
whenever the offset exceeds one second.
HTH,
Richard
|
|
From: Aman S. <ama...@gm...> - 2013-07-04 05:16:31
|
Hi Richard, Can You please explain me Actually what this* pi_offset_const *does.I am not unable to get what you want to say. Regards Aman Sharma On Thu, Jul 4, 2013 at 12:09 AM, Richard Cochran <ric...@gm...>wrote: > On Wed, Jul 03, 2013 at 05:46:04PM +0530, Aman Sharma wrote: > > > > When we start the PTP on both machine, for first time slave time syncs to > > Master time. > > But during runtime if we are changing time(manually changing System time) > > of Master or Slave both are not syncing to each other. > > The master will not synchronize to the slave! > > Also, it does not make sense to change the time on the slave. That is > the job of the ptp4l program. > > > We have to restart the PTP(i.e have to run command again) to make them > sync > > to each other. > > > > Please let us know the command or process, so that Master and Slave can > > sync there time during run time > > As Jacob wrote, there is an option for jumping the clock. > From the man page: > > pi_offset_const > The maximum offset the PI controller will correct by changing > the clock frequency instead of stepping the clock. When set to > 0.0, the controller will never step the clock except on start. > The default is 0.0. > > If you set this to, say 1.0, then the slave will jump the clock > whenever the offset exceeds one second. > > HTH, > Richard > |
|
From: Aman S. <ama...@gm...> - 2013-07-12 14:04:16
|
Hi Richard, Thanks for the response. After increasing it to 3600, slave time is updated according to the master time.But the master machine is not sending Sync message and no follow_up is there.Both machines directly sending Peer delay message to each other using gPTP. Is there any reason for not sending Sync meassage to slave machine. -- Regards Aman On Fri, Jul 12, 2013 at 6:48 PM, Richard Cochran <ric...@gm...>wrote: > On Fri, Jul 12, 2013 at 05:06:35PM +0530, Aman Sharma wrote: > > > > when running linuxPTP with gPTP configuration between two linux machines, > > then they are just sending peer delay request to each other and other one > > also gives response for the same request.Means Master is not sending Sync > > message and no follow_up is there.And in log messages ,it shows : > > Probably you are using software time stamping. If so, you need to > increase neighborPropDelayThresh. Read the ptp4l.8 man page, and you > will find: > > neighborPropDelayThresh > Upper limit for peer delay in nanoseconds. If the estimated peer > delay is greater than this value the port is marked as not > 802.1AS capable. > > HTH, > Richard > |
|
From: Richard C. <ric...@gm...> - 2013-07-12 17:07:04
|
On Fri, Jul 12, 2013 at 07:34:10PM +0530, Aman Sharma wrote: > > After increasing it to 3600, slave time is updated according to the master > time.But the master machine is not sending Sync message and no follow_up is > there.Both machines directly sending Peer delay message to each other using > gPTP. > > Is there any reason for not sending Sync meassage to slave machine. Read this once again ... > > neighborPropDelayThresh > > Upper limit for peer delay in nanoseconds. If the estimated peer > > delay is greater than this value the port is marked as not > > 802.1AS capable. and notice the part about "not capable." Why don't you try a more reasonable value? Richard |
|
From: Richard C. <ric...@gm...> - 2013-07-04 05:26:32
Attachments:
signature.asc
|
On Thu, Jul 04, 2013 at 10:46:22AM +0530, Aman Sharma wrote: > Hi Richard, > > Can You please explain me Actually what this* pi_offset_const *does.I am > not unable to get what you want to say. It lets the slave jump clock again, after the first time. Try it out, like I already explained, and see what happens. HTH, Richard |
|
From: Aman S. <ama...@gm...> - 2013-07-04 09:30:05
|
