Modify

Opened 2 years ago

Closed 22 months ago

Last modified 22 months ago

#87 closed defect (fixed)

Bad latency

Reported by: regina.operation@… Owned by: stuerze
Priority: normal Component: BNC
Version: Keywords:
Cc:

Description

Hello,

We installed the BNC 2.12.2 and we saw bad latency from different streams. Some of the streams used satellite communication, and we observed negative latency (we have a configuration in order to retrieve the mean latency and the number of epochs for checking our receiver network for real time)
Our system used NTP, and before that we used BNC 2.11.1 and the latency seemed good regarding the network.

Attachments (4)

log_160807.7z (166.0 KB) - added by regina.operation@… 23 months ago.
log + config file
log_160717.7z (108.3 KB) - added by regina.operation@… 23 months ago.
log_160724.7z (130.8 KB) - added by stat.log_160731.gz 23 months ago.
BNCConfigStat.bnc (5.8 KB) - added by regina.operation@… 23 months ago.

Download all attachments as: .zip

Change History (12)

comment:1 Changed 2 years ago by stuerze

Hello,

The latency is computed from time differences between observations epoch time and current time.

The current time is related to the computer time, where the latency determination takes place.
Hence, the clock of the host computer has to be properly synchronized.

Because we cannot observe negative latencies, it is difficult to explain your phenomenon. Maybe you can check if your NTP server is working. From my point of view, the satellite communication link should not manipulate the observations epoch time.

Best regards, Andrea.

comment:2 Changed 2 years ago by regina.operation@…

Dear Andrea,

Sorry for the delay i saw a little bit late tha my reply was rejected by 'trac@…'
So, our server is synchronized to ntp.
Before using version 2.12.2 we used bnc in version 2.11.1 and we installed the version 2.12.1 on the same machine. We made test and comparison of latency between the 2 versions and we observed bad latency (on version 2.12.1) with some negative latency or latency inferior to 0.3s for data providing by a satellite transmission. So we decided to not install this version and keep version 2.11.1.
Unfortunally, without changing configuration, the version 2.11.1 of BNC stopped by itself (on two machine in the same time) and we decided to installed BNC in version 2.12.2. and it works well for our needs except latency. And we don't understand why because ntp is OK (with ntpstat we have 23ms on one server and 11ms for the other server). Is there a specific configuration parameters on version 2.12.2 which can be affect the latency... maybe we missed it ?

Thank you for your help

comment:3 Changed 2 years ago by stuerze

Dear Colleague,

Unfortunately, there is no special parameter to setup the latency determination more correctly.
We have a lot of latency checks running as well. But without negative results.
Hence, would it be possible to get a configuration file from you, with a stream that provides such strange results?

Best regards, Andrea.

comment:4 Changed 23 months ago by regina.operation@…

Dear Andrea,

After analyses, the issue appears some times from different streams and different hardware (novatel, leica and trimble. it's not very often but when it appears, there is a lot of shift with the utc time. In the log provided (see attached file) by the tool, there is a message concerning OWMG1 stream at the begining of the day (-647s) "wrong observation epochs".
All the negative latency observed appears at the begining of the week (sunday).
You will find on attached file, the log where we encountered the issue and the config file of BNC.

Thank you for your support
Best regards
Gregory

Changed 23 months ago by regina.operation@…

Attachment: log_160807.7z added

log + config file

Changed 23 months ago by regina.operation@…

Attachment: log_160717.7z added

Changed 23 months ago by stat.log_160731.gz

Attachment: log_160724.7z added

Changed 23 months ago by regina.operation@…

Attachment: BNCConfigStat.bnc added

comment:5 Changed 23 months ago by stuerze

Dear Gregory,

Thank you verry much for the hopefully determining hint and the config file. I think the reason was, that the check regarding wrong observation epochs was done after the latency check..
I will test the source code changes now and let you know the results.

Best regards and have a nice weekend.

Andrea

comment:6 Changed 22 months ago by stuerze

The issue seems to be solved now:

16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
16-08-18 06:08:04 KITG1: Wrong observation epoch(s)
..
16-08-18 06:15:00 KITG1: Mean latency 0.71 sec, min 0.42, max 3.53, rms 0.22, 635 epochs, 1 gaps
16-08-18 06:30:00 KITG1: Mean latency 0.7 sec, min 0.32, max 1.74, rms 0.15, 900 epochs, 0 gaps

Best regadrs, Andrea

comment:7 Changed 22 months ago by stuerze

Resolution: fixed
Status: newclosed

comment:8 Changed 22 months ago by anonymous

Thank you very much Andrea for your help

Modify Ticket

Change Properties
Action
as closed The owner will remain stuerze.
The resolution will be deleted.

Add Comment


E-mail address and name can be saved in the Preferences.

 
Note: See TracTickets for help on using tickets.