after i was update client from tixati 2.43 to 2.47 , noticed that the statistics became wrong on the tracker/site for both download and upload , so i re-install 2.43 again , then statistics work normal , and this happen to some another members in private tracker site not just me  , Tixati user :-)

Explain, with pictures or copy of the text.
The only thing changed in the tracker announce was the "incremental" update to the "total update" according to the changelog.

As another user on the same forum - I do not think a photo of my upload and download measured in Tb would be very helpful.

I made a few timings on the download on one day - in space of 1 hour and 16 mins my download total went from 8.95Tb to 9.74Tb.

Reverting to 2.34 in my case (I did not have 2.43 to hand) cured the problem.

I did not try versions 2.44, 2.45 or 2.46 so I cannot say in which release the bug was introduced. But 2.47 has the problem.

Most likely the tracker you are using is not compatible with "total update" announcement. I had this kind of problem a few years ago. That is why I requested "per session stats" change back then.

Most popular torrent clients reset upload and download bytes every session. There are some trackers that assume every "started" event is 0 bytes uploaded/downloaded. It isn't true with Tixati, because each session after the first one will start with some bytes at event started. For example:
session 1
started1 = 0 bytes uploaded
stopped1 = 2000 bytes uploaded

session 2
started2 = 2000 bytes uploaded
stopped2 = 3000 bytes uploaded

Good trackers will calculate:
session 1 = stopped1 - started1 = 2000 - 0 = 2000
session 2 = stopped2 - started2 = 3000 - 2000 = 1000
total = 2000 + 1000 = 3000

Bad trackers will calculate:
session 1 = stopped1 - 0 = 2000 - 0 = 2000
session 2 = stopped2 - 0 = 3000 - 0 = 3000
total = 2000 + 3000 = 5000

In reality it is a bit more complicated, but you should get the idea. Basically it is tracker's fault not Tixati but tracker staff usually don't care and just ban client. Client who announces "per session stats" will be correct even with bad trackers. Of course I'm not 100% sure this is the cause in your case.

I presume the change to "total" update was for the behaviour that once the failure to announce twice happened, you lost the previous session.

Proof you can't please 100% of the people 100% of the time.

I'm not sure why the announcement is reverted to the old way. I remember seeing a thread about "wrong" numbers on a tracker webpage but I can't find it. I guess that was the cause of recent change.

I think an option somewhere in tracker settings could make everyone happy.

option would be welcomed

i see both results of this change

one private tracker, shows totals reported by tixati and counts everything normally (positive impact, as you can see actual all time traffic in torrent details on the site, while counting properly down/up traffic)

second private tracker is completely broken by this change, counting hundreds of gigabytes on download/upload, just waiting for ban ....

it was an unnecessary change (why?) and reason why i had to revert back to 2.43

I was the first guest response. Like @Pete pointed out, there are multiple ways to be "incorrect" with statistics depending on the math in the backend.
There is a way to turn on announce URLs for a transfer. That text would help, corroborated with your tracker statistics. If you can't provide that information with the "experienced" difference for bytes D/U, it is up to development whether they can parse what it is you are trying to say because i know Pete Bugmagnet and others did explore this perceived but not actual difference before: search "tracker" on this forum right up top there.

didn't find single instance of protocol definition that would indicate you can send all time traffic with downloaded/uploaded parameters, its by definition "since announcing start event"

yet another reason to ban tixati on private trackers, and we tried so hard ...

Yet another reason... because you drink FUD coolade... maybe?

See if this behavior is "normal" with 2.48? It was mentioned in changelog

It seems the problem has been solved with 2.48. There is also tracker option to use total stats if someone wishes to. Thanks!