Hi Richard, Thanks for your reply. We are facing one more issue, in one of our Laptop we are not able to run linux ptp but it got compiled correctly. Machine details: OS: 12.04 ubuntu Kermel version: 3.2.0-40-generic Ethernet controller: INTEL 82579LM with e1000e kernel driver Following are log messages: sudo ./ptp4l -S -i eth0 -m -2 ptp4l[327.331]: port 1: get_ts_info not supported ptp4l[327.464]: port 1: INITIALIZING to LISTENING on INITIALIZE ptp4l[327.464]: port 0: INITIALIZING to LISTENING on INITIALIZE ptp4l[333.464]: port 1: LISTENING to MASTER on ANNOUNCE_RECEIPT_TIMEOUT_EXPIRES ptp4l[334.032]: port 1: new foreign master 0023ae.fffe.1c6e61-1 ptp4l[334.469]: poll tx timestamp timeout ptp4l[334.469]: port 1: send sync failed ptp4l[334.469]: port 1: MASTER to FAULTY on FAULT_DETECTED (FT_UNSPECIFIED) ptp4l[350.472]: port 1: FAULTY to LISTENING on FAULT_CLEARED We are not able to figure out why "send sync failed" is happening interface is UP and connected to other linux machine with linuxptp runnning. Thanks for your help. -- Regards Aman sharma On 7/4/13, Richard Cochran <ric...@gm...> wrote: > On Thu, Jul 04, 2013 at 10:46:22AM +0530, Aman Sharma wrote: >> Hi Richard, >> >> Can You please explain me Actually what this* pi_offset_const *does.I am >> not unable to get what you want to say. > > It lets the slave jump clock again, after the first time. > Try it out, like I already explained, and see what happens. > > HTH, > Richard > |
|
From: Richard C. <ric...@gm...> - 2013-07-04 10:33:06
|
On Thu, Jul 04, 2013 at 02:59:57PM +0530, Aman Sharma wrote: > Machine details: > OS: 12.04 ubuntu > Kermel version: 3.2.0-40-generic > Ethernet controller: INTEL 82579LM with e1000e kernel driver ... > We are not able to figure out why "send sync failed" is happening > interface is UP and connected to other linux machine with linuxptp > runnning. It is because the e1000e needs kernel version 3.5 or newer to support software transmit time stamps. See the Driver Support Matrix at: http://linuxptp.sourceforge.net/ Thanks, Richard |
|
From: Aman S. <ama...@gm...> - 2013-07-05 08:10:33
|
Hi Richard, Thanks for the Support. We are able to run this now. -- Regards Aman Sharma On Thu, Jul 4, 2013 at 4:02 PM, Richard Cochran <ric...@gm...>wrote: > On Thu, Jul 04, 2013 at 02:59:57PM +0530, Aman Sharma wrote: > > > Machine details: > > OS: 12.04 ubuntu > > Kermel version: 3.2.0-40-generic > > Ethernet controller: INTEL 82579LM with e1000e kernel driver > > ... > > > We are not able to figure out why "send sync failed" is happening > > interface is UP and connected to other linux machine with linuxptp > > runnning. > > It is because the e1000e needs kernel version 3.5 or newer to support > software transmit time stamps. > > See the Driver Support Matrix at: > > http://linuxptp.sourceforge.net/ > > Thanks, > Richard > |
|
From: Aman S. <ama...@gm...> - 2013-07-12 11:03:13
|
Hi Richard, Can you please tell me how to enable 802.1as in the linuxPTP source code. By default,802.1as is in off state and what are the dependencies for the same. -- regards Aman On Fri, Jul 5, 2013 at 1:40 PM, Aman Sharma <ama...@gm...>wrote: > Hi Richard, > > Thanks for the Support. > We are able to run this now. > > -- > Regards > Aman Sharma > > > On Thu, Jul 4, 2013 at 4:02 PM, Richard Cochran <ric...@gm...>wrote: > >> On Thu, Jul 04, 2013 at 02:59:57PM +0530, Aman Sharma wrote: >> >> > Machine details: >> > OS: 12.04 ubuntu >> > Kermel version: 3.2.0-40-generic >> > Ethernet controller: INTEL 82579LM with e1000e kernel driver >> >> ... >> >> > We are not able to figure out why "send sync failed" is happening >> > interface is UP and connected to other linux machine with linuxptp >> > runnning. >> >> It is because the e1000e needs kernel version 3.5 or newer to support >> software transmit time stamps. >> >> See the Driver Support Matrix at: >> >> http://linuxptp.sourceforge.net/ >> >> Thanks, >> Richard >> > > |
|
From: Richard C. <ric...@gm...> - 2013-07-12 11:11:01
|
On Fri, Jul 12, 2013 at 04:33:03PM +0530, Aman Sharma wrote: > Hi Richard, > > Can you please tell me how to enable 802.1as in the linuxPTP source code. > By default,802.1as is in off state and what are the dependencies for the > same. ptp4l -f gPTP.cfg HTH, Richard |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X