NTP Polling Interval

[RebusCom]

All our switches and routers have the same NTP source but I notice the WS Netonix switches vary by up to several minutes. We're still running 1.5.8 if that matters. What is the NTP polling interval?

[RebusCom]

I find it interesting that the switch logs show "system: time set by NTP server" following a reboot, but that message is never logged again for as long as the switch runs. Does this mean the switches never poll the NTP server following a reboot? That would explain the drift. There's also no button in the GUI to manually refresh NTP.

[sakita]

Hmm, there was a fix like this in 1.5.6...

viewtopic.php?f=17&t=6198&p=32695&hilit=+NTP#p32695

Also wonder if changing the NTP address to an invalid one and then back to the valid one still works or if there is a command line command to get it to get NTP?

Agreed, a [NTP Sync] button would be cool as well as a settings for how often it updates (e.g. how many minutes between updates).

I'll have to check some switches at work this week to see if they are also not updating (currently running 1.5.14 except for one location testing latest rc).

[Dave]

we will look into this, basically NTP is suppose to poll every ten minutes....please post logs from systems you say are out of sync...thanks.

[RebusCom]

we will look into this, basically NTP is suppose to poll every ten minutes....please post logs from systems you say are out of sync...thanks.

None of our switches poll regularly -- they don't keep accurate time. The NTP service appears to start after reboot but may take multiple tries before it actually starts and then it gets the time. Once the time is set successfully it never polls thereafter.

I'll PM you logs from five of our switches representing each model of switch.

[Dave]

thanks for logs, we are going to set up a test bed & investigate this issue.

[sakita]

Confirmed same behavior in both 1.5.14 and 1.5.15rc3... lots of tries (usually then the port with the uplink radio comes up) and then it gets the time once. A bunch of "system: starting ntpclient" events and then one "system: time set by NTP server" event. Likewise, every switch time was not correct.

Ran top in the command line and didn't see ntpclient in the list of stuff running (by comparison, top on a Isostation radio does show ntpclient running).

[sirhc]

Confirmed same behavior in both 1.5.14 and 1.5.15rc3... lots of tries (usually then the port with the uplink radio comes up) and then it gets the time once. A bunch of "system: starting ntpclient" events and then one "system: time set by NTP server" event. Likewise, every switch time was not correct.

Ran top in the command line and didn't see ntpclient in the list of stuff running (by comparison, top on a Isostation radio does show ntpclient running).

Well yea web the switch is booting up it does not have net access and NTP will keep trying for a reasonable time.

I am pretty sure we are NOT getting time updates from NTP after boot. I will discuss putting this in next release.

[Katlybrtoletti83]

I'm intrigued by the fact that the switch logs display "system: time set by NTP server" after a reboot, yet this message never appears again while the switch remains operational. Could this indicate that the switches do not continue to poll the NTP server after reboot, which could explain the time drift? Additionally, there is no button available in the GUI to manually refresh the NTP synchronization.