[OpenSIPS-Users] time going backwards?
Răzvan Crainea
razvan at opensips.org
Tue Jul 13 08:40:30 EST 2021
Hi, Kingsley!
For these purposes, instead of using $Ts and $Tsm, we advise you to use
get_accurate_time[1] and ts_used_delta[2].
[1]
https://opensips.org/docs/modules/3.2.x/cfgutils.html#func_get_accurate_time
[2] https://opensips.org/docs/modules/3.2.x/cfgutils.html#func_ts_usec_delta
Best regards,
Răzvan Crainea
OpenSIPS Core Developer
http://www.opensips-solutions.com
On 7/12/21 4:53 PM, Kingsley Tart wrote:
> Gah, thanks - I was having a mega brain fart, clearly. I even already
> had this function in my config (!!) ... :-/
>
> BTW, apologies for the lateness of my reply - I've been on holiday.
>
> route[get_timestamp_hires] {
> $var(N25_resultAVP) = $param(1);
> # returns a floating point number as a string "$seconds.$decimals" representing the timestamp
> # accurate to a millisecond. Because $Ts and $Tsm are separate variabls in OpenSIPS, there is
> # a tiny chance that $Ts might advance to the next clock second while reading $Tsm, which could
> # result in a reported timestamp almost exactly a second wrong.
> # Therefore, here we generate a timestamp string where we mitigate against that possibility.
>
> $var(N25_Ts) = $Ts;
> $var(N25_Tsm) = $Tsm/1000;
> if ($Ts > $var(N25_Ts)) {
> # clock has advanced, take new values
> $var(N25_Ts) = $Ts;
> $var(N25_Tsm) = $Tsm/1000;
> }
> if ($var(N25_Tsm) < 10) {
> $var(N25_Tsm) = "00" + $var(N25_Tsm);
> } else if ($var(N25_Tsm) < 100) {
> $var(N25_Tsm) = "0" + $var(N25_Tsm);
> }
>
> $avp($var(N25_resultAVP)) = "" + $var(N25_Ts) + "." + $var(N25_Tsm);
> }
>
> Cheers,
> Kingsley.
>
> On Sat, 2021-06-26 at 19:37 +0200, Gregory Massel wrote:
>> Per https://www.opensips.org/Documentation/Script-CoreVar-3-1
>> $Tsm - reference to current microseconds of the current second
>> This means that if the second counter cycles up, the microsecond counter will cycle back to zero and you'll get a negative.
>> e.g. Let's assume that at the time of the first reading of $Tsm it is 12:00:00.99999 and, at the time of the second reading, it's 12:00:01.00003, then you'll get a result of -0.99996, because you're looking at the fraction of a second in isolation to the second counter.
>> So you probably need something like:
>> $var(t0s) = $Ts;
>> $var(t0ms) = $Tsm;
>> $var(node_number) = route(get_my_node_number);
>> $var(dur) = (($Ts - $var(t0s))*1000000) + $Tsm - $var(t0ms);
>>
>> Note: I'm not sure whether the syntax above is correct (I've never tried to do arithmetic operations within OpenSIPS), but the basic idea is there.
>> --Greg
>> On 2021-06-25 22:01, Kingsley Tart wrote:
>>> Hi,
>>>
>>> For testing, this code at the top of my route script:
>>>
>>> $var(t0) = $Tsm;
>>> $var(node_number) = route(get_my_node_number);
>>> $var(dur) = $Tsm - $var(t0);
>>> xlog("My node number is $var(node_number) and it took $var(dur) microsecconds to find out\n");
>>>
>>> has just logged this:
>>>
>>> Jun 25 20:55:01 [598]: My node number is -1 and it took 65 microsecconds to find out
>>> Jun 25 20:55:03 [598]: My node number is -1 and it took 67 microsecconds to find out
>>> Jun 25 20:55:05 [598]: My node number is -1 and it took 64 microsecconds to find out
>>> Jun 25 20:55:05 [578]: My node number is -1 and it took 602523 microsecconds to find out
>>> Jun 25 20:55:06 [579]: My node number is -1 and it took -79396 microsecconds to find out
>>> Jun 25 20:55:07 [598]: My node number is -1 and it took 63 microsecconds to find out
>>> Jun 25 20:55:07 [580]: My node number is -1 and it took -455503 microsecconds to find out
>>> Jun 25 20:55:07 [579]: My node number is -1 and it took 108 microsecconds to find out
>>> Jun 25 20:55:09 [598]: My node number is -1 and it took 78 microsecconds to find out
>>> Jun 25 20:55:11 [598]: My node number is -1 and it took 68 microsecconds to find out
>>> Jun 25 20:55:13 [598]: My node number is -1 and it took 66 microsecconds to find out
>>> Jun 25 20:55:15 [598]: My node number is -1 and it took 68 microsecconds to find out
>>> Jun 25 20:55:17 [598]: My node number is -1 and it took 65 microsecconds to find out
>>> Jun 25 20:55:18 [578]: My node number is -1 and it took 107 microsecconds to find out
>>> Jun 25 20:55:19 [598]: My node number is -1 and it took 71 microsecconds to find out
>>>
>>> which is weird because time seems to have gone backwards a couple of times.
>>>
>>> Using OpenSIPS 3.1.2 in a VM running Debian 10, with the host being a
>>> CentOS 7 machine running Qemu KVM.
>>>
>>> Any idea what's happening here?
>>>
>>> Cheers,
>>> Kingsley.
>>>
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opensips.org
>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
More information about the Users
mailing list